Está en la página 1de 182



Economic Implications of
Remittances and Migration 2006
Economic Implications of
Remittances and Migration

© 2006 The International Bank for Reconstruction and Development / The World Bank
1818 H Street, NW
Washington, DC 20433
Telephone: 202-473-1000

1 2 3 4 09 08 07 06

This volume is a product of the staff of the World Bank. The findings, interpretations, and
conclusions expressed herein do not necessarily reflect the views of the Board of Executive
Directors of the World Bank or the governments they represent.

The World Bank does not guarantee the accuracy of the data included in this work. The
boundaries, colors, denominations, and other information shown on any map in this work do
not imply any judgment on the part of the World Bank concerning the legal status of any
territory or the endorsement or acceptance of such boundaries.

Rights and Permissions

The material in this work is copyrighted. Copying and/or transmitting portions or all of this
work without permission may be a violation of applicable law. The World Bank encourages
dissemination of its work and will normally grant permission promptly.

For permission to photocopy or reprint any part of this work, please send a request with
complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,
MA 01923, USA, telephone 978-750-8400, fax 978-750-4470,

All other queries on rights and licenses, including subsidiary rights, should be addressed to
the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, USA,
fax 202-522-2422, e-mail

ISBN: 08213-6344-1
E-ISBN: 0-8213-6345-X
13-digit E-ISBN: 978-0-8213-6345-4
DOI: 10.1596/978-0-8213-6344-7
EAN: 978-0-8213-6344-7

ISSN: 1014-8906

Library of Congress Cataloging-in-Publication data has been applied for.

Cover design: Naylor Design

Cover photo: Panos

Foreword vii

Acknowledgments ix

Overview xi

Abbreviations xvii

Chapter 1 Prospects for the Global Economy 1

Global growth 3
Long-term prospects and poverty forecast 8
International finance 10
Commodity markets 14
World trade 16
Risks and uncertainties 18
Notes 22
References 23

Chapter 2 The Potential Gains from International Migration 25

International migration trends 26
The demographic challenge 29
Migration and its development impact 31
Returns to households 34
Returns to factors of production 41
Caveats—what the model leaves out 48
Notes 51
References 53

Chapter 3 The Policy Challenges of Migration: The Origin Countries’ Perspective 57

The migration decision and its impact on migrants and their families 59
Low-skilled migration 64
High-skilled emigration 66
Diasporas 70
The return of expatriates can benefit development 70
Temporary migration and international agreements 72
Notes 76
References 78


Chapter 4 Trends, Determinants, and Macroeconomic Effects of Remittances 85

Remittance data and trends 86
Factors affecting remittance flows 92
Macroeconomic effects of remittances 99
Annex 4A.1 World Bank data on remittances 105
Annex 4A.2 A model-based estimation of informal remittance flows 108
Notes 110
References 112

Chapter 5 Remittances, Households, and Poverty 117

Remittances, poverty, and inequality 118
Remittances and household consumption smoothing 122
Remittances and indirect effects on household income 123
Remittances, savings, and investment 125
Annex 5A.1 Poverty simulation model: description and results 127
Notes 129
References 131

Chapter 6 Reducing Remittance Fees 135

Remittance fees and costs 136
Factors underpinning high remittance fees 144
Policies to reduce remittance costs 147
Remittances and financial institutions 149
Annex 6A.1 A stylized remittance transaction—structure, players, instruments 151
Annex 6A.2 Licensing and registration requirements for remittance service
providers 152
Annex 6A.3 A brief history of some remittance service providers 153
Notes 154
References 156

1.1 Industrial production 3
1.2 A sharp slowdown 5
1.3 Regional growth 6
1.4 Dollar-euro interest rate differentials 10
1.5 Financing of the U.S. current account deficit 11
1.6 Emerging market spreads 11
1.7 Real long-term interest rates in G-7 countries 12
1.8 World savings rate 12
1.9 Inflation rates 13
1.10 Cumulative real increase in housing prices, 2005 14
1.11 Commodity prices 14
1.12 Levels of spare oil capacity 15
1.13 World trade volumes 16
1.14 Change in textile exports to the developed world, first half of 2005 17
1.15 Estimated change in textile exports as share of total merchandise exports 18
1.16 Some countries are particularly at risk 20
2.1 International migrants as a share of destination countries’ population 27


2.2 Share of females in international migration 28

2.3 Immigration to selected countries, reasons for admittance, 2001 28
2.4 Labor force and dependency rates 30
2.5 Factor returns and migration 42
2.6 Source of gains for native workers 45
2.7 Factor returns and migration in high-income countries, 2005 46
3.1 Median wage levels for workers in the same occupation, relative to high-income
economies, 1988–92 59
3.2 Major developing country diasporas in developed countries 62
3.3 Estimates of stock of irregular migration 62
3.4 Emigration rates for low-skilled workers 64
3.5 Emigration rates for those with a tertiary education, 2000 67
3.6 Number of temporary workers admitted under skill-based programs 72
4.1 Top 20 remittance-recipient countries, 2004 90
4.2 Estimated remittance payment, by country group, 2004 91
4.3 Remittances as percent of private consumption, two years before and two years after
natural disasters 100
4.4 Remittances as a share of personal consumption, two years before and two years after
conflict 100
4.5 Remittances as a share of personal consumption, two years before and two years after
financial crises 101
4.6 Indebtedness classification including and excluding remittances, 2003 101
4.7 Remittance securitization structure 102
4.8 Securitization of remittances, 1994–2004 103
5.1 Sri Lankan migration 122
6.1 Remittance costs are high and regressive 138
6.2 Remittance fees in the United States–Mexico corridor 138
6.3 Fees and foreign exchange spreads for $200 in Western Union transfers from New
York City 144
6.4 Exclusive arrangements with post offices skew competition 145
6.5 Barriers perceived by remittance service providers 146

1.1 The global outlook in summary 4
1.2 Long-term prospects 8
1.3 Regional breakdown of poverty in developing countries 9
1.4 Terms-of-trade impacts of commodity price changes 16
1.5 Impact of a 2 million bpd negative supply shock 19
1.6 Interest rate scenarios 20
2.1 Growth in international migration by destination, 1970–2000 27
2.2 Labor force structure in the base case and after increases in migrants 33
2.3 Change in real income across households in 2025 relative to baseline 34
2.4 Real income impacts across developing regions 38
2.5 Impact of different assumptions on the consumption of public goods and services by
selected groups in 2025 39
3.1 Fees charged by recruitment agencies 60
3.2 Emigration rates of skilled workers, 2000 68
4.1 Workers’ remittances to developing countries, 1990–2005 88


4.2 Recorded remittances have grown faster than private capital flows and ODA 88
4.3 Choice of remittance channel in selected countries 91
4.4 Estimated increase in formal remittances if transaction costs were reduced to
2 to 5 percent and dual exchange rates were eliminated 92
4.5 Impact of remittances on country credit rating and sovereign spread 102
4A.1.1 Countries with alternative estimates in 2004 106
4A.2.1 Regression results: determinants of worker remittances 109
4A.2.2 Regression results: determinants of transaction costs 109
4A.2.3 Panel regression results: determinants of remittances 110
5.1 Simulated impact of eliminating remittances on poverty rate 120
5A.1 Effect of removing remittances on the poverty headcount rate 128
6.1 Approximate cost of remitting $200 137
6.2 Operating profits of major MTOs 140
6.3 Estimating the cost of a remittance transaction 140
6.4 Remittances are more cost-elastic when costs are higher 143
6.5 Policies to reduce costs, regulate informal providers, and provide remittance-linked
financial services 148

2.1 The model used in this study 32
2.2 Calculating and interpreting global welfare gains from migration 36
2.3 The impact of immigrants on fiscal balances 40
2.4 Empirical studies of the impact of immigration on wages 43
2.5 Increased migration and its impact on wages 47
3.1 Internal versus international migration 65
3.2 Mode 4 and international migration 75
4.1 International working group on improving data on remittances 87
4.2 The recent surge in remittance flows to India 89
4.3 Collective remittances through hometown associations and matching schemes 95
4.4 Forced remittances 97
4.5 Unlike oil windfalls, remittance inflows do not weaken institutional capacity 105
5.1 Estimating a cross-country poverty change model 119
6.1 Decline in remittance costs in the United States–Mexico corridor 139
6.2 Estimating remittance industry costs 141
6.3 Even charitable donations are sensitive to cost 142
6.4 United States–Mexico FedACH 148
6.5 The World Bank/CPSS task force on general principles for international remittance
systems 149
6.6 Smart’s phone-based remittance system in the Philippines 150


or millennia people have migrated in The World Bank’s research department, in
search of economic opportunity. In the partnership with others, has launched a pro-
nineteenth and early twentieth centuries, gram to expand knowledge in an area that de-
technological advances and untapped natural serves greater attention. The program ad-
resources drove movements of population from dresses the issues surrounding remittances;
Europe and Asia to the Americas. International migration of high-skilled workers; the deter-
migration generated enormous improvements minants of migration; temporary movements
in people’s lives. Immigrants enjoyed higher of persons; social protection and governance;
wages, countries of destination profited from and the links among trade, foreign direct in-
increased supply of labor, and countries of ori- vestment, and migration.
gin saw labor market pressures ease. An integral part of this program, Global
Current trends indicate that pressures for Economic Prospects 2006 focuses on policies
migration from the south to the north are set to improve the developmental impact of remit-
to rise again. This movement is driven largely tances. It documents the high level of transac-
by income gaps and the rising number of tions costs facing migrants sending small re-
young adults in developing countries seeking mittances to their families, and it outlines the
better opportunities abroad. The economic, regulatory issues and market imperfections
social, and political implications that come that keep costs high.
with the movement of people differ from the Fewer barriers to remittance flows and
movement of goods or money. As a result, the greater competition among remittance service
topic of international migration has prompted providers could substantially reduce costs and
much political debate in the international boost remittance flows to developing coun-
community today. tries. Global Economic Prospects 2006 shows
The prospects for migration flows are crit- how sound domestic policies and an invest-
ical for development. Developing countries ment-friendly climate can significantly increase
benefit through the money that migrants send the contribution of remittances and migration
home to their families (remittances), through to improved living conditions back home.
reduced labor market pressures, and through Migration remains an important force for
contacts with international markets and ac- fighting poverty, the key mission of the World
cess to technology. Bank, and it is our hope that this report will
But migration is not always beneficial. Mi- contribute to this important debate.
grants can be subject to exploitation and Paul Wolfowitz
abuse, and the loss of highly skilled personnel President
through migration has hindered development World Bank
in some countries. November 2005


HIS REPORT WAS prepared by the Development Prospects Group (DECPG). The lead authors
of this report were Dilip Ratha and William Shaw, with direction by Uri Dadush. The
principal authors of the chapters were Andrew Burns (chapter 1), Dominique van der
Mensbrugghe (chapter 2), William Shaw (chapter 3), and Dilip Ratha (chapters 4, 5, and 6). The
report was prepared under the general guidance of François Bourguignon, chief economist and
senior vice president of the World Bank.
The main macroeconomic forecasts in chapter 1 were prepared by the Global Trends Team of
DECPG led by Hans Timmer and including John Baffes, Andrew Burns, Maurizio Bussolo,
Annette de Kleine, Betty Dow, Himmat Kalsi, Fernando Martel Garcia, Donald Mitchell, Gauresh
Shailesh Rajadhyaksha, Mick Riordan, Cristina Savescu, Shane Streifel, and Shuo Tan. The out-
look for the East Asia and Pacific region was carried out with the cooperation of Milan Brahmb-
hatt and Louis Kuijs. The team also benefitted from in-depth consultations and comments from
the regional chief economists and their staff, as well as country economists. The long-term growth
and poverty forecasts were prepared by Dominique van der Mensbrugghe, Shaohua Chen, and
Martin Ravallion. The companion Prospects for the Global Economy web site was prepared by
Andrew Burns, Sarah Crow, and Cristina Savescu, in collaboration with Reza Farivari, Saurabh
Gupta, David Hobbs, Shahin Outadi, Raja Reddy Komati Reddy, Malarvizhi Veerappan, and
Cherin Verghese.
Maddalena Honorati and Prabal De provided research assistance. Chapter 2 benefitted from
collaboration with Hans Timmer and from comments received from seminar participants, no-
tably Lindsay Lowell and Susan Martin. Special thanks are due for the background material
provided by Riccardo Faini, Robert Lucas, Julia Nielson, Kathleen Newland, and Irena Omela-
niuk for chapter 3; Swaminathan S. Aiyar, Ralph Chami, Neil Fantom, Caroline Freund, Gary
McMahon, Irena Omelaniuk, Serdar Sayan, Nikola Spatafora, and K. M. Vijayalakshmi for
chapter 4; John McHale for chapter 5; John Gibson, David McKenzie, George Kalan, Dilek
Aykut, Nikos Passas, and Jan Riedberg for chapter 6. Ole Andreassen, Jose de Luna Martinez,
Raul E. Hernandez-Coss, Massimo Cirasino, and Roger Ballard also contributed background
notes for chapter 6. Thanks also to colleagues in the International Organization for Migration
who helped collect information on remittance-related government policies (for chapter 4) using
their extensive international network, and Bernd Balkenhol of the International Labour Organi-
zation for preparing a background paper on forced remittances.


Many colleagues provided excellent comments at various stages of the report’s preparation.
L. Alan Winters provided comments on the report and guidance throughout its preparation. Luca
Barbone, Kevin Barnes, Augusto de la Torre, Shantayanan Devarajan, Mustapha Nabli, John
Page, Bryan Roberts, John Whalley, and Dean Yang were peer reviewers at the Bankwide review.
Richard Adams, William Easterly, Isaku Endo, Jose Maria Fanelli, Shahrokh Fardoust, Ian
Goldin, Daria Goldstein, Yevgeny Kuznetsov, Ali Mansoor, Phil Martin, Maria Soledad Martinez
Peria, Fernando Montes-Negret, Nayantara Mukerji, Latifah Osman Merican, Christopher Par-
sons, Guillermo Perry, Sonia Plaza, S. Ramachandran, and Terrie Walmsley also provided useful
comments. Johan Mistiaen and Romeo Matsas provided excellent help in designing and imple-
menting a survey of migrant remitters from Congo, Nigeria, and Senegal residing in Belgium.
Maria Amparo Gamboa, Araceli Jimeno, Katherine Rollins, Sarah Crow, and Michael Paul pro-
vided invaluable administrative support, including the collection of remittance fee data from all
over the world.
The report team held consultations in July 2005 in Accra, Brussels, Geneva, London, and
Paris. Thanks are due to Haleh Bridi, Barbara Genevaz, Carlos Braga, Sonia Plaza, Michelle
Bailly, and other colleagues in these country offices for efficiently and enthusiastically arranging
consultations with several international, academic, financial, and non-governmental institutions.
Thanks are also due to the International Organization for Migration for assistance with organiz-
ing consultations in Geneva and to the International Labour Organization, the Global Commis-
sion on International Migration, the European Commission, and the Commonwealth Secretariat
for participating in consultations and providing useful feedback.
This report also benefitted from the comments of the Bank’s executive directors made at an
informal board meeting on October 20, 2005.
Marilou Uy, Alan Gelb, Jeff Lewis, Amar Bhattacharya, Shaida Badiee, Robert Keppler, and
Misha Belkindas provided guidance and encouragement to the team at various stages. Dorota A.
Nowak managed production and dissemination activities by DECPG. Steven Kennedy’s contri-
bution as an editor is gratefully acknowledged. Book design, editing, and production were coor-
dinated by the World Bank Office of the Publisher.


HE THEMES OF this year’s Global Eco- This publication has two goals. The first is
nomic Prospects are international re- to explore the gains and losses from interna-
mittances and migration, their eco- tional migration from the perspective of devel-
nomic consequences, and how policies can oping countries, with special attention to the
increase their role in reducing poverty. Interna- money that migrants send home. The second
tional migration can generate substantial wel- goal is to consider policy initiatives that could
fare gains for migrants and their families and improve the developmental impact of migra-
for the countries involved (countries of origin tion, again with particular attention to remit-
and destination). The money that migrants tances. Our focus (for economic purposes) is
send home—remittances—is an important on international migration from developing
source of extra income for migrants’ families countries to high-income countries. Despite
and for developing countries: in aggregate, re- their importance, internal migration, migra-
mittances are more than twice as the size of in- tion among developing countries, and the po-
ternational aid flows. However, migration litical and social impacts of migration are
should not be viewed as a substitute for eco- beyond the scope of this work.
nomic development in the origin country— It is important to keep in mind three basic
development ultimately depends on sound principles. First, migration is a diverse phe-
domestic economic policies. nomenon, and its economic impact in one
Over the past two decades, barriers to location or another depends heavily on the par-
cross-border trade and financial transactions ticular circumstances involved. Second, basic
have fallen significantly, while barriers to the data on migration and remittances are lacking,
cross-border movement of people remain so predicting the impact of policy changes can
high. Despite its economic benefits, migration be problematic. This underlines the need for
remains controversial and, for some people, better data and more research. Third, migra-
threatening. In part, this is because migration, tion has social and political implications that
like trade and capital movements, has distrib- may be just as important as the economic
utional consequences, whereby net gains for analysis provided here. These are ably and
society may mask important losses for some comprehensively discussed in the recent report
individuals and groups. But migration also of the United Nations’ Global Commission on
sparks resistance because the movement of International Migration. For all of these rea-
people has economic, psychological, social, sons, the analysis and policy recommendations
and political implications that the movement for migration must remain qualified. This
of goods or money do not. report draws conclusions where they can be


supported by adequate data and points to an The impact of remittances

agenda for research where they cannot. and migration
The impact on migrants
Global economic prospects The bulk of the economic gains from migra-

T he slowdown among industrial economies

that began in the second half of 2004 con-
tinued in 2005, with GDP growth expected to
tion accrue to migrants and their families, and
these gains are often large. Wage levels (ad-
justed for purchasing power) in high-income
come in at 2.5 percent, down from 3.1 percent countries are approximately five times those of
the year before. The pace of the expansion in low-income countries for similar occupations,
the high-income countries is forecast to in- generating an enormous incentive to emigrate.
crease slightly over the next two years, with ac- Moreover, to the extent that migrants devote
celeration in Europe offsetting a modest slow- a portion of their income to remittances, the
ing in Japan and stable growth in the United gains are even greater. Essentially migrants can
States. Economic activity also slowed in devel- earn salaries that reflect industrial-country
oping economies during 2005. Higher oil prices and spend the money in developing
prices, domestic capacity constraints, and countries, where the prices of nontraded goods
slower demand for their exports brought GDP are much lower. Migrants, however, incur sub-
growth down from a very strong 6.8 percent in stantial costs, including psychological costs,
2004 to an estimated 5.9 percent this year. and immigrants (particularly irregular mi-
While GDP growth has remained robust, grants) sometimes run high risks; many suffer
higher oil prices have sharply slowed real in- from exploitation and abuse. The decision to
come growth among oil importers from 6.4 to migrate is often made with inaccurate infor-
3.7 percent. Looking forward, continued high mation. Given the high costs of migration—
oil prices, coupled with inflationary pressures, including the risks of exploitation and the ex-
are expected to restrain growth in most devel- orbitant fees paid to traffickers—the net
oping countries over the next two years. Nev- benefit in some cases may be low or even neg-
ertheless, GDP in these economies should ative. There are costs, too, for family members
expand by around 5.5 percent—much more left behind—particularly children—although
quickly than during the past two decades. these costs must be balanced against the bene-
This relatively positive outlook is subject to fits of the extra income that migrants send
important downside risks. The outlook for oil back home to their families.
prices is particularly uncertain. Low excess ca-
pacity has introduced a risk premium into oil The impact on destination countries
prices and will make it more difficult to con- Destination countries can enjoy significant eco-
tain the impact of a future supply shock, nomic gains from migration. The increased
should one arise. As a result, a significant sup- availability of labor boosts returns to capital
ply disruption could slow global growth, with and reduces the cost of production. A model-
large negative consequences for global eco- based simulation performed for this study indi-
nomic prospects. The future path of interest cates that a rise in migration from developing
rates, which despite recent increases are still countries sufficient to raise the labor force of
low, is another source of uncertainty. Persis- high-income countries by 3 percent could boost
tent global imbalances, signs of rising infla- incomes of natives in high-income countries by
tion, and concerns about the sustainability of 0.4 percent. In addition, high-income countries
government finances in industrialized coun- may benefit from increased labor-market flexi-
tries are all factors that could push rates bility, an increased labor force due to lower
higher and possibly provoke a much more se- prices for services such as child care, and per-
rious slowdown. haps economies of scale and increased diversity.


Nevertheless, there are losers within destina- remittances at all, even those sent through for-
tion countries. Some workers may see an ero- mal channels, or they report remittances under
sion of wages or employment, although this ef- other balance of payments entries.
fect is found to be small in most empirical Despite the prominence given to remit-
studies. In the model-based simulation of the tances from developed countries, South-South
impact of increased migration, earlier migrants remittance flows make up between 30 and 45
suffer significant income losses, while the im- percent of total remittances received by devel-
pact on natives’ wages is small. (The differential oping countries, reflecting the fact that over
impact is reduced if foreign-born workers are half of migrants from developing countries
viewed as closer substitutes for natives.) Easing migrate to other developing countries.
rules that limit labor-market flexibility, and While the impact of remittances on growth
strengthening institutions that provide educa- is unclear, remittances do play an important
tion and training, will help workers displaced role in reducing the incidence and severity of
by immigration (both natives and resident mi- poverty (with no significant effect on income
grants) to find work. Note that the simulation inequality). Remittances directly increase the
results are not intended to incorporate all of the income of the recipient and can help smooth
economic impacts of migration, nor do they household consumption, especially in response
capture important social and political implica- to adverse events, such as crop failure or a
tions. The goal is not to forecast the overall im- health crisis. In addition to bringing the direct
pact of increased migration, but rather to give benefit of higher wages earned abroad, migra-
us insights into the economic gains that might tion helps households diversify their sources of
be expected from changes in policy or circum- income (and thus reduce their vulnerability to
stances, as well as insights into the channels risks) while providing a much needed source of
through which migration affects welfare. savings and capital for investment. Remit-
tances appear to be associated with increased
The impact on origin countries household investments in education, entrepre-
Migration also generates economic benefits for neurship, and health—all of which have a high
origin countries, the largest being remittances. social return in most circumstances.
International remittances received by develop- Measuring the poverty impact of remit-
ing countries—expected to reach $167 billion tances is difficult: data are scarce, and calcu-
in 2005—have doubled in the past five years as lating the income gains from remittances re-
a result of (a) the increased scrutiny of flows quires assumptions concerning what migrants
since the terrorist attacks of September 2001, would have earned if they had stayed at home.
(b) changes in the industry that support remit- Careful analyses of the available household
tances (lower costs, expanding networks), (c) survey data indicate that remittances have
improvements in data recording, (d) the depre- been associated with declines in the poverty
ciation of the dollar (which raises the dollar headcount ratio in several low-income
value of remittances denominated in other cur- countries—by 11 percentage points in
rencies), and (e) growth in the migrant stock Uganda, 6 in Bangladesh, and 5 in Ghana, for
and incomes. However, records still underesti- example. In Guatemala, remittances may have
mate the full scale of remittances, because pay- reduced the severity of poverty by 20 percent.
ments made through informal, unrecorded Cross-country regressions and simulations
channels are not captured. Econometric analy- also indicate that increases in remittances help
sis and available household surveys suggest that to reduce the incidence of poverty.
unrecorded flows through informal channels By generating a steady stream of foreign
may conservatively add 50 percent (or more) of exchange earnings, remittances can improve a
recorded flows. Several countries with signifi- country's creditworthiness for external bor-
cant migrant populations do not report data on rowing and, through innovative financing


mechanisms (such as securitization of remit- ties, such as education and health (particularly
tance flows), they can expand access to capital for the control of transmissible diseases), may
and lower borrowing costs. While large and be impaired; (c) opportunities to achieve
sustained remittance inflows can contribute to economies of scale in skill-intensive activities
currency appreciation, this outcome may be may be reduced; (d) society loses its return on
less severe than it is in the case of natural re- high-skilled workers trained at public ex-
source earnings, because remittances are dis- pense; and (e) the price of technical services
tributed more widely and may avoid exacer- may rise. Highly educated citizens, if they
bating strains on institutional capacity that are stayed in their countries, could help to im-
often associated with natural resource booms. prove governance, improve the quality of de-
Migration has economic implications for ori- bate on public issues, encourage education of
gin countries beyond remittances. The small size children, and strengthen the administrative
of migration flows relative to the labor force sug- capacity of the state—contributions that
gests that the effects of South–North migration would be lost through high-skilled emigration.
on working conditions for low-skilled workers It is impossible to reliably estimate the net
in the developing world as a whole must be small benefit, or cost, to origin countries of high-
as well. However, in some countries low-skilled skilled emigration because data are limited
emigration can raise demand for the remaining and a myriad of individual country circum-
low-skilled workers (including poor workers) at stances enter into the calculus of that benefit
the margin, leading to some combination of or loss. We can only offer two rough observa-
higher wages, lower unemployment, less under- tions, which reflect the wide variation in high-
employment, and greater labor force participa- skilled emigration rates among countries:
tion. Thus low-skilled emigration can offer a
valuable safety valve for insufficient employ- • Very high rates of high-skilled emigra-
ment at home. In the long run, however, devel- tion are found in countries that represent
oping country policies should aim to generate a small share of the population of the de-
adequate employment and rapid growth, rather veloping world. Many of those countries
than relying on migration as an alternative to de- have poor investment climates that likely
velopment opportunities. limit the productive employment of high-
High-skilled emigration has more complex skilled workers. Of course, the loss of
implications. Like low-skilled migration, it high-skilled workers may aggravate the
can greatly benefit migrants and their families poor investment climate and limit the
and help relieve labor market pressures. How- potential benefits of economic reform.
ever, a well-educated diaspora can improve ac- • Some countries find it difficult to pro-
cess to capital, technology, information, for- vide productive employment for many
eign exchange, and business contacts for firms high-skilled workers because of their
in the country of origin. The return of expa- small economic scale or because mis-
triates and the maintenance of close contacts guided educational policies have resulted
with high-skilled emigrants have played an in a large supply of university graduates
important role in the transfer of knowledge to for whom no suitable jobs exist.
origin countries. At the same time, large out-
flows of high-skilled workers can reduce Policies to improve the
growth in the origin country for these reasons: developmental impact of
(a) the productivity of colleagues, employees, remittances and migration
and other workers may suffer because they
lose the opportunity for training and mutually Migration policies
beneficial exchanges of ideas; (b) the provision Greater emigration of low-skilled emigrants
of key public services with positive externali- from developing to industrial countries could


make a significant contribution to poverty re- igrants to return by identifying job opportuni-
duction. The most feasible means of increas- ties for them, cooperating with destination
ing such emigration would be to promote countries that have programs to promote re-
managed migration programs between origin turn, permitting dual nationality, and helping
and destination countries that combine tem- to facilitate the portability of social insurance
porary migration of low-skilled workers with benefits.
incentives for return. Temporary programs By providing authoritative information on
have several advantages, and some disadvan- migration opportunities and risks, govern-
tages, relative to permanent migration. From ments could help avoid unfortunate, costly-to-
the perspective of the destination country, reverse migration decisions and limit the abuse
managed, temporary migration programs ease of vulnerable migrants. Labor recruiters can
social tensions by limiting permanent settle- play a valuable role in promoting migration,
ment; they limit the potential burden on pub- but emigrants’ lack of information often en-
lic expenditures because immigrants are guar- ables recruiters to capture the lion’s share of the
anteed a job and are less likely to bring rents generated by constraints on immigration
dependents; and they allow for controlled and imperfect information. Origin countries
variation of the number of immigrants in re- with effective public sector institutions might
sponse to changes in labor-market conditions, consider the regulation of recruitment agents
thus limiting adverse effects on low-skilled na- to limit rents and improve transparency.
tive workers. However, temporary migration
can be less efficient than permanent migration Remittance policies
for firms in destination countries because of Governments in destination and origin coun-
high training costs. From the origin-country tries can sharpen the developmental impact of
perspective, managed, temporary migration remittances through the application of appro-
may be the only means of securing deliberate priate policies. Access of poor migrants and
increases in low-skilled emigration and may their families to formal financial services for
raise remittances and improve the skills of re- sending and receiving remittances could be
turning workers. On the other hand, managed improved through public policies that encour-
migration programs do not guarantee future age expansion of banking networks, allow
access to labor markets (and thus to remit- domestic banks from origin countries to oper-
tances), because it is easier for destination ate overseas, provide identification cards to
countries to suspend temporary programs migrants, and facilitate the participation of
than to expel immigrants. Overall, however, microfinance institutions and credit unions in
such programs do represent a feasible ap- providing low-cost remittance services. Remit-
proach to capturing the efficiency gains from tances, in turn, can be used to support finan-
labor migration. cial products—housing and consumer loans
Origin countries that are adversely affected and insurance—for poor people.
by high-skilled emigration face challenges in A second set of promising policies could
managing it better. Service requirements for improve competition in the remittance trans-
access to publicly financed education can be fer market and thereby lower fees. The price
evaded and are likely to discourage return; of remittance transactions is often unnecessar-
and proposals for the taxation of emigrants to ily high for the small transfers typically made
the benefit of the origin country have made lit- by poor migrants. The cost of such transac-
tle progress. Origin countries can help to re- tions is often well below the fees paid by cus-
tain key workers by improving working con- tomers. Reducing transaction charges in-
ditions in public employment and by investing creases the disposable income of poor
in research and development. Origin countries migrants and increases their incentives to
can also take steps to encourage educated em- remit, because the net receipts of recipients


increase. The overall result would be stronger ings and improve the allocation of expendi-
remittance flows to developing countries. tures should be accomplished through im-
Competition among providers of remittance provements in the overall investment climate,
services could be increased by lowering capital rather than by targeting remittances. Similarly,
requirements on remittance services and open- because remittances are private funds, they
ing up postal, banking, and retail networks to should not be viewed as a substitute for offi-
nonexclusive partnerships with remittance cial development aid.
agencies. Disseminating data on remittance
fees in important remittance corridors and es-
tablishing a voluntary code of conduct for de- Organization of this study
livering fair-value transfers would improve
transparency and reduce prices for remittance
transactions. Governments could help reduce
A s is customary in this report, chapter 1
reviews recent developments in and
prospects for the global economy and their
costs by supporting the introduction of mod- implications for developing countries. Chapter
ern technology in payment systems. Alleviating 2 uses a model-based simulation to evaluate
liquidity constraints by providing a credit line the potential global welfare gains and distribu-
either to the sender or the recipient, based on tional impact from a hypothetical increase of 3
past remittance activity, would enable senders percent in high-income countries’ labor force
to take advantage of the lower fee rates avail- caused by migration from developing coun-
able only for larger remittances. Reducing ex- tries. Chapter 3 surveys the economic literature
change-rate distortions could also lower the on the benefits and costs of migration for mi-
cost of remittance transactions. Finally, regula- grants and their countries of origin, focusing
tory regimes need to strike a better balance be- on economically motivated migration from de-
tween preventing financial abuse and facilitat- veloping to high-income countries. We then
ing the flow of funds through formal channels. turn to remittances, the main theme of the re-
Several origin countries have attempted to port. Chapter 4 investigates the size of remit-
improve the developmental impact of remit- tance flows to developing countries, the use of
tances by introducing incentives to increase formal and informal channels, the role of gov-
flows and to channel them to more productive ernment policies in improving the development
uses. Such policies are more problematic than impact of remittances, and, for certain coun-
efforts to expand access to financial services tries, their macroeconomic impact. Chapter 5
or reduce transaction costs, because they pose addresses the impact of remittances at the
clear risks. Tax incentives to attract remittance household level, in particular their role in re-
inflows, for example, may also encourage tax ducing poverty, smoothing consumption, pro-
evasion, while matching-fund programs to at- viding working capital for small-scale enter-
tract remittances from migrant associations prises, and increasing household expenditures
may divert funds from other local funding pri- in areas considered to have a high social value.
orities. Efforts to channel remittances to in- The last chapter investigates policy measures
vestment, meanwhile, have met with little suc- that could lower the cost of remittance trans-
cess. Fundamentally, remittances are private actions for poor households and measures to
funds that should be treated like other sources strengthen the financial infrastructure support-
of household income. Efforts to increase sav- ing remittances.


ADB Asian Development Bank

AML Anti-money laundering
AML/CFT Anti-money laundering/countering financing of terrorism
ARTEP Asian Regional Team for Employment Protection (ILO)
ATC Agreement on Textiles and Clothing
BIS Bank for International Settlements
BoA Bank of America
BOP Balance of payment
CECEI Comité des établissements de crédit et des enterprises, d’investissement
CEMLA Center for Latin American Monetary Studies (Centro de Estudios
Monetarios Latino Americanos)
CEPAL Economic Commission for Latin America and the Caribbean (Comisión
Económico para América Latina y el Caribe)
CES-IFO Center for Economic Studies at the University of Michigan—Institute
for Economic Research
CGAP Consultative Group to Assist the Poorest
CPI Consumer price index
CPSS Committee on Payment and Settlement Systems
DDP Development data platform
DFID Department for International Development
DHS Department for Homeland Security
DPRs Diversified payment rights
DPS Deferred payment scheme
ECA Europe and Central Asia
ECOSOC UN Economic and Social Council
EU European Union
EV Equivalent variation
FATF Financial Action Task Force against Money Laundering


FDC First Data Corporation

FDI Foreign direct investment
FFMC First Financial Management Corporation
FinCEN Financial Crimes Enforcement Network
FISDL Social Investment Fund for Local Development (Fondo de Inversión
Social para el Desarrollo)
GATS General Agreement on Trade in Services
GDP Gross domestic product
GE General equilibrium
GTAP Global Trade Analysis Project
HIV/AIDS Human immunodeficiency virus/acquired immune deficiency syndrome
HSBC Hong Kong Shanghai Bank Corporation
HTAs Hometown associations
IADB Inter-American Development Bank
IBM International Business Machines
ICICI Industrial Credit and Investment Corporation of India
ICMC International Catholic Migration Commission
IFAD International Fund for Agriculture and Development
ILO International Labour Organization
IMF International Monetary Fund
IOM International Organization for Migration
IRCA Immigation Reform and Control Act
IRD Institut de recherche pour le développement
IRnet International Remittance Network
LAC Latin American and the Caribbean
LDCs Least-developed countries
MCIC Matricula Consular identification card
MDG Millennium Development Goal
MIDSA Migration for Development in Southern Africa
MFIs Microfinance institutions
MFN Most-favored nation
MICR Magnetic ink character recognition
MIDA Migration for Development in Africa
MIDWA Migration for Development in Western Africa
MPG Migration Policy Group
MSB Money service business
MTOs Money transfer operators
NAFTA North American Free Trade Agreement


NBER National Bureau of Economic Research

NGOs Non-governmental organization
ODA Official development aid
OECD Organisation for Economic Co-operation and Development
OFAC Office of Foreign Assets Control
OPEC Organization of the Petroleum Exporting Countries
OLS Ordinary least squares
PADF Pan-American Development Foundation
PPP Purchasing power parity
PRIO International Peace Institute, Oslo
PROFECO Procuraduría Federal del Consumidor
R&D Research and development
SADC Southern African Development Community
SPV Special purpose vehicle
TDS Telephone and data systems
TSG Technical Sub-Group on the Movement of Persons
UAE United Arab Emirates
UCSD University of California, San Diego
UN United Nations
UNDP United Nations Development Programme
UNESCAP United Nations Economic and Social Commission for Asia and the
UNESCO–ICSU United Nations Educational, Scientific and Cultural
Organization–International Council of Scientific Unions
USAID United States Agency for International Development
USD US dollar
WEO World Economic Outlook
WOCCU World Council of Credit Unions
WTO World Trade Organization

Prospects for the Global Economy

Following very strong growth, the world have reduced growth among oil-importing de-
economy slowed in late 2004 and into 2005 as veloping countries from 6.9 percent to 6.1
output began to push against capacity con- percent. In terms of real incomes, the slow-
straints. High oil prices cut into the incomes of down was much sharper—from 6.4 percent to
oil importers, but the expansion remained 3.7 percent. Despite still growing oil revenues,
strong, partly because of favorable conditions reduced opportunities to expand production
in financial markets, including still low infla- in the petroleum sector meant that output
tion, interest rates, and interest-rate spreads. growth in oil-exporting developing countries
Tightness in the oil market, the threat of even also eased, from 6.6 percent to 5.6 percent.
higher fuel prices, and the possibility that in- During 2006 the expansion among high-
terest rates may rise pose major threats to the income countries is projected to be stable, at
expansion. about 2.5 percent, before picking up a bit in
2007. This reflects a combination of improved
Slower but still strong growth performance in Europe and stable growth in
World GDP is estimated to have increased by the United States and Japan. In the United
3.2 percent in 2005, down from 3.8 in 2004. States, higher oil prices and tighter monetary
Growth is projected to be stable in 2006, be- policy are expected to offset the positive stim-
fore strengthening somewhat in 2007. The ulus to growth from past depreciations. The
slowdown that began in the second half of projected pickup in Europe occurs despite a
2004 was experienced throughout the indus- significant drag on growth from high oil prices
trialized world, with growth in Europe still whose effects are expected to be more than
underperforming its potential. In contrast, the offset by low interest rates, pent up investment
economies of the United States and Japan, de- demand, and a dissipation of most of the neg-
spite having slowed, are expanding at close to ative consequences following the euro’s real-
their maximum sustainable rates. effective appreciation. In Japan, strengthening
Among large developing economies, GDP domestic demand and supportive macroeco-
in 2005 continued to expand rapidly in China nomic policies should enable growth to re-
and India (in excess of 9 percent and about main close to potential, despite high oil prices.
7 percent, respectively), but slowed in Russia Growth in developing economies is pro-
as growth in oil production weakened. High jected to slow modestly from an estimated
oil prices, in combination with domestic ca- 5.9 percent in 2005 to 5.5 percent by 2007.
pacity constraints and slower import demand In East and South Asia, the expansion is
from high-income countries, are estimated to projected to moderate somewhat but remain

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

very strong, particularly in China and India. could push prices to more than $90 a barrel
In the Middle East and in both North and for more than a year, resulting in a 1.5 percent
Sub-Saharan Africa, strong oil revenues reduction in global growth by the second year
should buoy internal demand among oil ex- following the shock. The terms-of-trade im-
porters and partially offset capacity con- pact for low-income oil-importing economies
straints that will slow production growth. The would reduce incomes in these countries by
projected easing of growth in Latin America more than 4 percent of their GDP (much more
and the Caribbean reflects weaker non-oil than for high-income countries) because their
commodity prices as well as a return to trend economies are relatively oil intensive, and be-
growth in several countries that rebounded cause a supply shock–induced increase in oil
very strongly in 2004. In Europe and Central prices is unlikely to be accompanied by higher
Asia, the waning of the growth bonus follow- non-oil commodity prices.
ing EU accession and capacity constraints in
oil-producing countries are expected to con- Global imbalances remain an issue
tribute to a modest slowing of the expansion. Global current account imbalances and the
U.S. current account deficit (which is exp-
Tight commodity markets ected to exceed $750 billion in 2005) remain
Weaker global growth should reduce the important medium-term problems. During
strain in non-oil commodity markets. Already late 2004 and early 2005 tensions eased
there are signs of stabilization, and even of somewhat. Rising interest rate differentials
decline, in the prices of agricultural products, relative to European short- and long-term as-
where supply has responded to high prices. sets made private sector purchases of dollar–
Metals and shipping prices also show signs of denominated assets more attractive. As a re-
easing, although to a lesser extent. sult, the dollar appreciated some 2.5 percent
In oil markets, the projected slowdown is in real-effective terms during the first seven
not expected to be sufficient to generate a sub- months of 2005, and reserve accumulation by
stantial easing of prices. While crude oil supply foreign central banks became less important in
is growing marginally faster than demand, the financing of the current account deficit.
supply conditions are expected to remain tight. This respite appears to have been short-
As a result, crude oil prices, which currently lived. To some extent, the increased private
embody a large risk premium, are not expected flows represented a one-off portfolio adjust-
to fall rapidly. The baseline assumes that no ment toward U.S. assets by investors. Begin-
major supply disruptions occur and that there ning in the second quarter of 2005, the flows
will be a gradual decline in oil prices toward diminished, and the dollar faced renewed
$40 per barrel by 2010. This implies an aver- downward pressure. As a result, foreign re-
age price of $56 for a barrel of oil in 2006 and serve accumulation once again became a criti-
$52 in 2007. cal component in the financing of the U.S. cur-
Future spikes in oil prices form a potential rent account deficit, restoring the risk that a
risk to global prospects. A price hike gener- change in behavior on the part of foreign cen-
ated by a sustained negative supply shock tral bankers could prove destabilizing. Recent
would be particularly disruptive, because out- decisions by China and Malaysia to widen the
put would be constrained directly by the re- range of currencies to which their own cur-
duced availability of oil and petroleum-based rencies are pegged could help ease future pres-
inputs. This would be in contrast to the recent sures, especially if the scope for appreciation
past, when prices rose in the context of rapidly included in the regime is exercised in practice.
growing supply. A supply shock that reduced Globally, policy should continue to focus on
oil deliveries by 2 million barrels per day increasing public and private savings in deficit


countries and increasing spending (notably on same way that other countries have profited
investment goods) in surplus countries. from freer trade in the manufacturing and raw
materials sectors.
Low interest rates are a source
of uncertainty
The future path of long-term interest rates and Global growth
spreads, which have been at historically low
levels for an extended period, is an important
uncertainty. A number of factors have helped
T he global economy slowed markedly in
2005, but still continued to expand at an
estimated 3.2 percent pace, compared with
maintain interest rates at low levels, including 3.8 percent in 2004 (table 1.1). The slowdown
several years of very loose monetary policy was widespread, reaching virtually every eco-
throughout the developed world; increased nomic region. It was precipitated by higher oil
aging-related savings in Europe; balance-sheet prices, resource-sector capacity constraints,
consolidation in the United States and tightening monetary policy in the United
Asia; and a low inflationary environment— States, and in some countries, the maturation
thanks, in part, to increased competition fol- of the investment cycle following a year of
lowing the entry into global markets of China very fast growth.
and members of the former Soviet bloc. Most
of these factors are temporary and are expected
Outturns and prospects in
to gradually abate, resulting in a steady rise in
high-income countries
long-term rates in the baseline. Indeed, yields
Growth among industrialized economies in
on 10-year U.S. Treasuries have risen 50 basis
2005 is estimated at 2.5 percent, substantially
points since September.
lower than the 3.1 percent recorded the year
However, these temporary factors could
before. Industrial production and trade flows
continue to hold sway, reversing or bringing
among high-income countries were particu-
to a halt the recent increase in long-term rates
larly weak. Growth rates of the former de-
(as they have in the past). This would prompt
clined from over 5 percent in mid-2004 to less
stronger-than-projected demand, but also ex-
than 1.5 percent in the middle of 2005 (figure
acerbate capacity constraints. As a result, oil
1.1). High oil prices, rising short-term interest
prices could get pushed higher, which would
rates, and an unusually disruptive hurricane
provoke a more brutal inflationary cycle, and
season1 slowed growth in the United States to
ultimately, a recession.
Alternatively, these forces could dissipate
more rapidly, causing long-term interest rates
to rise more quickly toward long-term equilib- Figure 1.1 Industrial production
rium levels, which would provoke a more pro- % change, monthly, year over year
nounced slowdown. While not the most likely 15
Developing countries
scenario, the recent rise in long-term yields
and inflation suggest that a higher interest- 10

rate scenario is a real possibility.

Finally, this environment of slowing
growth and global imbalances raises the risk 0
of rising protectionism. In this regard, policy- Developing
⫺5 World
makers need to make a concerted effort to en- countries
(excluding China)
(excluding China)
sure that the Doha round reaches a successful ⫺10
conclusion so that developing countries spe- 1997 1998 1999 2000 2001 2002 2003 2004 2005
cializing in the export of agricultural products Source: World Bank.
can benefit from trade liberalization in the

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 1.1 The global outlook in summary

Percentage change from previous year, except interest rates and oil price

2003 2004 2005e 2006f 2007f

Global Conditions
World trade volume 5.9 10.2 6.2 7.0 7.3
Consumer prices
G-7 countriesa,b 1.5 1.7 2.2 2.0 1.7
United States 2.3 2.7 3.4 3.0 2.4
Commodity prices (USD terms)
Non-oil commodities 10.2 17.5 11.9 ⫺5.9 ⫺6.3
Oil price (US$ per barrel)c 28.9 37.7 53.6 56.0 51.5
Oil price (percent change) 15.9 30.6 42.1 4.5 ⫺8.0
Manufactures unit export valued 7.5 6.9 2.4 2.4 2.1
Interest rates
$, 6-month (percent) 1.2 1.7 3.8 5.0 5.2
€, 6-month (percent) 2.3 2.1 2.2 2.1 2.8

Real GDP growthe

World 2.5 3.8 3.2 3.2 3.3
Memo item: world (PPP weights)f 3.9 5.0 4.4 4.3 4.4
High income 1.8 3.1 2.5 2.5 2.7
OECD countries 1.8 3.0 2.4 2.5 2.7
Euro area 0.7 1.7 1.1 1.4 2.0
Japan 1.4 2.6 2.3 1.8 1.7
United States 2.7 4.2 3.5 3.5 3.6
Non-OECD countries 3.7 6.3 4.3 4.2 4.0
Developing countries in 5.5 6.8 5.9 5.7 5.5
East Asia and Pacific 8.1 8.3 7.8 7.6 7.4
Europe and Central Asia 6.1 7.2 5.3 5.2 5.0
Latin America and Caribbean 2.1 5.8 4.5 3.9 3.6
Middle East and N. Africa 5.2 4.9 4.8 5.4 5.2
South Asia 7.9 6.8 6.9 6.4 6.3
Sub-Saharan Africa 3.6 4.5 4.6 4.7 4.5

Memorandum items
Developing countries
excluding transition countries 5.3 6.8 6.1 5.8 5.6
excluding China and India 4.1 6.0 4.9 4.7 4.6

Note: PPP ⫽ purchasing power parity; e ⫽ estimate; f ⫽ forecast.

a. Canada, France, Germany, Italy, Japan, the UK, and the United States.
b. In local currency, aggregated using 1995 GDP weights.
c. Simple average of Dubai, Brent, and West Texas Intermediate.
d. Unit value index of manufactured exports from major economies, expressed in U.S. dollars.
e. GDP in 1995 constant dollars;1995 prices and market exchange rates.
f. GDP measured at 1995 PPP weights.

an estimated 3.5 percent, compared with 1.2 percent (1.1 percent in the euro zone), was
4.2 percent the year before. The slowdown much weaker. The relatively low oil-intensity
was not as marked as it could have been, of European economies and relaxed macro-
because low long-term interest rates boosted economic policy stance help explain why the
domestic demand, and the cumulative effect slowdown in Europe was not more pro-
of past dollar depreciations improved net nounced. In Japan, GDP is estimated to have
exports. increased 2.3 percent. Rising domestic de-
In Europe, the growth slowdown was less mand and household incomes, as a result of
pronounced, but the expansion, at an estimated tighter labor market conditions and reduced


industrial restructuring, compensated for

much slower Chinese import demand. Figure 1.2 A sharp slowdown
Looking forward, the increase in oil prices % change, GDP
observed in 2005 is expected to slow global 7
Developing countries
growth by about one quarter of a percentage 6
point in 2006, compared with what it would 5
have been had prices remained stable. In the
United States, the pace of the expansion is
projected to remain broadly stable, because 3
the negative effects of further expected in- 2
creases in interest rates and high oil prices will
be partly offset by a deficit-financed pickup in











post-hurricane investment and additional in-





creases in the contribution of the external sec- Source: World Bank.
tor to growth. In Europe, economic activity is
projected to accelerate despite a significant
drag on growth from high oil prices, because
of low interest rates, pent up investment again outperform high-income economies by a
demand, and a dissipation of most of the wide margin through 2007.
negative consequences following the euro’s
real-effective appreciation. Meanwhile, in Regional outlooks
Japan, the negative consequences of higher oil
prices are expected to be substantially offset Detailed descriptions of economic develop-
by strengthening domestic demand and con- ments in developing regions can be found
tinued supportive macroeconomic policies. in the Regional Outlooks section of http://www.
Developing economy outturns
and prospects The economies of the East Asia and
Despite a slowdown of almost a full percent- Pacific region continued to expand rapidly
age point, growth in developing economies re- in 2005. Regional GDP is estimated to have
mained very robust, at an estimated 5.9 per- increased by 7.8 percent, down from 8.3 per-
cent in 2005 (figure 1.2). In part this reflects cent in 2004. Growth in China remained
the strong performance of China and India, very strong—despite a substantial slowing in
where output continued to expand at rapid both private consumption and investment
rates (in excess of 9 percent and about 7 per- demand—because exports continued to
cent, respectively). The slowdown among the grow rapidly, and import growth declined by
oil-importing countries (excluding China and half. China appears to have been a major
India) was sharper, from 5.6 percent to beneficiary of the expiration of quotas on
4.3 percent.2 At the same time, dwindling textiles (see the global trade discussion
spare capacity in the petroleum sector caused below), which contributed to rapid export
growth in oil-exporting developing countries growth in the first half of the year. Since
to ease from 6.6 percent to 5.7 percent, even then, the re-imposition of quotas by the
though oil revenues continued to rise. United States and the European Union (EU)
High oil prices, rising interest rates, and have attenuated this positive force. For other
building inflationary pressures are expected to countries in the region, the slowdown in
restrain growth in most developing regions in Chinese imports, weak global high-tech de-
2006 and 2007 (figure 1.3). As a group, how- mand, and elevated oil prices have translated
ever, low- and middle-income countries should into reduced export growth, rapidly rising

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 1.3 Regional growth

% change, GDP
7 2004 2007

East Asia and Europe and Latin America Middle East South Asia Sub-Saharan Oil importers Oil importers,
the Pacific Central Asia and the and North Africa except China
Caribbean Africa

Source: World Bank.

producer prices, and a deterioration of cur- revenues in oil-exporting countries helped off-
rent account balances. set much slower growth in the oil sector itself.
Even higher oil prices on average in 2006,3 Reflecting these capacity constraints and the
the longer-term implications of reduced in- very strong growth recorded last year, infla-
vestment levels of China, and a tightening of tionary pressures have built up in many coun-
monetary policy are expected to slow regional tries in the region, notably Russia. Turkey,
growth to 7.6 and 7.4 percent in 2006 and where improved macroeconomic policy has
2007, respectively. The changes in the cur- pushed inflation below 10 percent, represents
rency regimes of China and Malaysia are not an important exception. The expected acceler-
expected to have a major impact on growth. ation of demand in Europe, continued high oil
Nevertheless, as discussed below, these prices—which for many countries in the re-
regimes should improve financial stability gion are a positive factor—and additional
both domestically and internationally. gains in European market share, suggest that
Economic activity in the Europe and Cen- growth for the region as a whole should re-
tral Asia region decelerated sharply in 2005, main relatively stable—at about 5 percent in
with GDP growing by an estimated 5.3 per- 2006 and 2007, which is close to the region’s
cent, down from 7.2 percent in 2004. Slower potential growth rate.
increases in oil production, a peaking of the Economic activity in Latin America and the
investment cycle (especially among economies Caribbean is estimated to have increased by
that recently joined the EU), and less robust some 4.5 percent during 2005, substantially
world demand for the region’s exports con- slower than the 5.8 percent recorded in 2004
tributed to the slowdown, which was particu- but much faster than the region’s 0.4 percent
larly intense in a number of the larger average growth rate during the preceding
economies of the region. Russia decelerated three years. Supply constraints and tight mon-
from 7.2 percent to 6.0 percent; Ukraine from etary policy are estimated to have slowed GDP
12.1 percent to 4.4 percent; Poland growth in Brazil to some 3.8 percent (down
from 5.4 percent to 3.5 percent; and Turkey from 4.9 percent in 2004), while in Mexico
from 8.9 percent to 4.8 percent. five fewer working days in 2005 than in 2004
Higher oil prices constrained domestic are expected to contribute to a significant
demand in oil-importing countries, but oil slowing.4 Excluding these countries, regional


growth in 2005 is estimated at a robust the ATC. The region’s strong performance re-
5.9 percent, boosted by both strong world flects, in part, past reforms, such as steps to
demand for the region’s exports (particularly improve transparency in the oil sector in
oil, coffee, and copper, which account for 65 Algeria, as well as banking-sector reform, re-
percent of the regions’ commodity exports) ductions in customs duties, privatization, and
and low interest rates. Domestic factors that regulatory reform in other Maghreb countries.
contributed to the strong performance include These efforts, and in particular, the substantial
past efforts to open the region up to interna- reforms underway in Egypt, help to raise the
tional trade, more responsible budget policy, region’s growth potential by improving both
the introduction of more flexible exchange infrastructure and the overall investment cli-
rate regimes, and lower inflation. mate. While heartening, the pace of reform
Slower global growth is already easing ten- outside of Egypt appears to have waned, per-
sions in the non-oil commodity markets that haps because high oil prices have reduced
have driven the recovery in the Latin America the sense of urgency attached to reform in oil-
and Caribbean region, and this trend is ex- exporting countries.
pected to continue. Moreover, while many In contrast to the slowdown elsewhere in
countries in the region benefit from high oil the world economy, growth in South Asia is
prices, many others, particularly those in the estimated to have picked up a bit in 2005,
Caribbean, are heavily oil dependent and face coming in at 6.9 percent, compared with
substantial income losses.5 As a result, re- 6.8 percent in 2004. This mainly reflects im-
gional GDP growth is projected to decline to proved performance in Pakistan, where GDP
3.6 percent by 2007. is estimated to have increased 8.4 percent (up
High oil prices and strong oil demand con- from 6.6 percent in 2004), thanks to a broad-
tinue to be key drivers for the economies of based acceleration in the manufacturing and
the Middle East and North Africa, where GDP agricultural sectors. Like Pakistan, other
is estimated to have increased by 4.8 percent countries in the region have enjoyed very
in 2005. Very high oil revenues generated strong export performance, in part because of
double-digit advances in public spending, the recent removal of ATC quotas. However,
which have helped to increase GDP in oil- the sharp rise in oil prices and solid regional
producing economies by an estimated 5.4 per- growth over the past several years have
cent. Strong demand from these economies contributed to an acceleration of inflation.
spilled over to the labor-abundant economies Addressing this issue will require a further
of the region through higher remittances and tightening of monetary policy, which, in com-
increased intraregional tourism flows. How- bination with rising oil bills, is expected to
ever, weak growth in Europe, high oil bills, result in a modest deceleration of economic
and a one-off negative effect from the removal activity to about 6.3 percent by 2007.
of quotas under the Agreement on Textiles GDP in Sub-Saharan Africa is estimated to
and Clothing (ATC) reduced growth of re- have increased 4.6 percent in 2005, bolstered
gional oil-importing countries from 4.6 per- by very strong growth among resource-rich
cent in 2004 to about 4.0 percent in 2005. countries. Output in South Africa, the region’s
Looking forward, high oil prices are ex- largest economy, is estimated to have acceler-
pected to continue feeding demand in oil- ated to 4.2 percent, lifted by high metal prices,
producing countries, whose economies should strong confidence, low nominal interest
expand by 5.4 percent in 2006 and 5.1 percent rates and the rand’s recent depreciation.
in 2007. In the oil-importing economies, The economies of oil-exporting countries,
growth is expected to accelerate to about the including Nigeria (the region’s second largest
same level, supported by stronger European economy), grew an estimated 5.5 percent in
growth and a weaker negative effect from 2005, reflecting rapid increases in petroleum

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

production and investment inflows. Growth single digits as a result of lower food prices
in some oil-exporting countries may exceed and prudent monetary policies. Recent eco-
25 percent in 2006 and 2007, as new oil fields nomic reforms, and increased donor sup-
come on stream. However, the pace of the port—as more countries reach the Heavily In-
expansion will taper off in other countries as debted Poor Country (HIPC) completion
they reach capacity constraints. point—will also help support growth, which is
In West Africa, strong commodity prices in projected to be at or above 4.5 percent over
2005, improved rainfall, and more vigorous the medium term.
use of insecticides are expected to lift regional
growth. In East and Southern Africa the ex-
pansion is projected to slow somewhat, partly Long-term prospects and poverty
because the removal of quotas under the ATC forecast
will continue to put textile exports under pres-
sure. Political strife and insecurity in Côte
d’Ivoire and the Great Lakes region are likely
T he recent strong economic performance of
developing economies and the relatively
rapid growth projected for these economies
to impact growth there. Countries are increas- over the medium term owe much to the eco-
ingly passing higher crude-oil prices through nomic reforms undertaken over the past sev-
to consumers with the aim of containing bud- eral years. Improved macroeconomic policies,
get deficits but will cut into consumer demand reflected in lower inflation, trade liberalization
and add to inflationary pressures. (average tariffs have fallen from 30 percent to
The balance of payments and economic less than 10 percent since the 1980s), more
consequences of higher oil prices are expected flexible exchange rate regimes, and lower fiscal
to intensify over the next year as other com- deficits have reduced uncertainty and im-
modity prices, which have attenuated the proved the overall investment environment.
terms-of-trade impact of high oil prices, ease. More microeconomic structural reforms, such
Despite higher oil prices and increased pass- as privatization and regulatory reform initia-
through, inflation is expected to remain in the tives, have also played a key role.

Table 1.2 Long-term prospects

Real GDP per capita, annual average percentage change


Medium-term Long-term
1980s 1990s 2001–06 2006–15

World Total 1.3 1.2 1.5 2.1

High-income countries 2.5 1.8 1.6 2.4

OECD 2.5 1.8 1.6 2.4
United States 2.3 2.0 1.8 2.5
Japan 3.4 1.1 1.1 1.9
European Union 2.1 1.8 1.4 2.3
Non-OECD 3.5 4.0 2.0 3.5

Developing economies 0.7 1.5 3.7 3.5

East Asia & Pacific 5.8 6.3 6.4 5.3
Europe & Central Asia 0.9 ⫺1.8 5.0 3.5
Latin America & Caribbean ⫺0.9 1.6 1.2 2.3
Middle East & North Africa ⫺1.1 1.0 2.5 2.6
South Asia 3.3 3.2 4.5 4.2
Sub-Saharan Africa ⫺1.1 ⫺0.5 1.8 1.6

Source: World Bank.


These factors are expected to contribute population living in extreme poverty is ex-
to better long-term growth performance pected to decline in all developing regions.6
as compared with past decades (table 1.2). With the exception of Sub-Saharan Africa,
Consistent with recent improvements in eco- all regions are expected to achieve their
nomic performance, per capita incomes in de- Millennium Development Goal of reducing
veloping countries are projected to grow poverty by 50 percent from its 1990 level. In
some 3.5 percent a year, more than twice as East Asia, the target has already been
fast as the 1.5 percent growth rates recorded achieved. Moreover, based on the current
during the 1990s. Projected future growth long-term forecast, extreme poverty would be
rates are higher than during the 1980s and almost eliminated by 2015 in both the East
1990s in every developing region except East Asia and Pacific and the Europe and Central
Asia, where they are expected to decline Asia regions. Overall, the number of people
somewhat due to an aging population. living on $1 a day or less will fall to around
Table 1.3 reports poverty projections 620 million, from 1.2 billion in 1990 and an
based on these real per capita income growth estimated 1.0 billion in 2002.
rates and the (re)distribution of income Despite these heartening prospects, there is
within the population. The table indicates no room for complacency. The percent of the
that over the next 15 years the share of the population in developing economies living at

Table 1.3 Regional breakdown of poverty in developing countries

Millions of people living on

less than $1 per day less than $2 per day

Region 1990 2002 2015 1990 2002 2015

East Asia and the Pacific 472 214 14 1,116 748 260
China 375 180 11 825 533 181
Rest of East Asia and the Pacific 97 34 2 292 215 78
Europe and Central Asia 2 10 4 23 76 39
Latin America and the Caribbean 49 42 29 125 119 106
Middle East and North Africa 6 5 3 51 61 40
South Asia 462 437 232 958 1,091 955
Sub-Saharan Africa 227 303 336 382 516 592
Total 1,218 1,011 617 2,654 2,611 1,993
Excluding China 844 831 606 1,829 2,078 1,811

Percent of population living on

less than $1 per day less than $2 per day

Region 1990 2002 2015 1990 2002 2015

East Asia and the Pacific 29.6 14.9 0.9 69.9 40.7 12.7
China 33.0 16.6 1.2 72.6 41.6 13.1
Rest of East Asia and the Pacific 21.1 10.8 0.4 63.2 38.6 11.9
Europe and Central Asia 0.5 3.6 0.4 4.9 16.1 8.2
Latin America and the Caribbean 11.3 9.5 6.9 28.4 22.6 17.2
Middle East and North Africa 2.3 2.4 0.9 21.4 19.8 10.4
South Asia 41.3 31.3 12.8 85.5 77.8 56.7
Sub-Saharan Africa 44.6 46.4 38.4 75.0 74.9 67.1
Total 27.9 21.1 10.2 60.8 49.9 32.8
Excluding China 26.1 22.5 12.9 56.6 52.6 38.6

Source: World Bank.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

or below $2 per day is projected to remain dis-

turbingly high. Moreover, notwithstanding Figure 1.4 Dollar-euro interest rate
that inroads have been made recently, the inci-
dence of extreme poverty in Sub-Saharan Percentage points
Africa in 2002 was actually higher than in
1990. While current projections suggest 8 per- 1.5 Long-term
cent of the subcontinent’s population will be 1.0
lifted above the extreme poverty line by 2015, 0.5
some 38 percent of Africans will still be living 0
in extreme poverty. Worse, the absolute num- ⫺0.5
ber of Africans living at or below the $1-a-day Short-term
level is projected to increase. And, because per difference
capita incomes elsewhere are projected to
grow faster, the continent will continue to fall ⫺2.0

Ap 003

Ju 03

n. 3
Ap 04

Ju 04


Ap 05

Ju 05
farther behind the rest of the world—unless















steps are taken to greatly improve economic




growth in Africa. Source: World Bank.

International finance

T he significant adjustments of international

exchange rates over the past several years
paused in 2005. In particular, notwithstanding
inflows in the first quarter of 2005 as in-
vestors adjusted their portfolios. Not only did
these inflows help strengthen the dollar, they
the persistence of the U.S. current account also financed a large share of the U.S. current
deficit (expected to exceed $750 billion this account deficit (figure 1.5). As a result, the
year), the dollar’s trend decline with respect to dollar was much less reliant on the accumula-
major currencies came to an end. Initially, the tion of reserves by foreign central banks (for-
currency appreciated against its trading part- eign official asset purchases) than in 2004.
ners by some 3.5 percent in real-effective terms In the second quarter of 2005, however,
as of July 2005. It then lost value in August private inflows eased, and foreign central
and September, before showing signs of banks once again assumed a large role in the
strengthening in October. financing of the dollar. Moreover, toward the
The strengthening of the dollar during the end of July the dollar came under renewed
first seven months of 2005 is partly explained downward pressure and depreciated some
by rising U.S. short-term interest rates (as the 1.7 percent in real-effective terms during
Federal Reserve Bank continued its policy of August and September. The dollar began to
gradual tightening) and falling long-term rates appreciate again only after the long-term
in Europe (possibly in response to the conti- interest rate started to rise again. By October
nent’s relatively weaker economic perfor- 2005, the long-term interest rate differential
mance). By July, these developments had had widened to about 120 basis points.
generated a 300 basis-point swing in the dif- The apparent sensitivity of the dollar and
ference between U.S. and European short the financing of the U.S. current account
rates, along with a 75 basis-point gap in favor deficit to interest-rate differentials highlight
of long-term U.S. bonds (figure 1.4). the problems posed by the large financing
These growing interest rate differentials in- requirements of the U.S. current account
creased the financial incentive to hold dollar- deficit.
versus euro-denominated assets, temporarily Until global imbalances are resolved, the
producing stronger net private sector capital dollar is likely to continue to come under


Figure 1.5 Financing of the U.S. current Figure 1.6 Emerging market spreads
account deficit Basis points
$ billions, annualized 1,000
600 EMBI global bond spreads










2004 2005 2005








Q1 Q2

Other private flows Private sector Source: World Bank; Datastream.

treasury bills
Official assets Net equity and FDI

Source: U.S. Department of Commerce; BEA; U.S.

Treasury Bulletin. managed. While technically the announced
rules could allow the renminbi to depreciate as
much as 9 percent per month, similar possibil-
ities for flexibility existed under the former
downward pressure, unless foreign central regime but were not exercised.
banks accumulate substantial quantities of
dollars or interest-rate differentials widen fur-
ther. In the baseline, interest rate differentials Interest rates and spreads remain low
are projected to widen further, and the dollar The recent period of very low real interest
is projected to decline gradually, falling by rates has been particularly beneficial to devel-
about 5 percent per year. Should central banks oping economies. Together with narrower
cease to be willing to accumulate reserves at risk premia (figure 1.6), low rates have al-
current rates, there could be a disruptive hike lowed developing countries to reduce their
in interest rates or a more precipitous fall in financing costs, restructure their debt, and
the dollar (World Bank 2005). pursue strong investment growth. Early re-
The recent decision of the Chinese and payment of Paris Club debt has already
Malaysian authorities to move from an ex- reached $22 billion in 2005, and among
change rate regime linked to the dollar alone emerging-market economies, virtually all
to one focusing on a basket of currencies rep- financing requirements for this year had been
resents a major and welcome move toward a met by August.8
more flexible currency regime. While it will Short-term rates have been rising, and
not resolve current account imbalances, it they can be expected to continue to rise as
should increase the stability of the renminbi monetary policy tightens, initially in the
and ringitt with respect to the currencies of United States, but eventually in Europe as
their trading partners (other than the United well. In contrast and notwithstanding re-
States) and reduce the amount by which the cent increases, longer-term interest rates
dollar would have to depreciate relative to have remained low longer than expected
other currencies to achieve a given level of ad- (figure 1.7), while spreads on more risky
justment.7 How effective the new regimes will emerging market and corporate assets have
be, depends importantly on how they are fallen even further.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 1.7 Real long-term interest rates in Figure 1.8 World savings rate
G-7 countries % of nominal GDP a
Percent 25
8 Period Global
average 24 savings rate

4 23

21 Average
⫺2 GDP-weighted average savings rate
of the 10-yeara government
⫺4 20
bond yields
⫺6 1981 1986 1991 1996 2001 2005
1970 1975 1980 1985 1990 1995 2000 2005
Source: World Bank.
Source: World Bank. a. Sum of national savings divided by the sum of national
Note: For data prior to 1972, exclude Italy, a 10-year or GDP expressed in U.S. dollars at market exchange rates.
nearest government bond yields.

Many reasons for these low interest rates activity, principally in the developed world, has
have been proposed (see IMF 2005 for a re- failed to keep pace with savings as they have
cent overview), including the following: returned to historical levels (see IMF 2005).
Most of these explanations for lower long-
• Excess liquidity stemming from an ex-
term rates involve temporary factors, imply-
tended period of very low short-term
ing that long-term rates will eventually rise
interest rates in almost all developed
toward their long-run equilibrium level9 (fre-
quently defined as the long-run potential
• A low inflation environment, thanks to
growth rate of the economy). In this context,
improved credibility of monetary policy,
the question is not so much why long-term
and the disinflationary impact of in-
rates are low, but how much longer they will
creased competition following the entry
remain so. In the baseline, increased invest-
into global markets of China and mem-
ment in Europe and tighter monetary policy
bers of the former Soviet bloc.
result in a gradual rise in interest rates, which
• An increase in global savings, due to
will nevertheless remain below recent esti-
– increased savings in Europe following
mates of the long-term growth potential of the
heightened recognition of the need to
U.S. economy. The final section of this chapter
prepare for the impending retirement
explores some of the economic implications
of the baby-boom generation; and
should interest rates stay low for an extended
– increased corporate savings in dy-
period of time or, alternatively, should they
namic East Asia (caused by corporate
rise more quickly than anticipated.
restructuring following the currency
crisis) and in the United States (follow-
Signs of rising inflation
ing the stock market decline in 2000).
Low interest rates have contributed directly to
However, while global savings have in- the strong economic performance of recent
creased recently, this follows a period where years. Growth has, in turn, provoked a pickup
they declined substantially, making it difficult in inflation in many developing countries. The
to argue that the world savings rate is currently largest hikes have been in commodity prices
too high (figure 1.8). Rather, investment (see below). However, producer price inflation


has jumped by more than 4 percentage points productivity growth suggest that core infla-
in some regions and exceeds 5 percent in tion, which has been more stable, may begin
every developing region except Sub-Saharan to rise soon. In Europe, high oil prices have
Africa.10 limited disinflation despite significant slack
Consumer price inflation has also been and the appreciation of the euro.
rising (if less spectacularly). Weighted by These same factors should continue to limit
GDP, aggregate inflation among developing price inflation in Europe. However, in the
economies increased from 4.0 percent in the United States, high oil prices plus the pro-
fourth quarter of 2003 to 5.4 percent by July jected further depreciation of the dollar are
of 2005. It has since eased somewhat. Region- expected to generate additional upward price
ally, inflation has picked up strongly in South pressure.
Asia, Sub-Saharan Africa, and East Asia (fig- Low interest rates have resulted in higher
ure 1.9). Inflation in developing countries prices of interest-sensitive assets in markets
is projected to continue rising in 2005, as with strong financial intermediation—notably
growth remains at or above trend rates, and in the United States and some European
the pass-through from high oil prices contin- countries—contributing to strong consumer
ues to exert upward pressure on prices. demand (World Bank 2005, IMF 2005). As in-
In high-income countries, there are only terest rates rise, housing prices are expected to
limited signs of rising inflation. In the United plateau and even decline, which has already
States, where output is close to potential, in- begun in the United Kingdom. As they do so,
flation has been rising steadily. It jumped to the rate at which household wealth increases
4.7 percent in September 2005, but the in- will moderate and its contribution to con-
crease is not expected to be permanent, be- sumer demand should abate.11
cause it reflects very high gasoline prices that Data indicate that house prices have also
month, which have since declined. Neverthe- been rising rapidly in a number of middle-
less, data pointing to rising wages and lower income countries, such as Bulgaria, India,

Figure 1.9 Inflation rates

Annual % change
7 2004
6 Most recent available dataa

East Asia Europe and Latin America Middle East and South Asia Sub-Saharan OECD
and Pacific Central Asia and the Caribbean North Africa Africa

Source: World Bank.

a. Data concern August 2005, except for East Asia (September), the Middle East and North Africa (June), and Sub-Saharan Africa

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 1.10 Cumulative real increase in Figure 1.11 Commodity prices

housing prices, 2005 Index, 1990 ⫽ 100
Percent 300
100 Metals and
Energy Agricultural
80 200 products


2000 2001 2002 2003 2004 2005 2006
Source: World Bank.

(B ha a

na a
So ba ia



gk nd
um Ind


ut i)

an ila
ni ok)







Although metals and minerals prices rose


Source: World Bank; BIS. during the first months of the year, they have
a. Data for Russia and China reflect increases between 2000
and 2004; data for other countries reflect increases between
since stabilized, and in October 2005, they
2000 and 2005. were at the same level as in March 2005.
Conditions in some metals and minerals mar-
kets remain tight, due to low inventories. In
Indonesia, Malaysia, and South Africa (fig- the case of copper and aluminum, prices re-
ure 1.10). While fast economic growth and main elevated (partly reflecting higher energy
changes in the regulatory environment have content in the production of these goods).
certainly played a role in these countries, so Demand has weakened markedly for lead, tin,
have low interest rates. Unfortunately, data and zinc.
limitations prevent a thorough analysis of the Analysis of past non-oil commodity cycles
causes and consequences of rising housing suggests that this one may have run its course.
prices in low- and middle-income economies.12 Already it distinguishes itself from previous
episodes by having lasted longer, in part,
because energy prices have also been high,
Commodity markets which was not always the case during previ-

A fter several years of rising commodity

prices, there are indications of a stabiliza-
tion and even reversal of gains in the markets
ous episodes. In so far as high fuel prices in-
crease production costs in both agriculture
and metals and minerals, they may have re-
for agricultural products and for metals and duced the supply response, keeping prices
minerals (figure 1.11). higher longer.
Agricultural prices have been declining In line with the projected slowdown in
most of this year and are down 5 percent since global growth and increased supply, prices of
March 2005. However, prices of agricultural agricultural products and metals and minerals
raw materials are rising, partly because of are projected to decline somewhat in 2006.
higher prices for commodities that are close
substitutes for crude oil-based products (for Limited spare capacity to keep
example, natural rubber prices are up 41 per- oil prices high
cent because of increases in synthetic rubber In contrast with other commodity prices,
costs). oil prices continued to strengthen during the


first nine months of 2005. During this period, refining capacity, particularly of lower-quality
they averaged some $52 per barrel, a 38 per- crude oil.
cent increase compared with the average for Spare production capacity is now about
2004. These increases occurred despite an eas- 2 million barrels per day (mbpd), compared
ing of conditions in the oil market. Demand with almost 6 mbpd three years ago, and ca-
growth slowed from more than 3.5 percent in pacity will remain tight over the near term.
2004 (the highest growth since the late 1970s) This reflects long lags in bringing significant
to a 1.4 percent annualized rate during the new quantities of oil into production15 and
first three quarters of 2005. As a result, supply shorter lags before demand substitution can
is actually increasing faster than demand,13 have an effect.16 Moreover, approximately
and inventories have begun to accumulate, half of the expected new capacity is being pro-
although they remain low.14 duced by OPEC, suggesting that the organiza-
Rising prices over the first eight months of tion will continue to exercise significant mar-
the year reflected the market’s concern that ket power over the near term.
existing spare capacity was insufficient to deal In this environment, prices are likely to re-
with a major disruption to supply or an in- main volatile, as small events or even minor
crease in demand (figure 1.12). In some sense, changes in expectations may provoke signifi-
hurricane Katrina was the kind of serious cant price swings. As a result, the World Bank
shock the market feared. Although oil prices has adopted a technical assumption for the
spiked briefly to more than $70 a barrel, they future path of oil prices based on a slow de-
are back below $60, following the release of cline toward $40 per barrel by 2010. This
some 29 million barrels of crude oil from the implies an average price of $56 in 2006 and
stockpiles of the International Energy Agency $52 in 2007, which is somewhat higher than
and the U.S. government. Moreover, gasoline the current consensus forecast. The economic
prices in the United States have returned to the consequences of alternative scenarios, notably
levels observed before hurricane Katrina, and a sharp negative supply shock, are discussed in
market concerns have switched from a focus the final section of this chapter.
on inadequate oil supply to insufficient
The impact of oil prices on
developing economies
The world economy in general and developing
Figure 1.12 Levels of spare oil capacity countries in particular have shown consider-
able resilience to higher oil prices. This reflects
Million barrels per day
increases in non-oil commodity prices and a
very robust global economy, which have, until
recently, muted the impact of higher oil prices.
6 The first round of oil price hikes (1999–
5 2000) adversely affected low- and middle-
income countries. The price increase was very
large in percentage terms (rising from just
3 under $12 to almost $30 per barrel between the
2 first quarter of 1999 and the end of 2000) but
smaller than the most recent increases in dollar
terms (table 1.4). Current account deficits
Jan. July Jan. July Jan. July Jan. July
among low-income oil-importing African
2002 2002 2003 2003 2004 2004 2005 2005 countries increased by 0.5 percent of GDP on
Source: World Bank.
average.17 Moreover, government deficits in
those countries that did not pass on the price

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 1.4 Terms-of-trade impacts non-oil commodity prices. Indeed, the esti-
of commodity price changes mated terms-of-trade shock from price move-
ments since January 2004 is more than three
1999–00 2001–03 2004–05
times as large for various groups of
Cumulative price change low-income countries than the cumulative
Oil 120.3 18.9 88.0 shock over the preceding three years. As a re-
Agricultural products 0.3 15.7 8.9
sult, non-oil imports from poor, current
Metals and minerals 25.0 10.2 47.9
Manufactures ⫺5.0 3.0 10.4 account–constrained countries are expected to
Total terms-of-trade effect (% of GDP)
come under pressure in the coming months.
Oil Importers Moreover, the impact on oil-importing poor
Low and Middle income ⫺1.8 ⫺0.1 ⫺0.9 countries could be significantly aggravated if
Low income ⫺3.8 ⫺0.9 ⫺2.9
Sub-Saharan Africa ⫺2.5 1.4 ⫺1.2
oil prices remain at or close to current levels
South Asia ⫺3.9 ⫺1.5 ⫺2.7 and if non-energy commodity prices return to
Highly indebted ⫺4.3 1.5 ⫺3.3 pre-shock levels.
poor countries

Source: World Bank.

Note: Periods Jan. 1999–Dec. 2000, Dec. 2000–Dec. 2003, World trade
Dec. 2003–July 2005.

T he expansion of world trade slowed sig-

nificantly during 2005 (figure 1.13).
While merchandise exports were growing at a
hikes rose by about the same amount.18 Among 16 percent or more annualized pace in the
those countries with limited access to interna- middle of 2004, they subsequently slowed and
tional finance, non-oil imports fell by 3.8 per- were expanding at an 8.5 percent pace during
cent in 2000, partly because insufficient foreign the third quarter of 2005.
exchange was available to finance imports at
previous levels.
During the second bout of (more gradual)
price increases (2001–3) the same countries
performed much better. Current account posi- Figure 1.13 World trade volumes
tions actually improved as a percent of GDP, Annual growth rates, 6-month moving average
and there were no discernible slowdown in 40
non-oil imports. Part of this improved perfor- 36
mance is explained by real exchange rate
movements (which diminished the domestic
currency cost of the oil-price hike) and by a 28

number of non-oil commodity prices that also 24 World

increased rapidly during that time, thus pro- 20
Other developing
viding the necessary foreign currency to meet
the additional oil burden without cutting into
non-oil imports. On average, high non-oil 12

commodity prices reduced the negative terms- 8

of-trade shock from higher energy costs by High-income
4 countries
more than half.
Drawing from this experience, the impact Jan. July Jan. July Jan. July
of the latest hike in oil prices on poor oil im- 2003 2003 2004 2004 2005 2005
porters is a concern, principally because it has Source: World Bank.
not been accompanied by as much strength in


Most of the deceleration concerned the ex-

Figure 1.14 Change in textile exports to
ports of high-income economies, volumes of
the developed world, first half of 2005
which grew by less than 4 percent (annualized)
in the first quarter of 2005, before strengthen- Tajikistan
ing more recently. Merchandise export volumes
of developing countries (excluding China) were Kenyaa
relatively robust, increasing at an estimated Philippines
12 percent pace toward the middle of 2005.
Chinese exports, boosted by the removal of Mauritius
ATC quotas, grew at a 24 percent pace. South Africa
Commodity markets have significantly
shaped developments in world trade. The mer- Lesothoa
chandise exports of oil-exporting countries, Madagascar
which had been rising at some 5.8 percent in
2004, increased by an estimated 5.2 percent Macedonia, FYR
in 2005. The deceleration among developing Pakistan
oil importers (excluding China) was steeper in
percentage terms (from 15 percent to 11 per- Swazilandb
cent), but growth rates remained much higher. Indonesia
Growth in the production of non-oil com-
modities in general is moderating, both be- Lao PDR
cause demand is easing and because of supply Egypt
As a reflection of the slowdown already ob- Perua
served, international trade is forecast to slow Jordan
down relative to 2004 as a whole. Merchan-
Sri Lanka
dise trade volumes are expected to increase India
by around 7.7 percent. The goods and service Chinab
trade is expected to increase 6.2 percent in ⫺60 ⫺40 ⫺20 0 20 40 60
2005 before strengthening somewhat in % change, year over year, $ values

2006–7. Source: World Bank, IMF, U.S. Department of Commerce,

Exports of developing economies continue and EuroStat.
a. Data refer to the first five months of each year only.
to be heavily influenced by developments in the b. Data for China include Hong Kong and Macao.
volatile high-tech market. After falling sharply
in the third quarter of 2004, global sales of
semiconductors and other high-tech products
picked up before weakening once again in the January 2005 (figure 1.14). While the removal
second quarter of 2005. This volatility is ap- of quotas was done in phases, and ten years of
parent in East Asian export volumes (high-tech adjustment time provided, backloading of the
products represent as much as two-thirds of the removal of quotas meant that they were still
exports of some economies in this region).19 binding when they were finally removed. As a
High-frequency data suggest a strengthening of result, there were significant changes in pat-
demand for these products, implying a pick up terns of trade among affected goods in 2005,
in export flows from the region. most notably in the form of increased exports
Trade growth for some countries was from China (and other countries, whose mar-
heavily influenced by the removal of quotas ket share had been artificially held back under
under the ATC of the Multifiber agreement in the old quota scheme) to the detriment of

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

income imports over the same period. The tex-

Figure 1.15 Estimated change in textile
tile sectors in Kenya, Myanmar, Nepal, the
exports as share of total merchandise
Philippines, and Tajikistan, on the other hand,
saw the dollar value of their exports decline by
Bangladesh 4 or more percent.
However, many of these countries are not
Sri Lanka
large exporters of textiles. As a result, these
Jordan figures may exaggerate the overall economic
India impact of the relaxation of textile quotas.
Expressed as a percent of these countries’
total merchandise exports, the biggest gains
Chinaa were experienced by Bangladesh, Cambodia,
Bulgaria Jordan, India, Pakistan, Sri Lanka, and
Turkey, while the largest losses were those of
Egypt, Arab Rep.
Macedonia, FYR
Kenya, Nepal, Myanmar, Mongolia, and
Madagascar Tajikistan (figure 1.15).
Albania Risks and uncertainties
L ow interest rates, moderate inflation, and
the robust growth projected in the baseline
constitute a relatively benign scenario. However,
South Africa
the outlook is dominated by downside risks.
Philippines An oil-market supply shock could cause
serious disruption
Myanmar The most important potential risk comes
Nepal from the oil market. As global demand and
Tajikistan supply are projected to increase broadly in
⫺15 ⫺10 ⫺5 0 5 10 15 step, excess capacity (currently estimated at
1.9 million barrels per day) will remain very
Source: World Bank, IMF, U.S. Department of Commerce, constrained. In this context, the market can be
and EuroStat.
expected to continue reacting to events in a
Note: Data reflect the change in textile and clothing exports
between 2004 and 2005 as a percent of total merchandise relatively volatile manner. Rather than gradu-
exports. ally declining, as in the baseline scenario,
a. Data for China include Hong Kong and Macao.
b. Data refer to the first five months of each year only. prices could remain at current levels, rise fur-
ther, or even fall.
More fundamentally, with spare produc-
tion capacity so low, the market is particularly
those exporters that had benefited most from vulnerable to a supply shock. Because no
the old quota system. country can easily ramp up production, if out-
Using U.S. and European imports of tex- put in another producing country were to fall
tiles as a proxy for developments in the world significantly, world supply would fall, provok-
as a whole, China,20 India, Jordan, Peru, ing a decline in economic activity to the extent
Sri Lanka, and Turkey saw the dollar value of that the global economy could not quickly
their exports increase between the first half of adopt an alternative energy source.21
2004 and the same period in 2005 by more Table 1.5 presents results from a simulation
than the 20 percent average increase in high- of the impact of a 2-million-barrels-per-day


Table 1.5 Impact of a 2 million bpd income and middle-income countries, both be-
negative supply shock cause of higher energy intensities and a greater
inflationary impact, which requires a larger
2006 2007 2008 2009
contraction to eliminate.
Price of oil 90 70 44 40 While the impact in terms of GDP for cur-
(Change from base line) 34 28 3 0 rent account–constrained low-income coun-
Change in GDP % of baseline tries is smaller, it is more severe in terms of
World ⫺1.0 ⫺1.5 ⫺1.1 0.2 domestic consumption and investment. Such
High income ⫺0.7 ⫺1.3 ⫺1.3 ⫺0.3
countries have limited access to international
Middle income ⫺1.6 ⫺1.6 ⫺0.1 1.4
Large low income ⫺1.7 ⫺2.8 ⫺1.8 0.7 capital markets, and their capacity to pay
higher oil prices is limited by their export rev-
Impact on inflation rate
World 2.6 0.6 ⫺0.9 ⫺0.2 enues. If these revenues are stable, they are
High income 1.4 0.0 ⫺1.0 ⫺0.4 forced to reduce domestic demand and non-
Middle income 5.8 2.0 ⫺0.9 0.5
oil imports in order to pay their higher oil
Large low income 2.8 0.9 ⫺0.7 ⫺0.2
bill. As a consequence, when oil prices rise,
Impact on real interest rates (levels)
oil consumption remains relatively constant
World 1.0 0.2 ⫺0.1 0.1
High income 1.0 0.1 ⫺0.2 0.0 in volume terms (being generally inelastic in
Middle income 1.1 0.7 0.2 0.2 the short run), but the oil bill rises. To com-
Large low income 0.5 0.1 0.1 0.4
pensate, non-oil imports and domestic de-
Impact on current account balance (% of GDP) mand tend to decline in unison—leaving GDP
World ⫺1.1 ⫺0.5 ⫺0.1 ⫺0.1
relatively unchanged. For these countries, the
High income ⫺1.1 ⫺0.7 ⫺0.2 ⫺0.2
Middle income ⫺0.9 ⫺0.2 ⫺0.5 ⫺0.3 terms-of-trade shock of the initial increase in
Large low income ⫺1.9 ⫺0.2 1.7 1.0 oil prices is estimated at 4.1 percent of their
Impacts on low-income current account– GDP, which would translate into a 2.7 per-
constrained countries (1) cent decline in domestic demand, with poten-
Terms of trade ⫺4.1 .. ..
tially serious impacts on poverty.
GDP ⫺0.3 0.1 0.0
Domestic demand ⫺2.7 ⫺1.1 0.0
Current account balance ⫺1.2 0.9 0 The future path of interest rates represents
Source: World Bank. an additional source of uncertainty
Note: Impacts on low-income, current account–constrained Persistent global imbalances continue to be a
economies were estimates based on the terms-of-trade impact
using a purpose-built VAR model. Other estimates were
serious source of uncertainty. The current ac-
simulated using the World Bank’s macroeconomic simulation count deficit of the United States and its fi-
nancing requirements are very large, and the
willingness of investors to finance it is sensi-
negative supply shock.22 The disruption is tive to both interest-rate differentials and
assumed to last throughout the projection exchange-rate expectations. As net foreign lia-
period, causing prices to rise to $120 for an bilities accumulate, markets will become in-
initial period of three months before easing creasingly sensitive to adverse shocks or
to $80 for three quarters. Thereafter, supply changes in sentiment, and the dollar is likely
and demand adjustments result in a gradual to come under downward pressure once
decline in oil prices toward $40. again, which would put upward pressure on
Global output responds to the initial shock interest rates.
by contracting, as compared with the baseline, Table 1.6 explores the possible implications
by 1.5 percent of GDP after two years, while of higher interest rates. In this scenario, faced
inflation picks up rapidly. On average, the with sustained downward pressure on the dol-
current account position of oil-importing lar, investors demand higher returns on U.S.-
countries deteriorates by about 1.1 percent of denominated assets to offset further expected
GDP. The impact is more severe in large low- depreciations. This, combined with concerns

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 1.6 Interest rate scenarios prices and consumer wealth. Slower growth
eases inflationary pressure and global ten-
2005 2006 2007 2008 2009
sions, including in the oil market. As mone-
A. A 200 basis-point increase in interest rates and in spreads tary policy loosens in response to increasing
output gaps, growth starts to pick up again,
Interest rates (change of Q4 level from baseline)
World 1.8 1.4 ⫺0.6 0.2 1.3 bringing output back to the levels in the base-
High income 1.7 1.1 ⫺1.1 ⫺0.3 0.9 line by the end of the simulation period.
Low and middle
Higher interest rates would also affect fi-
income 2.0 2.7 1.9 2.6 3.1
nancing conditions for developing countries
GDP (% change from baseline)
by increasing future borrowing costs. For
World ⫺0.1 ⫺1.7 ⫺2.9 ⫺1.9 ⫺0.6
High income 0.0 ⫺1.5 ⫺2.7 ⫺2.5 ⫺1.0 many countries this will not pose a serious
Low and middle short-term risk, because they have taken ad-
income ⫺0.2 ⫺2.4 ⫺3.5 ⫺3.0 ⫺1.5
vantage of low rates to reduce the share of
Inflation (change in inflation rate) short-term debt relative to their overall debt
World 0.0 ⫺0.3 ⫺1.1 ⫺1.1 ⫺0.3
and to prefinance some of their future bor-
High income 0.0 ⫺0.3 ⫺1.5 ⫺1.6 ⫺0.5
Low and middle rowing needs. For others, particularly those
income 0.0 ⫺0.3 0.7 1.2 0.9 with large debt-to-GDP ratios or those that
B. Persistently low interest rates have accumulated large short-term debt posi-
tions (figure 1.16), a rapid rise in interest rates
Interest rates (change of Q4 level from baseline)
World ⫺0.7 ⫺0.5 0.6 0.1 ⫺0.6
High income ⫺0.7 ⫺0.5 0.8 0.1 ⫺0.6
Low and middle
income ⫺0.8 ⫺0.8 ⫺0.1 ⫺0.1 ⫺0.7 Figure 1.16 Some countries are particularly
at risk
GDP (% change from baseline)
World 0.0 0.8 1.4 0.3 ⫺0.5 Debt/GNI ratio
High income 0.0 0.8 1.4 0.4 ⫺0.5 0 50 100 150 200
Low and middle
income 0.0 1.0 1.4 ⫺0.1 ⫺0.7 Argentina
Inflation (change in inflation rate)
World 0.0 0.2 0.6 0.6 0.1 Lebanon
High income 0.0 0.2 0.8 0.8 0.2 Papua New Guinea
Low and middle
income 0.0 0.1 ⫺0.2 ⫺0.5 ⫺0.3
Source: World Bank.

about rising debt and pension liabilities in Turkey

industrialized countries, and a more rapid dis- Philippines

sipation of the temporary factors depressing Jamaica

long-term interest rates, causes them to in- Bolivia

crease by some 200 basis points in high- Paraguay

income countries. Risk premia in developing Brazil
economies increase by an additional 200 basis Pakistan
points as investors’ appetites for risk decline. R.B. de Venezuela
The world economy reacts to the substan- Ukraine
tial tightening of monetary conditions by re- Dominican Rep.
ducing global growth by half for a period Guatemala
of two years, as higher interest rates cut into 0 5 10 15 20 25 30 35
investment and consumption demand, both Short-term debt/Total debt
through classic transmission mechanisms and Source: World Bank.
via the impact of interest rates on housing


could pose a real threat—particularly if the stitution toward alternative energy sources—
rise in base interest rates also provokes a re- to the ultimate detriment of oil-producing
turn of spreads to more usual levels. countries.
Alternatively, if excess liquidity and global In the developed world, efforts to increase
savings prevent long-term interest rates from energy efficiency by developing more fuel-
rising as quickly as in the baseline scenario, efficient technologies, such as hybrid cars,
the resulting higher levels of demand could in- could well pay important dividends. These
crease tensions in commodity markets, includ- technologies are already economic in some
ing the oil market. In addition, lower interest countries, where gasoline is heavily taxed, and
rates could cause a number of economies, in- could generate substantial savings in overall
cluding the United States, to overheat, gener- fuel demand.23 In addition to the ecological
ating additional inflation that forces a further benefits, making such technology available in
tightening of monetary policy. As a result, developing economies, where the increase in
while growth would be initially higher, the transportation-related energy demand is high-
subsequent tightening of policy could provoke est, would be particularly effective in limiting
a stronger-than-projected slowdown. overall demand.
The depth of the cycle would depend im- Finally, efforts to improve cooperation be-
portantly on the extent of the wealth effect tween users and suppliers concerning the qual-
generated by low interest rates on housing ity and transparency of oil market data could
prices. The more pronounced, the deeper the help reduce unwarranted volatility and per-
cycle. Moreover, because asset prices become haps contribute to lower prices by reducing
even more out of step with their long-run the oil-price risk premia.
levels, an even longer period of slow growth To further dissipate the risk from global
could be required to re-establish equilibrium. imbalances, policies need to promote both
public and private savings in countries with
Policy challenges large current account deficits. Recent mea-
Policy can help reduce both the economic sures to tighten fiscal policy in the United
severity of such unfavorable outturns and the States are headed in this direction, but more
likelihood that they will materialize. tightening is required. Tighter monetary pol-
High oil prices will naturally induce substi- icy is helping. Higher interest rates in the
tution toward alternative energy sources and United States promote private sector financing
conservation. In the current context, where of the deficit but also promote private sector
the increase in oil prices is expected to endure, saving. In Europe, policymakers should seek
countries that have not passed on recent price to maintain low interest rates in an effort to
hikes to consumers (and industry) may wish stimulate demand. As output picks up, fiscal
to revise their policies. Not only are the bud- policy (rather than monetary policy) should be
getary costs of such subsidies likely to be dif- used to restrict demand, if necessary. Indeed,
ficult to support, but these policies also im- given unfunded public pension liabilities in
pede adjustment. these countries, such a fiscal tightening is nec-
Moreover, countries with restrictive rules essary in its own right.
concerning the exploitation of oil reserves Developing economies should react flexibly,
might wish to re-examine them. Such policies seeking to maintain real effective exchange
may deny these countries access to technical rates in line with their fundamentals, rather
expertise and financial capital, thereby pre- than a particular alignment with any one cur-
venting them from investing in new produc- rency. In this regard, recent steps by some
tion to the extent that they might otherwise. countries to adopt an exchange rate regime
This may slow the aggregate supply response that reflects their overall trade patterns are
and encourage greater conservation and sub- positive and could be emulated by other

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

countries. Petro-dollars could help reduce the weekends. Occasionally, these fluctuations can have an
likelihood of a disruptive resolution of global important impact on annual GDP growth. While five
fewer days corresponds to roughly a 2.5 percent re-
imbalances if they are recycled into the global
duction in working time, in general, the actual reduc-
economy in a way that reduces tensions. In
tion in production is less pronounced.
particular, the financing of investment expen- 5. These losses are estimated at more than 5 per-
ditures both domestically and in other devel- cent of GDP for Antigua and Barbuda, Belize, Guyana,
oping countries would help stimulate demand Honduras, Nicaragua, and Jamaica.
outside of the United States, reducing that 6. This year’s projections differ somewhat from
country’s current account deficit. To the extent 2004’s partly because of a shift in the base year for cal-
culations from 2001 to 2002, which reduces the
that these funds are invested in U.S. financial
poverty level of the starting year in all regions that
securities, they could also help finance the U.S.
have experienced positive per capita growth. In addi-
current account deficit. tion, new survey data was employed for a large num-
Finally, global weakness could trigger in- ber of countries (more than half in the Europe and
creased protectionism or a slowing of trade Central Asia region), including important new house-
liberalization (which has been the basis for hold surveys in a number of Latin American countries
much of developing countries’ recent success). in the place of labor force survey data used in the past.
Finally, revisions to national income estimates of
The supply disruptions in Europe provoked by
GDP, inflation, and consumption play a role. For more
the re-imposition of quotas on Chinese im-
information concerning the changes to the poverty
ports of textiles, following their liberalization forecast, please visit the Long-term Prospects and
at the beginning of this year, is a good illus- Poverty Forecasts section of http://www.worldbank.
tration of how trade restrictions work to the org/globaloutlook.
detriment of both exporting and importing 7. The extent to which the new regime will con-
countries. Not only should countries resist the tribute to increased stability in world markets will also
depend on the extent to which other Asian currencies
temptation to intervene in already liberalized
follow suit, and how much flexibility is permitted in
domains, concerted efforts need to be made to
achieve meaningful liberalization in the agri- 8. As of August 2005, overall financing for emerg-
cultural and service sectors in the Doha ing market sovereign debt was already 74 percent
process. To date, liberalization has largely funded; this share reached 93 percent for emerging
omitted the politically sensitive agricultural Europe and Turkey, and 100 percent for Latin America.
sector, depriving many developing countries 9. Long-term interest rates tend to be determined
by the long-term growth potential of the economy and
of the benefits from trade liberalization that
expected inflation. Historically, temporary factors have
more manufacturing-oriented economies have
caused them to deviate from this measure, sometimes
enjoyed. for extended periods of time. However, they have al-
ways tended to return to this level.
10. In the first half of 2005, producer price infla-
Notes tion exceeded 15 percent in the Europe and Central
1. The Congressional Budget Office (2005) esti- Asia region, was more than 9 percent in Latin America
mates that hurricane Katrina reduced growth in the and the Caribbean, about 7 percent in the Middle
United States by 0.4 and 0.9 percent (annual rates) in East and North Africa, and around 6 percent in both
the third and fourth quarters. East and South Asia.
2. The importance of China to aggregate statistics 11. The stock of housing in the United States is es-
is also visible in the industrial production data. Growth timated by the Federal Reserve Bank to be equal to
rates for all developing countries showed little slowing, $15.2 trillion, or about 138 percent of GDP. A 10 per-
but excluding China, annualized growth rates declined cent change in the value of that stock would represent
from about 7.5 percent in mid 2004 to less than 5 per- 13.8 percent of GDP, or 19 percent of consumption.
cent a year later. Econometric estimates suggest that the long-term mar-
3. Although oil prices are projected to decline dur- ginal propensity to consume from housing wealth is
ing 2006, they will be higher, on average, than in 2005. 0.05 (see, for example Catte and others 2004 and
4. The number of working days each year varies, Benjamin, Chinloy, and Jud 2004), implying a reduc-
generally because certain holidays do or do not fall on tion in consumption of 1.35 percent.


12. Very few low- and middle-income countries 22. Beccue and Huntington (2005) estimate that
have housing data similar to the data available in high- the probability of such a disruption occurring during
income countries; and what does exist tends to be the next 10 years is high (70 percent for one lasting
limited to wealthy neighborhoods in single cities. 6 months and 35 percent for one of 18 months).
Moreover, there is little information on home owner- 23. In the United States, for example, hybrid cars
ship ratios, and mortgage-market completeness—all currently offer approximately an 80 percent improve-
critical components in determining housing-market ment in fuel efficiency. Were these vehicles to gain a
wealth effects. 10 percent share of new car sales, new energy demand
13. The U.S. Department of Energy estimates that would be reduced by about 12 percent (c. 0.3 percent-
both oil and gasoline consumption fell by 2.5 or more age points) per year for about seven years.
percent in September 2005.
14. In the second quarter, stocks equaled 54 days
worth of consumption versus an average of more than
58 days during the first half of the 1990s. References
15. Some oil fields can be brought into production Beccue, Phillip C., and Hillard G. Huntington. 2005.
within 1 year, but others would take as long as 10 years. “An Assessment of Oil Market Disruption
Three to five years would be needed to bring in two mil- Risks.” In Final Report EMF SR 8. Energy Mod-
lion barrels per day over and above expected increases eling Forum. October.
in demand. Benjamin, J.D., P. Chinloy, and G.D. Jud. 2004. “Real
16. Opportunities for demand substitution may be Estate versus Financial Wealth in Consumption.”
less plentiful than in the 1970s because of the substan- Journal of Real Estate Finance and Economics
tial conservation steps undertaken then. Nevertheless, 29(3): 341–54.
use of fuel-efficient cars and less intensive use of Catte, P., N. Girouard, R. Price, and C. André. 2004.
existing cars can have substantial impacts on overall “Housing Markets, Wealth and the Business
demand. Cycle.” Economics Department Working Paper
17. Simple average of 32 oil-importing Sub- 394. OECD, Paris.
Saharan economies. Congressional Budget Office. 2005. “Macroeconomic
18. Among countries that did not pass prices and Budgetary Effects of Hurricane Katrina.”
through fully, fuel subsidy spending rose substantially, Report. September 6. Washington, DC.
for instance in the Central African Republic, Guinea Huntington, Hillard G. 2005. “The Economic Conse-
Bissau, Malawi, and the Seychelles. quences of Higher Crude Oil Prices.” In Final
19. In 2003, high-tech products represented 13 per- Report EMF SR 9. Energy Modeling Forum,
cent of Thailand’s exports, but more than 50 percent of October.
Taiwanese, Malaysian, and Philippine exports. International Energy Agency. 2004. World Energy
20. China here is taken as the sum of Hong Kong, Outlook. Paris.
Macao, and China, based on the assumption that prior IMF (International Monetary Fund). 2005. World
to liberalization some of the exports from Hong Kong Economic Outlook. Washington, DC.
and Macao had actually originated in China, and OECD (Organisation for Economic Co-operation and
therefore, the changes in their market share reported in Development). 2004. Economic Outlook, no. 74.
official statistics are exaggerated. Paris.
21. During the 1980s (and before), OPEC acted as World Bank. 2005. Global Development Finance
a swing producer, stepping up production in response 2005. Washington, DC.
to shortfalls elsewhere. With its spare capacity now
measured at less than 2 million barrels per day, its
capacity to act as a swing producer is limited.

The Potential Gains from
International Migration

International migration can generate substan- migrants and for their origin countries. Here
tial welfare gains for migrants, their countries we can refer to a broader range of economic
of origin, and the countries to which they mi- issues than are captured by the model, al-
grate. The main focus of this report is on gains though without the ability to quantify that the
from the remittances that migrants send home model-based simulation provides.
(discussed in chapters 4–6); chapters 2 and 3 Starting from the base-case forecast of eco-
address the economic costs and benefits of nomic activity described in chapter 1, we in-
migration and the impact of migration on troduce an additional increase in migration
poverty. In this chapter, we use an economic from developing to high-income countries suf-
model to estimate the size of the welfare gains ficient to raise the labor force of high-income
resulting from migration from developing to countries by 3 percent over the period
high-income countries.1 It must be recognized 2001–25. The assumed increase, roughly one-
at the outset that the model fails to capture eighth of a percentage point a year, is close to
some known costs and benefits of migration; that observed over the 1970–2000 period. We
that the results are dependent on the specifica- imply no judgment concerning whether such
tion of the model and its key parameters; an increase is likely or politically feasible, but
and that the model cannot incorporate social rather view the rise in migration as an exoge-
or political considerations.2 The results of this nous shock. As discussed in chapter 3, pres-
simulation do not provide a precise forecast of sures to migrate are likely to rise over the next
the likely impact of migration; instead, they few decades, but the actual size of the migrant
provide a consistent framework that offers in- flows will depend heavily on political deci-
sights into (a) the economic gains that can be sions in destination countries. This exercise
expected from changes in policy or circum- presents us with the following key findings.
stances, and (b) the channels through which
migration affects welfare—and both are diffi- The expected decline in the labor force in
cult to measure in reality. The conclusions high-income countries will increase depen-
drawn from the model are supported by sev- dency ratios, which could add to the benefits
eral empirical studies, and they hold up well from migration. However, such increases in
under various alternative assumptions for migration are unlikely to be large enough to
model specification and parameters. have a significant impact on dependency ra-
In chapter 3, we complement this model- tios in high-income countries.
based approach to measuring the gains from
migration with a review of the economic liter- Under the assumptions adopted in this model-
ature, which covers the implications for ing exercise, the rise in migration—small

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

relative to the labor force of high-income coun- The costs of adjusting to increased migration
tries, but large relative to the existing stock of and the gains from migration depend, in part,
migrants—would generate large increases in on the investment climate. Adjustment costs
global welfare. Migrants, natives in destination as a result of migration will be lower if more
countries, and households in origin countries flexible labor markets and more efficient cap-
would experience gains in income, although mi- ital markets in high-income economies reduce
grants already living in high-income countries transitional unemployment and the cost of re-
would see a decline in wages relative to the base placing capital as economies adjust to the rise
case. Estimates of these gains and losses are par- in immigration. Similarly, developing coun-
ticularly sensitive to assumptions about the de- tries with strong investment climates will be
gree of differentiation among workers (between able to use increased remittances more effi-
natives and migrants and between old and new ciently, and enable workers who do not mi-
migrants), the impact of migrants on fiscal bal- grate to respond to improved labor market
ances, and the extent of remittances. conditions. The cost of adjustment may also
be lower if migration is spread over time
Empirical studies of the impact of migration rather than concentrated in spurts.
on natives’ wages have had mixed results. In
this simulation exercise, the rise in migration A principal conclusion from this exercise is
leads to a small decline in average wages in that migration can generate significant eco-
high-income countries relative to the baseline, nomic gains for migrants, origin countries,
which one would anticipate from a labor sup- and destination countries—but migration also
ply shock. But the decline has a barely per- can have important political and social conse-
ceptible impact on the long-term growth rate quences. For example, natives in destination
for wages. countries may become concerned about main-
Native households in high-income countries taining cultural identity in the middle of a
enjoy a rise in income, on average, as returns growing diversity, which also has implications
to capital increase, offsetting the mild decline relative to minority languages and other issues
in wages. The impact on developing countries surrounding the integration of migrants. To
is nearly the reverse, with wage income rising some extent, opposition to migration is driven
as labor-market conditions for workers im- by these concerns, and not by an economic
prove, while returns on capital decline with calculation of the gains and losses.
the smaller supply of workers. In developing We begin with a discussion of recent trends
countries the gain from increased remittances and discuss how migration to high-income
greatly exceeds that from changes in factor countries has grown over the past 30 years.
returns. We then turn to the prospects for migration,
including the intense pressures generated by
The economic benefits for high-income demographic changes. We describe the base-
economies could be even larger than those pre- case scenario for migration and the model-
dicted by the model, due to several factors: the based analysis of the welfare gains from in-
model excludes the increased productivity of creased migration. We conclude with issues
migrants (and the benefits to their offspring) that the model does not consider.
over time; investment levels could increase
substantially in response to higher returns to International migration trends
capital; labor-force participation could rise
among natives with the greater availability of Migration to high-income countries
migrant labor (for household help, for exam- has accelerated
ple); the labor market would become more The United Nations (UN) estimates that mi-
flexibile, and diversity would increase. grants account for some 3 percent of the


Table 2.1 Growth in international migration by destination, 1970–2000

Percent change per year in stock of migrants
1970–80 1982–90 1990–2000

World 2.0 4.4 1.3

High-income countries 2.4 2.9 3.1
Developing countries 1.8 5.5 –0.1
Excluding former USSR 1.9 2.1 0.0
Former USSR 0.5 25.0 –0.3

Source: United Nations.

world’s population, or about 175 million tion growth, the share of migrants in develop-
persons.3 The stock of immigrants to high- ing countries’ population (excluding the for-
income countries increased at about 3 percent mer Soviet Union) fell (figure 2.1).4
per year from 1980 to 2000, up from the Most high-income countries saw immigra-
2.4 percent pace in the 1970s (table 2.1). At tion rise by at least 2 percent per year from
that rate of growth, the share of migrants in 1980 to 2000.5 This increase reflected, in
high-income countries’ population almost part, increased demand for services accom-
doubled over the 30-year period, and popula- panying rising incomes, global competition
tion growth (excluding migration) fell from for highly educated workers as technological
0.7 percent per year in the 1970s to 0.5 per- advances boosted the premium for skills, the
cent in the 1990s. Immigration has had a par- growth of networks of immigrants in high-
ticular impact on population growth in several income countries that facilitated new immi-
high-income countries. For example, without gration, and increased refugee movements.
immigration Germany, Italy, and Sweden Almost 70 percent of the increase in immi-
would have experienced a decline in popula- gration is accounted for by the United States
tion in the past few decades (OECD 2005; and Germany, which together make up less
IOM 2005). By contrast, migration to devel- than 40 percent of the population of the
oping countries rose by only 1.3 percent per high-income countries. In the United States,
year from 1970 to 2000. With rapid popula- the Immigration Reform and Control Act
(IRCA) of 1986, which provided permanent
status to 2.7 million migrants, facilitated fur-
Figure 2.1 International migrants as a ther immigration through rules governing
share of destination countries’ population family reunification and may have encour-
aged further irregular immigration (Passel
9 2005) by encouraging expectations of future
8 amnesties.6 Germany saw a large inflow of
7 ethnic Germans following the breakup of the
6 Soviet Union (Dustmann and Glitz 2005), as
well as an increase in temporary migration
4 1970
3 2000
under bilateral agreements.
2 Though the stock of migrants has acceler-
1 ated sharply relative to the population in the
0 industrial countries, in some respects the com-
World Developed Developing Developing
countries countries countriesa position and patterns of international migra-
tion have exhibited continuity over the past
Source: United Nations.
Note: a. Excluding countries of the former USSR.
few decades. The share of female migrants
has remained almost unchanged (47 percent

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 2.2 Share of females in Figure 2.3 Immigration to selected

international migration countries, reasons for admittance, 2001
Percentage Percent
51 80
50 70 Kingdom Australia
49 2000 60
48 1970 50 States
47 40 France
46 30
45 20
44 10
43 0
World Developed Developing Workers Reuniting Refugee/
countries countries families asylum

Source: United Nations. Source: United Nations.

Note: Stock data, reported by countries of destination. Note: Stock data, reported by destination countries.

of global migrant populations in 1970, com- It should be emphasized that the migration
pared with 49 percent in 2000—figure 2.2), data on which these judgments are based
although women are the great majority of tend to be unreliable and incomplete. Many
migrants from some countries. More women countries and international agencies do not
today are migrating as independent wage distinguish between regular and irregular
earners, rather than to accompany their hus- migration or among types of temporary migra-
bands (IOM 2005). Migration continues to be tion. Some record migrants’ country of birth;
heavily determined by geographic proximity others their nationality (OECD 2005).
(from Mexico to the United States, from National estimates of the number of migrants
North Africa to Southern Europe, and from can be vastly different depending on whether
Eastern to Western Europe), as well as by “migrant” is defined as foreign born or of for-
colonial ties (from Latin America to Spain and eign nationality.
from a number of Sub-Saharan African coun-
tries to Belgium, France, Portugal, and the Migration is set to increase
United Kingdom—OECD 2005). The major It is likely that the number of people who wish
countries of destination continue to admit the to migrate from developing to high-income
largest share of permanent immigrants for countries will rise over the next two decades.
family reunification (or, in the case of the EU About 31 percent of developing countries’
countries, for humanitarian or refugee resettle- population is below the age of 14, compared
ment), although some countries are refocusing with 18 percent in high-income countries. We
their migration policy toward economic can thus anticipate a large influx in the age
(largely skilled) immigration (figure 2.3).7 But categories most suitable for emigration, as
international migration is also changing, par- lifetime earnings from migration tend to be
ticularly in the direction of flows. For exam- largest for those emigrating early in their
ple, more Asians are today seeking work in working life. The surge in immigration since
other Asian countries rather than in the Mid- the 1980s has established large diasporas in
dle East (Wickramasekera 2002; OECD 2005; high-income countries, which help to reduce
IOM 2005), while more Latin Americans are the costs and risks of migration (see chap-
turning to Europe for work opportunities, in ter 3). The demand for immigrant services in
addition to North America. high-income countries will also rise as the


aging of the population shrinks the workforce suming no change in labor-force participation
and increases demand for services that rates, the high-income countries may lose
immigrants can supply (such as nursing care). about 20 million workers by 2025, relative to
As income standards rise, the demand for peak employment.8
other services that employ migrants (such as The expected decline in the labor force is
household and restaurant help) should grow accompanied by a rise in the overall depen-
rapidly. The intensifying competition for dency ratio, defined as the ratio of nonworkers
skilled workers may also draw migrants, espe- to workers. For the high-income countries as a
cially from countries with strong systems of group, this ratio is forecast to remain at just
higher education. under one through 2009. However, by 2025,
100 workers will be supporting 111 depen-
Policies in destination countries dents, largely reflecting the increased number
can affect migration of the elderly (also, in most countries the num-
Forecasts of migration flows remain problem- ber of children under 15 will fall). The largest
atic. But with the underlying demand for and rise in the dependency ratio will be in Europe.
supply of migrants likely to increase in com- If we focus more narrowly on the number of
ing decades, the number of migrants will de- elderly per worker, every 100 European work-
pend on policy decisions governing admit- ers now support 36 elderly people; by 2025
tance and the effectiveness of efforts to police they will have to support 52. In Japan 100
borders and enforce workplace rules. Opposi- workers will support 60 elderly in 2025.
tion to immigration may grow as the number
of migrants increases, as it did in major In the developing countries the labor
countries of destination before World War I. force will expand
But it is likely that the main policy issue will Developing countries show considerable di-
be how best to manage and live with in- versity in demographic trends, but overall the
creased migration. In the simulations that fol- bulge of youths born over the last two decades
low, we explore the impact of an increase in is now entering the labor force, the number of
migration to 2025 in line with recent histori- elderly is as yet still rising slowly, and the
cal experience. number of births is falling rapidly. Thus
developing countries are forecast to add
nearly one billion workers to the world’s labor
The demographic challenge force by 2025, again assuming no change in
the labor-force participation rate, and depen-
The labor force in the high-income dency ratios are expected to fall.
countries is set to decline The expected expansion of the labor force
A key driver in the demand for international in developing countries, coupled with large
migrants over the next 20 years will be slow- wage premiums in high-income countries,
ing growth, and then decline, of the labor means that migration could help reduce de-
force in high-income countries. The age group pendency ratios in high-income countries.
that supplies the bulk of the labor force However, increases in immigration sufficient
(15–65 years old) is expected to peak near to have a noticeable impact on dependency ra-
500 million in 2010, and then fall to around tios would have to be very large. The scenario
475 million by 2025 (figure 2.4). In Japan this discussed below envisions an increase in the
age group has already begun to shrink, while labor force in high-income countries of 3 per-
in Europe the peak will be reached in cent through migration, or a hike of nearly
2007–08. In the other high-income countries, 50 percent in working migrants in high-income
the peak will occur later—around 2020 for countries. Even if migrants come with no
the United States and 2015 for the rest. As- elderly, the dependency ratio in the host coun-

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 2.4 Labor force and dependency rates

Europe Japan

Labor force, millions Dependency rate Labor force, millions Dependency rate
200 1.3 70 1.3

195 65
1.2 1.2

190 60
1.1 1.1
185 55
1.0 1.0
180 50

0.9 0.9
175 45

170 0.8 40 0.8

0 1 04 07 10 13 1 6 19 22 25 0 1 04 07 10 13 16 19 22 25
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20

United States Other high-income countries

Labor force, millions Dependency rate Labor force, millions Dependency rate
170 1.3 80 1.3

165 75
1.2 1.2

160 70
1.1 1.1
155 65
1.0 1.0
150 60

0.9 0.9
145 55

140 0.8 50 0.8

1 4 7 0 3 6 9 2 5 1 4 7 0 3 6 9 2 5
2 00 200 200 201 201 201 201 202 202 2 00 200 200 201 201 201 201 202 202

Total, High-income countries Total, Developing countries

Labor force, millions Dependency rate Labor force, millions Dependency rate
500 1.3 3,600 1.3

495 3,400
1.2 1.2

490 3,200
1.1 1.1
485 3,000
1.0 1.0
480 2,800

0.9 0.9
475 2,600

470 0.8 2,400 0.8

0 1 04 7 10 13 1 6 19 22 5 0 1 04 7 10 13 16 19 22 5
20 20 200 20 20 20 20 20 202 20 20 200 20 20 20 20 20 202

Labor force Dependency rate

Source: World Bank databanks (DDP).

Note: The dependency rate is measured as the ratio of the nonworking population to working population.


tries would fall by only about 3 percent under numbers hold up well as an approximation of
such a scenario. In the case of Japan, it would the gains to global output, regardless of vari-
lower the number of elderly dependents in ous assumptions made about taxes, non-wage
2025 from 60 per 100 workers to 59 per 100 income distribution, key model parameters,
workers—barely a dent. Nevertheless, as and other factors.
discussed in more detail below, selective Our modeling exercise uses a global gen-
migration—for example, of experienced and eral equilibrium model to measure the impact
skilled workers—can help mitigate the transi- of migration (box 2.1).10 One of the purposes
tional costs of financing pension benefits for of the global model is to verify the basic intu-
rapidly aging populations in high-income ition described above—that migration pro-
countries. duces a sizeable global gain. But it also is a
powerful tool to evaluate distributional
impacts—between skilled and unskilled work-
Migration and its development ers, between native- and foreign-born workers,
impact between capital and labor, and across

T o illustrate the potential gains from

increased migration, we compare the base-
case forecast for output and consumption
regions—and to show how these distribu-
tional impacts vary with policy choices and
parameters (for example, the role of fiscal
in chapter 1 with an alternative scenario, in policies or the propensity to remit).
which the stock of migrant workers is allowed
to increase in the high-income countries so The assumption is that migrants as a
as to raise the overall stock of workers by 3 per- share of population remain constant
cent (a movement of 14.2 million workers in the baseline scenario
from developing countries to high-income We begin with a base case for global eco-
countries by the year 2025). A first nomic activity (outlined in chapter 1), demo-
approximation of the global gains from such a graphic trends (described at the outset of this
scenario is simply to calculate the income gains chapter), and for migration. For the base case,
accruing to the new migrant workers—this the proportion of migrants in each region re-
will reflect the gains to the global economy, mains the same over time—somewhat con-
because it approximates the increase in global trary to the trends of the last two decades.
productivity derived from equipping the mi- This does not imply that gross migration is
grants with more and improved capital and stagnant, or even declining. The stock of mi-
technology. This back-of-the-envelope calcula- grants in any year will equal the previous
tion yields an increase in gross wage income of stock of migrants, plus new migrants, less the
$772 billion in 2025.9 As we will see later, attrition through death and return migration.
when corrected for differences in prices that We chose a relatively neutral assumption be-
migrants face in high-income versus develop- cause of the difficulty in forecasting these
ing countries, and taking into account other complex processes. For some countries—for
impacts of migration (on prices, for example) example Japan and those in Europe—the as-
as calculated by the model, global gains fall to sumption results in an absolute decline in the
$356 billion—an 0.6 percent increase in global stock of migrant workers. This decline paral-
income. The scenario is particularly beneficial lels the overall decline in the European and
to developing countries relative to high-income Japanese labor forces.11 For the high-income
countries. The aggregate percentage gain to countries as a group, the stock of migrant
developing countries (including the new mi- workers would increase by some 760,000
grants) is 1.8 percent, whereas the gains to between 2001 and 2025, just a small increment
natives in high-income countries amount to from the estimated 27.8 million in 2001. The
0.4 percent relative to baseline income. These main issue, however, is not the base case,

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Box 2.1 The model used in this study

including estimates of population and the stock of
T he underlying analytical framework used in this
chapter is the World Bank’s standard global
general equilibrium model—LINKAGE—which has
workers, both skilled and unskilled (Walmsley,
Ahmed, and Parsons 2005).a World Bank data (de-
been used in previous reports for trade policy analy- scribed in more detail in chapter 4 of this report)
sis. It has been modified to differentiate between mi- was used to provide the total level of remittances,
grant and native workers and to incorporate remit- and the bilateral stock of migrants was used to esti-
tances. The model is based on release 6.0 of the mate the bilateral remittance flows subject to the
GTAP database (base year 2001), developed by the overall total flows.
Global Trade Analysis Project (, a The standard horizon for the LINKAGE model has
global network of researchers and policymakers en- been 2015. For the work described here, the model
gaged in the quantitative analysis of international horizon has been extended to 2025, in part because
policy issues. It is supplemented for use in our
demographic dynamics play a more important role
model with a new database developed jointly by
over the longer-term horizon, and in part to allow
GTAP and the University of Sussex (Parsons and
for more time to phase in the increase in migration.
others 2005). That database contains a comprehen-
sive estimate of bilateral stocks of migrants for 226
countries and territories.
aThe 87 regions of GTAP have been aggregated into 21 re-
While the new migration database is undergoing
constant improvements as new data become avail- gions for the purposes of this study. Six of these are high-in-
able and obvious errors are corrected, its developers come regions using World Bank definitions—the European
Union and the European Free Trade Area, Canada, the United
have done a remarkable amount of detective work,
States, Japan, Australia/New Zealand, and the newly-industrial-
largely in national data sources. The GTAP center izing economies. The fifteen developing countries/regions in-
has used this underlying migration database to build clude China, the Philippines, India, Russia, Turkey, South
a bilateral migration database for the 87-region level Africa, and Mexico as individual countries, plus 6 regions that
of aggregation of the main GTAP database— represent the remaining countries in each geographical area.

but rather the impact of deviations from of high-income countries by 3 percent, phased
it—although significantly different base as- in from 2010 through 2020.12 As migrants
sumptions could affect the deviations aswell. make up about 6 percent of high-income
According to the base-case scenario, mi- countries’ labor force, a 3 percent rise in the
grant workers would make up about 6 percent labor force (through migration) implies a
of the labor force of high-income countries in 50 percent increase in the number of migrant
2025, though with sharp differences across workers. This may seem like a large change,
regions and skills (table 2.2). The vast major- but the resulting stock of migrants in Europe,
ity of migrant workers are unskilled—some Japan, and the United States would remain a
25.3 million migrant workers out of a pro- far smaller share of population than current
jected total of 28.5 million, or 7.8 percent of levels in some high-migration countries. (In
high-income countries’ labor force. Skilled mi- Australia, for example, about a quarter of the
grants, on the other hand, represent just 2.2 per- population are migrants, in Canada 19 per-
cent of the total skilled workforce on average. cent, in Kuwait 50 percent). The percentage
increase in migrants is large in Japan (as the
There are welfare implications if baseline share of migrants is relatively low),
migration rises significantly and lower in the United States. The increase
The alternative scenario involves a rise in corresponds to an annual growth rate of
migration sufficient to increase the labor force about 1.9 percent, somewhat slower than the


Table 2.2 Labor force structure in the base case and after increases in migrants
In millions except where noted
Baseline Migration shock

2001 2025 Change in millions 2001–25 Change in percent 2001–25

High-income countries
Total labor force 480.8 474.0 14.2 3.0
Developing-country migrant workers 27.8 28.5 14.2 49.9
Unskilled 24.6 25.3 9.8 38.6
Skilled 3.1 3.2 4.5 137.9
Developing-country migrant workers
as share of total labor force, percenta 5.8 6.0 8.8
Unskilled, percent 7.4 7.8 10.5
Skilled, percent 2.1 2.2 5.0
Developing countries
Total labor force 2,596.2 3,561.0 –14.2 –0.4
Unskilled 2,395.9 3,294.3 –9.8 –0.3
Skilled 200.4 266.7 –4.5 –1.7

Source: Initial 2001 data from migration database under development by GTAP/University of Sussex (Parsons and others 2005
and Walmsley, Ahmed, and Parsons 2005). Scenarios based on World Bank assumptions.
Note: a. The percentage of migrant workers as a share of the total labor force is assumed to be the same for each individual
region of the model throughout 2001–25, but the share averaged across all developed regions will change through aggregation

average increase over the period 1980–2000. reflect the likely migration pressures implied
Moreover, the growth rate is unbalanced, with by large differences in demographic trends in
an annual increase of only 1.5 percent in un- sending regions (for example, Sub-Saharan
skilled workers, but 3.8 percent in skilled Africa versus Latin America).
workers. A number of additional assumptions Third, foreign workers are assumed to
are critical to the results. bring family members in proportion to the
First, the high-income countries’ labor dependency ratio in their home country. As a
force of both skilled and unskilled workers result, the total number of migrants in high-
increases by 3 percent.13 As the share of skilled income countries increases from 65 million
workers among migrants is much smaller than (6.5 percent of high-income countries’ popula-
the share of skilled workers among high- tion) in the baseline for 2025, to 93 million
income country natives, the shock results in a (9 percent of population) after the shock. This
much larger percentage increase for skilled assumption can change the average depen-
migrants. The number of unskilled migrant dency ratio of the host country. It can
workers increases by 39 percent, while the also have other implications not modeled
number of skilled migrant workers rises by explicitly—including fiscal impacts, because
138 percent.14 the families of new migrants may require ad-
Second, the share of migrants by region of ditional public services (such as schooling),
origin remains constant; in other words, the not fully compensated by the taxes paid by the
new migrants reflect the same allocation by new migrants.
region of origin as existing ones. Thus if Fourth, remittances are assumed to be a
Mexicans constitute 30 percent of foreign mi- fixed proportion of migrants’ labor income,
grants in the United States in the base case, equal to the level in the base year. The average
they maintain the same share after the increase for developing countries is 17 percent, although
in migration. This assumption is made to sim- the level varies with the migrant’s origin and
plify the analysis, although it does fail to destination countries. New migrants are

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

assumed to send remittances to their home nearly 0.9 percent from baseline levels.18 A
country at the same rate (relative to income) significant portion of the increase is due to the
as existing migrants.15 remittances from the new migrants, with some
improvement in labor-market conditions for
remaining workers. Those who are likely to
Returns to households
lose—in the absence of any compensatory
The gains from increased migration mechanism—are the existing migrants in
are large high-income countries, who are relatively
With the labor force moving, it is best to close substitutes for the new migrants. Their
assess the effects on real income in terms of private consumption would decline by over
households as opposed to the national level 9 percent and overall consumption (including
(as is typically done in analyses of trade re- public services) by 6 percent compared to
form). Households are broken down into four baseline levels.
groups. First are the native households in
high-income countries.16 Second are previous New migrants and their countries
migrants from developing countries now liv- of origin reap benefits
ing in high-income countries, that is, those (through remittances)
who were in place in the baseline scenario. The main gains come from the higher incomes
Third are native households in developing the new migrant workers can earn in the des-
countries—households that do not migrate.17 tination country relative to what they would
And finally, we have the households of the have earned in their country of origin. New
new migrants. Each household’s welfare is migrants earn $481 billion in real (after-tax)
broken down between the change in private income in 2025 over the base case. However,
consumption and the change in the consump- the dollar increase in income overestimates the
tion of public services. welfare gains for migrants. Essentially, an
Natives in high-income countries gain additional $1 spent in the high-income coun-
$139 billion in real income, or 0.4 percent of tries does not provide the same amount of
the baseline, as a result of the rise in migration welfare as an additional $1 spent in the home
(table 2.3). Nonmigrating households in de- country, because prices are higher in high-
veloping countries see a rise in real income of income countries. Whereas the prices of

Table 2.3 Change in real income across households in 2025 relative to baseline
Real income adjusted for
Real income cost of living

Private Public Total Private Public Total

Change, $ billions Change, $ billions

Natives in high-income countries 139 –1 139 139 –1 139
Old migrants in high-income countries –88 0 –88 –88 0 –88
Natives in developing countries 131 12 143 131 12 143
New migrants 372 109 481 126 36 162
World total 554 120 674 308 48 356
Change, % Change, %
Natives in high-income countries 0.44 –0.01 0.36 0.44 –0.01 0.36
Old migrants in high-income countries –9.41 –0.02 –6.02 –9.41 –0.02 –6.02
Natives in developing countries 0.94 0.44 0.86 0.94 0.44 0.86
New migrants 584 607 589 198 203 199
World total 1.20 1.15 1.19 0.67 0.45 0.63

Source: World Bank model simulations.


traded goods (for example, cars and electron- using purchasing power parity (PPP) exchange
ics) are the same worldwide, at least in princi- rates from the World Bank’s database.20 Thus
ple, the prices of nontraded goods and services instead of an increase of $481 billion, the rise
(for example, housing and haircuts) are much in welfare for new migrants is $162 billion.21
higher in high-income countries. Table 2.3 shows the change in these com-
A simple example may clarify the idea. ponents for the four household groups and the
Take a household of two persons living in world. Measured in national accounting
their home country. One works and earns terms, that is, with no adjustments for the dif-
$200. The other does not work. Each spends ference in the cost of living for the new mi-
$100, half on tradable goods (each priced at grants, global real income rises under the
$1) and half on nontradable goods (likewise model by 1.2 percent relative to the baseline,
priced at $1). Now the worker moves to a or 0.6 percent with the cost-of-living adjust-
high-income country and earns $700. Assume ment. Global private consumption increases in
that spending patterns do not change. The real terms by $308 billion in 2025 (with the
worker remits $200 back to the home country, cost-of-living adjustment), with real govern-
so the income (and welfare, in money terms) ment expenditures increasing by an additional
of the other doubles. The new migrant buys $48 billion. The total real gain—with equal
the same goods—50 units of tradable goods weight for high-income—and developing-
and 50 units of nontradable,19 but the price of country gains—is $356 billion, with just
the latter is now $9 and not $1. The migrant under half accruing to the new migrants,
thus spends $500, but welfare is unchanged, though natives in both high-income and devel-
because the basket of purchased goods is oping countries also are better off. In percent-
identical. age terms—where relative weights between
Welfare evaluations are of course more high-income and developing countries are
complex than this simple example illustrates. irrelevant—the scenario clearly indicates that
For one thing, new migrants will have to adjust the relative gains are much higher for
their spending patterns to deal with their new developing-country households than high-
environment. Heating oil and warm clothes income country households, rivaling gains
are necessities that will not boost a migrant’s from global reform of merchandise trade.
welfare above what it was in the home coun- Obviously, global income and global gains
try. For another, the decision to migrate is not would also be larger if expressed in PPP terms.
taken for simply static reasons; there are sig- As the percentage increase in welfare for mi-
nificant dynamic reasons for migrating—for grants living (originally) in developing coun-
example, better opportunities for one’s chil- tries is larger than the percentage increase for
dren that are not captured in this simple frame- those living in high-income countries, a switch
work. Nonetheless, the difference in purchas- to PPP measures would also increase the
ing power illustrated in the example is a strong global gains as a percentage of global income.
motivation for migrating, even on a temporary If in the migration scenario presented here the
basis. The more wage income earned in high- gains are PPP-adjusted, the global gains would
income countries that can be spent in lower- amount to 0.9 percent of global income in the
income countries, the greater will be the wel- baseline, instead of 0.6 percent using the EV
fare benefits. Box 2.2 provides additional aggregation. This scenario illustrates that mi-
detail on the computation and interpretation grants living (originally) in developing coun-
of global welfare gains from migration. tries gain the most from migration in percent-
To account for the change in prices faced age terms.
by the new migrants, their “new” consump- The impact of higher migration on prices is
tion in the destination country is adjusted to mild in aggregate in high-income countries,
account for differences in the cost of living, with a small decline in the average price of

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Box 2.2 Calculating and interpreting global welfare

gains from migration
T wo sets of issues arise with respect to the so-
called global gains from a policy shock. First,
how should the gains of specific groups be evaluated
national accounting standards. However, the stan-
dard real income measure is still a good approxima-
tion of the welfare gains for the other households in
and how do the gains compare with traditional mea- the model.
sures, such as GDP or national accounting stan- To the extent that new migrants remit part of
dards? Second, how should the gains be aggregated their income to their country of origin and that
over groups and countries, and how should the income is spent in that country of origin, the increase
aggregated gains be interpreted? in the cost of living that new migrants face is not
Evaluation of the welfare gains of specific relevant. Therefore, the EV measure of remittances is
groups. In standard applications of general equilib- larger than the same nominal income spent by the
rium (GE) models, the welfare impacts of specific new migrant in the host country. This difference
groups are evaluated using a concept from welfare illustrates the incentive for new migrants to remit
theory called equivalent variation (EV). The con- income home.
cept is relatively straightforward. Welfare changes Aggregation. The second issue relates to the inter-
as a result of changes in nominal income and pretation of the “global” gains. Typically, to derive
changes in prices. EV calculations summarize this aggregate or global gains, EV (expressed in a com-
welfare change in terms of an equivalent change mon currency, typically the U.S. dollar) is summed
in income alone, showing by how much income across all households. For individual persons or ho-
at original prices would have to change to mogeneous groups this EV aggregation, expressed as
achieve the same change in welfare as observed in a percentage of original income, is a good approxi-
a simulation.a mation of the change in welfare (or more precisely, it
For most households, the standard notion of the is a good indication of the change in welfare).e
change in real income, that is, the difference in nomi- However, no clear link exists between global wel-
nal income adjusted by the change in the CPI, is a fare and the aggregation of EV across heterogeneous
good approximation of EV.b groups, because we do not know how to weigh indi-
This is not the case for new migrants, however. vidual welfare across heterogeneous groups (a partic-
There is no standard price index that can be used as ularly difficult issue in aggregating across countries
a deflator for the change in the nominal gains for the at very different stages of development, as is done
new migrants, since the prices they face in their new here). For example, while most groups gain from mi-
host country have no linkages to the prices they paid gration in the scenario discussed in the text, some
in their home countries. GE and macro models typi- lose. The fact that the change in global welfare (ex-
cally calibrate base-year prices in each region to one pressed as the aggregation of EV across groups) is
(or unit value) by choosing corresponding volume positive does not mean that the welfare gains of the
units.c winners are considered more important than the wel-
This approach does not allow one to take into ac- fare losses of the losers. Thus, global gains as ex-
count the price increases that new migrants face as pressed in aggregate EV should not be interpreted as
a result of their migration. In the simulations, the a value judgment on how to weigh individual or
macro PPP exchange rate (as an approximation of local welfare gains.
the rise in prices faced by migrants from developing The aggregation of EV across groups does,
to high-income countries) has been used to adjust the however, have a useful interpretation, which is
gains to the migrants—although this is just an ap- linked to the notion of compensation and Pareto
proximation of the true welfare gains.d optimality. As long as the global gains are positive—
Because of the cost-of-living adjustment to the using the standard practice of adding up EVs across
welfare gain of new migrants, the real gain reported households—then it is possible through redistribu-
is no longer equal to real income gains of tion to compensate households that lose (so that no
countries—and real output gains—measured using one is worse off relative to the baseline scenario),


Box 2.2 (continued)

when some households are better off. In that sense developing countries in global aggregates would in-
the global gain can be compared with an equal rise crease in the measurement of both global income
in global output plus redistribution. levels and global welfare gains. However, the per-
In this report we maintain this standard practice centage increase in income for developing countries
of reporting EV aggregates, making the gains compa- would not be affected.
rable to global gains in many other studies.
An alternative approach to calculating global aOne of the advantages of the EV measure is that it trans-

gains would be to add up changes in income mea- forms the ordinal concept of welfare into a cardinal concept of
sured in PPP terms. The rationale for that alterna- income. While it is impossible to measure how much one welfare
level differs from another (one can only conclude that one level is
tive is that because prices of nontraded goods are
preferred to another), the corresponding increase in income can
lower in developing countries, the addition of a dol- be measured, and the size of the increase has a clear meaning.
lar to a developing country would enable the pur- bFor example, in trade-reform scenarios, the change in the

chase of a larger amount of goods and services than price index is a relatively good approximation of the welfare
in an high-income country. In that case, both base impact, since the new price is approximately the old price less
income and gains for new migrants and for those the tariff.
cThere are exceptions. For example, in the case of climate-
who remain in developing countries would be
change models, it is necessary to know the relative prices of the
roughly three times as large as reported here. This different fuels to accurately determine the carbon tax.
is true for all gains, whether they come from migra- dSee Timmer and van der Mensbrugghe (2005) for more
tion itself, from remittances, or from changes in details.
eThe size of the change in individual welfare is undeter-
wages and prices in developing countries. As a re-
sult, the share of those who live (originally) in mined, since welfare is an ordinal concept.

absorption (private consumption, private in- For the new migrants, the real income
vestment, and government spending) of 0.1 gains—cost-of-living adjusted—increase by
percent. However, prices of some key nontrad- nearly 200 percent. There are large differences
ables decline by larger amounts—0.8 percent across regions, with the highest gains (in per-
on average for public services (including centage terms) accruing to migrants from Sub-
health-related services) and 0.2 percent for con- Saharan Africa (619 percent) and the lowest
struction and recreational services. These price to migrants from the Middle East and North
declines will be even sharper for specific sub- Africa and Europe and Central Asia. The main
sectors where migrant workers are concen- reason for the disparity is the relative differen-
trated (for example, household help), for which tial between wages in origin and destination
we currently have no comprehensive data. countries. Variations in wages paid to mi-
The allocation of the gains across develop- grants from different regions in destination
ing countries depends on various factors, in- countries are minor, whereas there are very
cluding the skill loss and the resulting impact wide variations in wages in countries of ori-
on production, the locations to which mi- gin. For example, the average wage for a mi-
grants move and the relative wage differential, grant in Europe in the base year is about
and the propensity to remit. By developing re- $16,500—with only minor variation across
gion, the gains to households under the model migrants. However, the average wage in Sub-
vary from 0.6 percent for Europe and Central Saharan Africa is only $470, whereas in the
Asia to 1.1 percent for South Asia and Latin Middle East and North Africa it is $2,700.
America and the Caribbean (table 2.4). Thus, the migrant from Sub-Saharan Africa

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 2.4 Real income impacts across developing regions

Change in 2025 relative to the baseline, adjusted for differences in cost of living
Natives in region New migrants from region

$ billions Percent $ billions Percent

Total developing 143 0.9 162 199

East Asia and Pacific 37 0.7 32 215
South Asia 21 1.1 2 175
Europe and Central Asia 14 0.6 25 138
Middle East and North Africa 18 0.9 11 134
Sub-Saharan Africa 7 0.9 7 619
Latin America and the Caribbean 47 1.1 85 224

Source: World Bank model simulations.

will gain much more in both absolute and per- several channels, some of which increase, and
centage terms than one from the Middle East others that decrease, trade flows:
and North Africa.
• First, the rise in incomes due to migra-
The impact of migration on trade tion produces a small rise in global trade
would be mild flows, with regional differentiation (be-
Whether migration and trade are substitutes cause income gains differ considerably
for each other is an old debate. For exam- among regions). In addition to higher in-
ple, in the discussions leading up to the sign- comes, the rise in migration changes the
ing of NAFTA—the free trade agreement size of regional economies, with implica-
among the United States, Canada, and tions for their demand for imports and
Mexico—one of the key arguments was that ability to export.
trade would replace migration and reduce • Second, the nature of the shock assumed
the pressure for Mexicans to migrate to the in our model differs from the standard
North. Likewise, allowing for increased debate over trade and migration. The
migration—for example of unskilled share of skilled workers in total migrants
workers—could reduce trade, because it is larger in the shock than in actual mi-
would enable the high-income countries to gration over the recent past. A large
continue producing low-skill-intensive prod- proportion of skilled workers will find
ucts at competitive cost. employment in nontraded sectors—for
Evidence of the link between trade and example, as doctors and nurses—rather
changing the comparative advantage emerges than in producing traded goods. This
in the migration scenario described here. For will have general equilibrium effects to
example, the largest gains in export revenue the extent that the price of nontraded
for high-income countries come in agriculture, goods will decline by more than the price
clothing, other manufacturing, recreational of traded goods. Thus there will be a rel-
services, and public services—all labor- ative shift to nontraded goods and a
intensive sectors, the first four being relatively potential reduction in demand for im-
intensive in unskilled workers and the last in ports of traded goods. Overall, the larger
skilled workers. share of skilled versus unskilled workers
Change in comparative advantage has only does tend to reduce trade flows.
a mild impact on trade flows in this scenario, • Third, the increase in remittances pro-
however, as migration affects trade through vides an opportunity for developing


countries to import more and export less, Migrants’ impact on government fiscal
as their current-account balance will in- accounts is broadly neutral
crease by the size of the remittances ($98 The assumption concerning the level of
billion in net terms). The model results consumption of public goods and services by
show that total imports into developing new migrants has important implications for
countries would increase by $58 billion individual gains, and global gains, under the
in 2025 (1.1 percent relative to the base- modeled scenario. We assume that the new
line), as aggregate exports decline by migrants’ level of consumption of public
$40 billion (0.7 percent).22 The change goods and services equals the amount they pay
in remittances leads to an appreciation of in taxes, that is, their impact on the public
the real exchange rate and therefore a budget is revenue-neutral. This is broadly con-
loss in relative export competitiveness.23 sistent with the available evidence (box 2.3).
For instance, the output price index in To provide some sense of how different ap-
developing countries rises by 0.6 percent proaches would affect the scenario results, we
on average, whereas it declines by 0.1 per- present two alternative assumptions regarding
cent for high-income countries. the distribution of public goods and services
to the new migrants (table 2.5). The default
In summary, the scenario provides evidence assumption had a largely neutral impact for
that changes in comparative advantage due to existing residents in the host country. Under
migration do influence trade flows. However, another assumption—new migrants pay taxes
overall migration and trade are not substitutes but receive no benefits from public goods and
for each other, because migration has many services—existing residents, native and mi-
other economic effects that have more power grant, enjoy a rise in real incomes of $126 bil-
to stimulate or reduce trade. One implication lion ($117 billion for natives and $9 billion
of this finding is that migration policies should for existing migrant households). Note that
not be pursued because of their specific impact the global welfare gains increase as well, since
on trade flows. Likewise, in trade policies the the income accruing to natives (and existing
impact on migration should not be a main migrants) is not adjusted for the differences in
focus.24 Trade and migration policies should the cost of living between developing and
be evaluated on their own merits. high-income countries.25 A second extreme

Table 2.5 Impact of different assumptions on the consumption of public goods

and services by selected groups in 2025
Change in cost-of-living-adjusted real income in 2025; billions of dollars
“New” migrants “New” migrants
receive benefits receive no public “New” migrants
equal to their benefits but pay receive per capita
taxes taxes average benefit

Private Public Total Public Total Public Total

Natives in high-income countries 139 1 139 117 256 85 54

Old migrants in high-income countries 88 0 88 9 79 6 94
Natives in developing countries 131 12 143 12 143 12 143
New migrants 126 36 162 18 108 75 201
World total 308 48 356 120 428 4 304

Source: World Bank model simulations.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Box 2.3 The impact of immigrants on fiscal balances

60s generally imposed a long-term burden. Studies
I mmigrants’ net contribution to fiscal revenues is
usually considered to be small. The net fiscal im-
pact of immigration on the United States has been
that follow immigrants over time generally conclude
that in net-present-value terms, immigrants and their
minimal (Coppel, Dumont, and Visco 2001; descendants tend to contribute more in terms of tax
Auerbach and Oreopoulos 1999). The U.S. revenues than they absorb in expenditures, but the or-
Binational Study on Migration (1997) found that ir- ders of magnitude are typically small (OECD 1997).
regular migrants did impose a significant fiscal bur- Intergenerational models are sensitive to the discount
den on state and local government. However, school rate used and assumptions concerning the allocation
expenses accounted for the bulk of these costs, and of the fiscal burden over future generations.
(as the authors note) education is an investment Third, the computation will depend on the level
that may readily be recovered in greater future pro- of skills, experience, education, and fertility of immi-
ductivity. Moreover, Lee and Miller (2000) found
grants. Rowthorn (2004) calculates that skilled mi-
that the overall fiscal consequences of altering the
grants to the United States typically make a large
volume of immigration to the United States would
positive contribution to the fiscal balance, whereas
be quite small. Gott and Johnston (2002) and
Sriskandarajah, Cooley, and Reed (2005) estimated unskilled immigrants cost more on average than the
that immigrants made a positive contribution to taxes they pay. Storesletten (2000) calculates that the
public finances in the United Kingdom. Gustafsson net-present-value contribution of the average high-
and Osterberg (2001) found that new immigrants to skilled immigrant to the U.S. budget is $96,000; the
Sweden generated a net fiscal cost, but this turned medium-skilled immigrant’s contribution is $2,000;
into a positive contribution after a few years. Nana and low-skilled immigrant’s contribution is
and Williams (1999) found that immigrants to New $36,000.
Zealand had a positive fiscal impact. Bonin, The results may change over time, as migrant
Raffelhuschen, and Walliser (2000) found that the characteristics and government policies change. The
net fiscal contribution of immigrants to Germany probability that an immigrant to the United States
could be significant if the government selects for will receive public benefits has risen since the 1970s,
skills. probably due to an increasing share of immigrants
Calculations of the net fiscal cost of immigration from poorer countries (Gustmann and Steinmeier
are fraught with difficulties, for several reasons. 1998).
First, the computation at any point in time An issue of particular concern has been the im-
depends heavily on the methodology used, what pact of migration on government-financed pensions.
expenditures and revenues are included, which public Likely increases in immigration can make only a
services should be regarded as pure public goods (and small net contribution to strengthening the financing
the extent of economies of scale in expenditures), and of pensions in the United States (Fehr, Jokisch, and
whether households or individuals are considered. Kotlikoff 2004), although selecting immigrants for
Second, static calculations of the current net fiscal working age and high skill levels could improve the
impact fail to take into account the age structure of picture (Storesletten 2000). By contrast, increases in
the immigrant population. Smith and Edmonston immigration could make a significant contribution
(1997) found that immigrants arriving between the to financing pensions in Germany (Bonin, Raffel-
ages of 10 and 25 years produced fiscal benefits under huschen, and Walliser 2000) and Spain (Collado,
most scenarios, while immigrants arriving in their late Iturbe-Ormaetxe, and Valera 2004).

assumption is that new migrants receive the taxes.26 In this case natives in high-income
same amount in public benefits as the average countries would lose $85 billion in aggregate
household in the destination country. This public goods and services, although this
would imply a net positive transfer to the new amount would not translate one-for-one into a
migrant households, since they would receive benefit for new migrants due to the cost-of-
more in public benefits than they paid in living adjustment. These simulations underline


the effect of public policy on the distribution those from an increase in migration, and those
of gains from migration. from global trade reform—are scaled to the
same reference year, 2001, the gains from
trade reforms are $155 billion versus $175 bil-
Additional gains from migration lion from the migration scenario.29 This
can be substantial leaves little doubt that easing restrictions on
The gains for migrants from this scenario the movement of labor could provide a sig-
essentially provide the same message as ear- nificant boost to the global economy. More-
lier estimates. In their seminal paper, Walmsley over, in comparison with the most recent
and Winters (2003) estimate that a relax- work on global merchandise trade reform,
ation on the movement of temporary work- the gains from an increase in migration are
ers on the same order as that modeled more balanced toward income increases for
here—that is, 3 percent of the labor force of developing countries relative to developed
the high-income countries—would yield countries. In a study by Anderson, Martin,
global income gains of $150 billion (using a and van der Mensbrugghe (2005), the gains
1997-based comparative static model). The to high-income and developing countries are
result from our scenario that is roughly 0.6 and 0.8 percent, respectively, relative to
comparable to their figures (that is, global baseline income. In the scenario modeled
gains before adjustment for cost of living here, the income increases are 0.4 percent
and measured relative to 2001, rather than for native households in high-income coun-
2025) are more than double their results.27 tries and 1.8 percent for developing coun-
However, our figures are comparable with tries (including the new migrants).
the more recent work done by Walmsley
and her colleagues.28 One of the key reasons
for the increase in the global welfare impact Returns to factors of production
is a reevaluation of the assumed wage dif-
ferential between the home and host coun-
try. In their initial work, Walmsley and
F our critical factors determine the distribu-
tion of gains from migration among skilled
workers, unskilled workers, and owners of
Winters had assumed that new migrants capital: (a) the size of the increase in migra-
made up 50 percent of the difference be- tion; (b) the distribution of nonwage income
tween the home and host country’s wages. (profits); (c) the degree of substitution between
Their new assumption (used in our model as workers by region of origin; and (d) the degree
well) is 75 percent, based in part on the fact of substitution or complementarity between
that the migrants are permanent rather than workers and capital. We have already posited
temporary. Hamilton and Whalley (1984) that the increase in migration is large, with an
and Moses and Letnes (2004) have shown average increase in the migrant labor force of
that removing all restrictions on labor around 50 percent over a 20-year period, and
movement, admittedly not a realistic sce- comparable (if somewhat less) to the rise in the
nario, would yield a huge increase in world share of migrants in high-income country pop-
output. Overall, these papers suggest that ulation over 1970–2000. In the absence of any
labor-market restrictions are imposing a specific data on the source of migrant income,
much larger burden on the global economy we assume that migrants—both existing and
than are trade restrictions. The World Bank’s new—receive no nonwage income. In essence,
trade model suggests that removing all re- their real income will be driven by changes in
maining merchandise trade barriers would wages. The effects of this simple assumption
yield $287 billion in global real income gains on the distribution of gains are significant, and
in 2015. For the purpose of comparison, when the implications of relaxing it are discussed
the gains from the two different scenarios— below.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Substituting between migrant and native empirical evidence of the extent to which mi-
workers determines who gains grants are substitutes for natives or for exist-
The key issue of who reaps the benefits in- ing migrants is sparse. Thus in addition to
volves the degree of substitution among dif- exploring the implications of the assumptions
ferent workers. The allocation of demand for made, we also devote attention to alternative
workers assumes differentiation among work- assumptions.
ers from different regions. This is done in two Finally, in a departure from previous
steps. First, “similar” workers are bundled to- work but in line with a developing consen-
gether into “native” workers and “foreign- sus, we assume that skilled workers are near
born” workers.30 In the second step, these complements with capital (meaning that they
two bundles are decomposed into labor de- are more productive, and thus earn higher
mand by region of origin. We assume that returns, when used together with capital),
there is more differentiation between a native whereas unskilled workers are substitutes
and a foreign-born worker (that is, a lower for capital and skilled labor.32 This specifi-
substitution elasticity) than between two cation has important consequences for the
workers from different countries of origin distributional impacts of increased migra-
within each of the two aggregate bundles. For tion. Whereas investment rises with in-
example, in the case of the United States, em- creased income, the overall increase in the
ployers see a greater difference between a U.S. stock of capital is modest, so that the rise in
worker and a generic immigrant from a devel- the supply of skilled workers is not matched
oping country than between a Mexican and a by an equivalent increase in capital. Thus the
Salvadoran worker. The implication is that a marginal productivity of additional skilled
rise in the supply of migrants has a greater im- workers declines, provoking a decline in the
pact on old migrants than on native workers, wage of skilled workers (by more than the
which plays a key role in the distributional fall in the wages of unskilled workers).
outcomes of the increase in migration. The
assumption of labor demand differentiation Increased migration can generate
operates for both skilled and unskilled labor substantial changes in income distribution
in the model. among workers and owners of capital
In the default case, we assume that wages The change in factor returns is depicted in
are flexible, with a substitution elasticity be- figure 2.5. In the high-income countries only
tween unskilled migrants and natives that is
roughly comparable to that implied in the Figure 2.5 Factor returns and migration
conclusions of the meta analysis in Longhi,
% change in factor returns, 2025, relative to baseline
Nijkamp, and Poot (2004); they conclude
that a “one percentage point increase in the Developing countries
proportion of immigrants in the labor force
lowers wages across the investigated studies
by only 0.119 percent.” (See box 2.4 for a re-
view of empirical studies of the impact of mi- 10

gration on wages.) In the scenario described 15

here, the 50 percent increase in the stock of 20 countries
migrants raises their proportion of the labor 25
force by about 3 percentage points, produc- Capital Unskilled Unskilled Skilled Skilled
native foreign- native foreign-
ing a 0.5 percent decline in the wages of na- workers born workers born
tives.31 We also assume perfect substitution workers workers

between new and old migrants (the large ma- Source: World Bank simulations.
jority of both categories being unskilled). The


Box 2.4 Empirical studies of the impact

of immigration on wages
M ost cross-sectional studies find that immigrants
have no impact, or a very limited impact, on
the wages or employment of natives (LaLonde and
increased the unemployment rate by 0.2 percentage
points. This shows more adjustment through em-
ployment than through earnings compared to U.S.
Topel 1997 and Borjas, Freeman, and Katz 1997 for studies, which may be due to French rules governing
the United States; Pischke and Velling 1994 for wages (Dustmann and Glitz 2005). The comparison
Germany). However, cross-sectional approaches re- underlines the importance of the investment climate,
late wage differences across local labor markets to and in particular, labor-market flexibility, in the effi-
the share of immigrants in each market. If immi- cient absorption of migrants. Some studies show that
grants are attracted to high-wage areas, which is immigration reduces native unemployment in the
likely, it is difficult to identify the exogenous impact
long term (Poot and Cochrane 2004), presumably
of immigrants on wages.a Studies of sudden, politi-
because increased consumption demand from immi-
cally driven inflows of immigrants likewise fail to de-
grants raises the demand for labor.d
tect a significant impact on natives’ wages or employ-
ment in affected areas.b However, native workers Thus some articles support the view that unskilled
may adjust to large, sudden inflows of immigrants by immigrants are relatively close substitutes for native
moving to other areas (or through reduced inflows workers (without attempting to distinguish between
from other areas), again obscuring the relationship the effects on native workers and old migrants). How-
between immigration and labor-market outcomes.c ever, the share of low-skilled native workers in desti-
This problem has encouraged the use of panel nation countries is falling. The share of U.S. adults
techniques that can discern the combined effects of with less than a high-school education declined from
time and cross-sectional effects. Some panel studies 47 percent in 1970 to 22 percent in 1998 (Massey
have found a significant impact on the wages of un- 2000). About 90 percent of new native entrants to the
skilled natives, who in addition have suffered de- U.S. labor force in 2004 had completed high school
clines in wages due to skill-biased technical change (U.S. Labor Survey 2005). By contrast, the average
and increased trade. Borjas (2003a), analyzed the im- migrant from rural Mexico has six years of education
pact of immigration in the United States across dif- and does not speak English (Mora and Taylor 2005).
ferent levels of skills and experience and estimates Many low-skill immigrants may have such limited ed-
that immigration reduced the wages of native high- ucation and language skills that they do not compete
school dropouts in the United States by 8.9 percent with native low-skilled workers at all, but instead
from 1980 to 2000. Jaeger (1996) finds that immi- take jobs that natives are unwilling to do. In this view,
gration lowered the real wage of U.S. high-school the rise in immigration in high-income economies
dropouts by as much as 3.6 percent in the 1980s. since 1980 has been accompanied by increases in na-
DeNew and Zimmermann (1994) find that a one tive educational levels; essentially natives moved out
percentage point rise in the share of migrants in the of certain kinds of jobs, creating a demand for immi-
labor force reduces the wages of blue-collar workers grant labor.
by almost 6 percent. By contrast, Dustmann and oth-
ers (2003) find that immigration has little impact on
native wages (or employment) in the United aThe researchers do attempt to correct for endogeneity by
Kingdom, including for the low-skilled. using instrumental variables.
Where wages are relatively inflexible, an inflow of bExamples include the Mariel boatlift from Cuba (Card

migrants may affect employment levels rather than 1989), the repatriation of Algerians of European origin to France
wages. Angrist and Kugler (2002) find that increased (Hunt 1992), the inflow of workers to Austria after the break-
immigration in Europe is associated with a signifi- down of the communist regimes (Winter-Ebmer and Zweimuller
1999), and the return of Portuguese from Africa in the 1970s
cant decline in native employment, particularly for
(Carrington and de Lima 1996).
the low-skilled. Hunt (1992) finds that a one per- cStill, Card (2001) finds no evidence that immigration into
centage point rise in the share of immigrants in the an area leads to offsetting net outflows of workers.
French labor force (following Algerian independence) dSee Gross (1999) for this result for France.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

capital enjoys an increase in returns under the larger than skilled migration, so that the wage
model—with wages declining for all labor impact of unskilled migration is easier to de-
categories, skilled and unskilled, native and tect in empirical work. Second, the shock
foreign-born. With essentially only a labor modeled here represents a one-time increase in
shock, the scarcity value of capital increases. skilled migration that is larger than the exist-
The negative impact on unskilled native wages ing stock, which has built up over time. And
is small, at around 0.3 percent, depending on third, the model assumes little change in the
the assumed elasticity of substitution between capital stock, while increased investment in re-
migrant and native workers.33 The greater sponse to migration would dampen the fall in
impact is felt by existing, unskilled migrants, skilled wages, a point to which we return in
whose wages decline by more than 10 per- the conclusion to this chapter.
cent.34 At least two factors mitigate that The impact in developing countries is
decline. First, labor markets are not com- nearly the reverse. Capital returns suffer and
pletely segmented, so that part of the adjust- labor returns improve, with larger improve-
ment falls on native workers. Second, other ments for skilled workers than for unskilled
general equilibrium effects are at work, such workers. The magnitudes differ because the
as a relative shift in the demand for unskilled relative size of the shock differs. For example,
workers as the price of capital (combined with the decline in unskilled workers in developing
skilled labor) rises and a relative shift in de- countries is only 0.3 percent, versus 1.7 per-
mand toward goods that use unskilled labor cent for skilled workers.
intensively, raising the relative demand for un- Assuming that all capital income accrues
skilled workers. to native households, native households in
The impact of the shock on the wages of high-income countries are on aggregate better
skilled workers is greater than for unskilled. off after the shock, with real incomes increas-
Wages decline by 1.1 percent on average for ing by 0.4 percent. That is, the increase in
skilled natives, significantly more than for capital income more than offsets the loss in
unskilled natives. Old skilled migrants suffer a wage income. Part of the old migrants’ 6 per-
wage decline of 20 percent, which is double cent decline in real income is due to the as-
that of old unskilled migrants. The impact on sumption that they own no capital, so enjoy
skilled workers is larger than for unskilled no nonwage income.35 An alternative, ex-
because skilled workers are assumed to be near treme assumption is that on a per capita basis,
complements with capital; with capital in- old migrants receive the same amount of non-
creasing only slightly, this would tend to drive wage income as natives. This alternative
down skilled wages. And the impact is largest would reduce old migrants’ loss to 3.4 percent
on old skilled migrants because the rise in of base real income.
migration of skilled workers is large relative to To summarize, the new migrants are
the stock of old skilled migrants, and the new clearly the large winners, particularly in per-
migrants are assumed to be closer substitutes centage terms. Under the assumptions of the
for skilled migrants than are native workers. model, existing migrants are likely to be
The greater impact of migration on skilled losers—though the extent of their loss will
than unskilled wages is not at first sight con- depend on their degree of substitutabil-
sistent with the limited evidence available. In ity with native workers and their share of
those studies that find any significant impact nonwage income. Native households in both
of migration on native wages, the largest im- high-income and developing countries are
pact tends to be on unskilled wages (see better off. The sources of their gains, though,
above). Our seemingly contrary result arises are very different (figure 2.6). In the high-
for three reasons. First, unskilled immigration income countries the gains are generated by
to high-income countries has been much higher returns to capital—somewhat offset by


Figure 2.6 Source of gains for native workers

Native workers in high-income regions Native workers in developing regions

Change in real income in 2025, $ billions Change in real income in 2025, $ billions
350 200


250 150




100 50


200 100




















Source: World Bank simulations.

lower wages. The gains for natives in high- linked to real-world dynamics since, over the
income countries would be lower if we as- long run, differences in labor characteristics
sumed a more even distribution of capital (to- could fade as migrants adjust to their new
ward migrants) and a greater degree of labor environment and as employers cease to see
substitutability. In developing countries, the them as different.37
gains to natives essentially are generated by The impacts of the alternative scenario on
higher wage income and higher remittances— factor returns are shown in figure 2.7. The
somewhat offset by lower returns to capital. most notable impact is that native wages (for
These gains would be lower should the skilled and unskilled workers) decline by more
propensity of the new migrants to remit be when natives and migrants are viewed as per-
lower than average. fect substitutes for each other, while the wages
of the foreign-born decline by significantly
If migrants are viewed the same as less.38 For skilled workers, the average decline
natives, then increased migration becomes negligible; the burden of adjustment
reduces natives’ wages is spread out more evenly between native and
The degree of labor-market differentiation foreign-born workers. For the given shock,
plays a critical role in determining the effect of and depending on the assumed elasticity of
the increase in migration on native and for- substitution between foreign and native work-
eign households. An alternative scenario— ers, the impact on native wages ranges from
maintaining the same increase in migration— a slight increase to a decline of 1 percent
assumes that employers are perfectly (box 2.5).
indifferent to hiring native workers versus for- Because increasing migration constitutes a
eign-born workers.36 This empirical issue is clear labor-supply shock, one would expect it

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

still positive but significantly lower. In the

Figure 2.7 Factor returns and migration
more optimistic scenario for existing mi-
in high-income countries, 2005
grants, where native workers bear a larger
% change relative to baseline part of the burden, the trend growth in wages
is virtually identical to the baseline, declining
0 to 3.4 and 4.8 percent, respectively, for un-
5 skilled and skilled workers.
Perfect The assumption of perfect substitutability
10 substitution
of native and migrant labor has a small impact
15 on the global income gain ($379 billion instead
20 of $356 billion), but significant distributional
effects. First, in high-income countries perfect
Capital Unskilled Unskilled Skilled Skilled substitution implies a more pronounced pro-
native foreign- native foreign-
workers born workers born
capital bias, as native labor suffers a larger
workers workers loss. However, the negative impact on existing
migrant households is much smaller: less than
Source: World Bank simulations.
1 percent, compared with the 6 percent suf-
fered with labor differentiation. This has a
positive impact for developing countries, be-
to affect wages (or employment, where move- cause their loss in remittances from existing
ment in wages is constrained). But it is impor- migrants drops dramatically. Second, with per-
tant to view these changes in dynamic terms. fect substitution the new migrants benefit from
First, assuming differentiation between native a larger wage differential and thus a higher in-
and foreign-born workers, the impact on na- come gain. And again, for developing coun-
tive workers’ wages in high-income countries tries, this translates into higher remittances.
is slight even in absolute terms (–0.04 percent Overall, the change in real net remittances is
for unskilled workers and –0.4 percent for $129 billion under perfect substitutability, as
skilled workers). More important, in dynamic opposed to $88 billion. The bottom-line is that
terms, these changes alter the rate of growth the real income of new migrants increases by
of wages over the next two decades only 250 percent, as opposed to 200 percent, and
slightly. In the base case, nominal wages will the gains for natives in developing countries
increase by 3.6 and 4.7 percent, respectively rise to 1.2 percent, instead of 0.9 percent.
(average annual growth between 2001 and Several other parameters affect the relative
2025), for unskilled and skilled workers in impact of increased migration on wages and
high-income countries. With an increase in mi- returns to capital. For example, as shown in
gration, the growth rate is unchanged for un- box 2.5, the relative impact will depend on the
skilled workers and drops to 4.6 percent for substitution between capital and labor. The
skilled workers. Even in the worst-case sce- lesser the degree of substitution (or the less
nario for native workers, where foreign-born flexible the economy), the greater will be the
workers are assumed to be perfect substitutes negative impact on wages. Most econometric
for natives, the dynamic trends are almost ex- evidence suggests a capital–labor substitution
actly the same as in the baseline scenario. elasticity of around 1, somewhat higher than
The effects of the modeled shock are obvi- that used in the model.39 Another crucial as-
ously larger for existing migrants. With labor sumption in the standard model is that skilled
differentiation, their long-term wage growth workers and capital are near complements. An
trend drops to 3.1 percent for unskilled work- alternative would be to assume that unskilled
ers and 3.9 percent for skilled workers from and skilled workers were both substitutes
their baseline trend of 3.6 and 4.9 percent— with capital. This would moderate the decline


Box 2.5 Increased migration and its impact on wages

T he impact of increased migration on wages can
be summarized by a few simple formulas. The
key parameters are the share of foreign workers in
Wage impact from a 50 percent increase in
foreign workers, under various substitution
the economy, the capital-labor elasticity of substitu- assumptions
tion, and the substitution between native and foreign % change in wages
workers. These relations are summarized in the table 5
below.a The values refer to an economy where for-
eign labor is 5 percent of the labor force (sfl). The 0
capital-labor substitution elasticity (␴v) is 0.9; and
the substitution (␴l) between native and foreign 5
workers is 4.b The point elasticity is given in the workers’
third column of the table. The estimated impact of a wages
50 percent increase in the stock of foreign workers is 15 Foreign
simply the point elasticity multiplied by 50. The workers’
actual impact comes from a calibrated numerical 20
model. The two are relatively close, despite the size
of the shock. This is because the results are largely 25
driven by the low share of foreign workers in the
economy; therefore most of the economy stays near 30
its initial equilibrium. The aggregate wage falls by
only 1 percent. This is allocated across the different 1.5 2.5 3 3.5 4 8 16 32 64
workers depending on their substitutability. With an
Substitution elasticity between native and foreign workers
elasticity of 4 (used in the default scenario), migrant
workers see a 10 percent decline in wages; domestic Source: World Bank staff calculation.
workers see only a 0.5 percent decline. If migrants

Summary of elasticities with respect to an increase in foreign workers

Point Estimated Actual
Expression Description elasticity impact impact
W, F   sf sk Aggregate wage 0.0222 1.1 1.0
snl sfl sk
FW, F   l  v Wage of foreign workers 0.2597 13.0 10.1
␴ ␴

slf sfl sk
NW, F   Wage of native workers 0.0097 0.5 0.5
␴l ␴v

and native workers are perfect substitutes, then timate the substitution elasticity with great pre-
wages of native workers would decline by 1 percent, cision, particularly since the shocks are unlikely
that is, by the same amount as the aggregate wage. to be as considerable as those modeled here.
The figure shows the impact on wages—of both
foreign and native workers—using different as- aThese relations are for a small single-sector closed econ-

sumptions about their substitutability. The shock is omy but line up relatively well with the impacts from the
a 50 percent increase in the stock of foreign work- global model. Because it is closed and single-sector, the rela-
tions may not hold exactly because of other general equilib-
ers with the same assumptions used for the numer-
rium effects. See van der Mensbrugghe 2005b.
ical example in the table. The impacts on wages bThe other parameters are the capital share (s ), the share
converge only at very high levels of substitutability. of foreign labor in output (sf  sl.sfl), and the share of native
In actual econometric work, it might be hard to es- workers in the labor force (snl).

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

of skilled wages and the rise in the returns to ing native households’ gains. On the other
capital, essentially because there would be hand, increased productivity as migrants im-
more (assumed) flexibility in the economy. prove their education may generate larger
When capital and skilled workers are comple- gains to owners of capital and could benefit
ments, a sharp increase in skilled workers, native workers through spillover effects such
without a concomitant increase in capital, as training. Remittances may decline as mi-
raises the scarcity value of capital. If we make grants become more removed from the origin
capital and skilled labor more substitutable, country.
the decline in the wages of skilled workers will The process of catch-up in productivity is
create more demand for them and dampen the not captured in our current model, but we
negative impact on their wages. have done some side calculations to see how
the results could be affected. The catch-up rate
and workers’ length of stay are two factors
that affect catch-up. How long does it take the
Caveats—what the model average migrant to achieve the level of pro-
leaves out ductivity of native workers? Borjas (2003b)

W hile the scenarios discussed here provide

a wealth of insights, they do not address
many important aspects of the impact of
provides mixed evidence on this point for
migrants in the United States. First, he shows
that the catch-up rate depends very much on
migration. Using side calculations, it is possible when the migrants arrived, with earlier mi-
to get a sense of two such aspects. grants doing better than later migrants. Sec-
ond, he finds no absolute convergence, with
migrants’ wages remaining below those of na-
The model does not account for changes tive workers.
in migrant characteristics over time The second factor relates to workers’
The model assumes that migrant characteris- length of stay. The longer workers stay, the
tics do not change over time. This is useful in better placed they are to improve their skills
highlighting the immediate impact of migra- and adapt to local work practices, including
tion, but less realistic over the medium term. language skills (if necessary). At one extreme,
As migrants remain in the destination coun- all migrants may be assumed to be temporary
try, they tend to take on the characteristics of workers staying for a short period to return
native workers. For example, they learn or home permanently. Or they may be assumed
improve their fluency in the language, they to be permanent workers arriving young and
better understand (and may tend to adopt) with high educational attainment or acquired
the social mores of the destination country, skills.
and they may become more educated. Em- Under the most optimistic scenario, where
ployers are likely to view a migrant that has catch-up occurs within a year, our simple,
lived in the country for 20 years as being calibrated model predicts gains in the output
more similar to a native worker than a mi- of high-income countries that are about
grant who arrived yesterday. This issue has 25 percent higher than in the case of no catch-
several implications for the calculation of the up.40 Under a more plausible scenario, where
gains from migration. Migrants who have the process of catch-up takes 10 years and
spent a longer time in the destination country annual attrition41 is around 10 percent, the
will be less perfect substitutes for new mi- output gain in high-income countries is about
grants, mitigating the drop in their wages 12 percent higher than with no catch-up. It
predicted by the model. Similarly, as the appears, therefore, that the catch-up phenom-
degree of labor differentiation declines, the enon could boost the gains from migration
impact on native wages will rise, thus reduc- substantially.


The model does not account for the Other factors not covered by the model
potential of migration to spur higher are more difficult to quantify
investment There are various costs associated with migra-
The model generates only a modest rise in the tion that the model does not take into
capital stock as a result of the increase in mi- account. One issue concerns adjustment costs,
gration. Increased returns to capital, and thus as changes in the technology of production
interest rates, do increase savings. However, and in the mix of goods imply transitional un-
this effect is marginal, in keeping with empiri- employment and changes in the pattern of
cal estimates of the responsiveness of savings investment. The magnitude of these costs de-
to changes in interest rates. In reality, higher pends in part on the structure of labor-market
investment in response to the higher returns to institutions (such as constraints on hiring and
capital may be financed by capital inflows, firing and minimum-wage legislation) and on
which do not change in the model. There is the efficiency of capital markets. Countries
some evidence, however, that large immigra- with more flexible labor markets and sound
tion can attract capital flows. For example, banking, stock, and bond markets are likely to
Davis and Weinstein (2002) argue that skilled experience lower adjustment costs, underlin-
labor, unskilled labor, and capital are all at- ing the importance of the investment climate
tracted to the United States, owing to U.S. for realizing the potential gains from migra-
technological superiority. The mass migration tion. The size of adjustment costs will also
from Europe to the new world before World depend on whether migration is concentrated
War I encouraged large inflows of capital. or spread over time. This point also has policy
Higher investment would lessen the decline implications. If a country anticipates needing
in wages suffered by skilled workers, dampen migrants in the future or recognizes that mi-
the rise in the return to capital, and increase gration pressures are bound to rise due to de-
the demand for unskilled workers in the high- mographic changes, it would be better to
income countries—but it could have the oppo- loosen constraints on migration earlier and
site effects in developing countries. To verify more gradually than to be confronted with a
this intuition, we simulated the same migra- sharp rise later on. Migration also involves
tion shock and added to the shock a 0.4 per- direct costs, including transportation and
cent decline in the level of investment in transitional expenses, as well as the noneco-
developing countries, with a concomitant nomic costs suffered by migrants separated
transfer of these resources to high-income from their families (for example, the impact
countries.42 Those assumptions indeed have a on children raised without one or both par-
positive impact for the high-income countries ents—see chapter 3). In general these are ei-
and dampen the capital-income gains and ther short-term costs that should not greatly
labor-income losses. Overall, the gain for change the calculation of benefits from per-
native households in high-income countries manent migration or problems that decline
improves by 4.5 percent (from $139 billion to over time as migrants and families adjust to
$145 billion). But it comes at the expense of permanent changes or take steps to reunite.
natives in developing countries, whose income Several other issues that may affect the
gain drops from $143 billion to $125 billion, gains from migration are impossible to quan-
a drop of 12.5 percent. The global gains fall to tify. First, our model does not distinguish be-
$345 billion, a 3 percent fall. This suggests— tween irregular and regular migrants. If mi-
in the absence of an increase in savings—that grants are irregular, they may be paid lower
the potential reallocation of global savings wages (see chapter 3), which would reduce
toward high-income countries is negative at their welfare gains (and remittances) relative
the global level and that capital is more to the model results, while it increases the
productive in developing countries. gains of natives. However, irregular migrants

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

also may impose costs on destination diversity are associated with lower growth
countries—among them the costs of enforce- rates, holding all else equal. However, diversity
ment (as governments seek to limit what some may be more beneficial to growth at higher in-
may view as undesirable changes in the coun- come levels.44 Clearly much will depend on
try’s culture and demographic characteristics); the kinds of diversity involved: immigrants
a possible burden on public spending, which who rely on national affinities to cement loy-
may be higher for irregular migrants (see box alty to violent gangs presumably have a very
2.3); and the potential for other forms of different impact on growth and welfare than
illegality generated by the presence of a large, immigrants who open ethnic restaurants.
undocumented group of foreign workers. Fourth, the model may not fully capture
Second, immigrants may improve the effi- the beneficial effect of immigration on in-
ciency of employment from the perspective of creasing the supply of labor in the service sec-
firms by providing a source of labor that can tor. Although reductions in the prices of ser-
easily be employed in new geographic loca- vices are captured, the resulting expansion in
tions, and hired or fired in response to changes the supply of native labor (as more parents
in cyclical conditions. Piore (1986) describes can afford child care and workers have more
how many migrants (at least initially) tend to time to devote to their jobs) is not.
view their stay as temporary, filling jobs with Fifth, the model does not reflect the possi-
lower salaries and less stability than those of bility that skilled migration may lower
natives.43 Large numbers of immigrants work growth in origin countries, for example, be-
in construction, which facilitates new develop- cause of positive externalities from the pres-
ments in areas that require a mobile labor ence of skilled workers or increases in the
force. However, over time, migrants will be- price of services that require technical skills
come more permanent and demand jobs simi- (see chapter 3).
lar to those held by natives. Finally, the model assumes constant re-
Third, our model does not reflect the social turns to scale, while immigration may be
or economic implications of increasing diver- more beneficial if significant sectors enjoy
sity in the destination country. The social im- increasing returns to scale. Increasing returns
pact lies outside our present scope, but diver- may be derived, for example, from fixed pro-
sity has potential economic costs and benefits duction costs, network effects (the unit price
that should be considered. Some writers argue of providing telephone service falls as the cus-
that increased diversity has an economic tomer base grows), reduced transport and
value. Glasser, Kolko, and Saiz (2001) empha- communications costs (as the local market ex-
size the role of a rich variety of services and pands), or increased productivity due to inter-
consumer goods in enhancing the attractive- actions among highly skilled workers. In their
ness of cities. Florida (2002) relates an index role as consumers and workers, immigrants
of diversity to a concentration on high-tech- may facilitate an expansion of the market,
nology industries. Ottaviano and Peri (2004) thereby raising productivity by increasing re-
find that cultural diversity has a net positive turns. On the other hand, large inflows of im-
effect on the productivity of U.S.-born citi- migrants may induce congestion, straining
zens. By contrast, Schiff (1998) uses a theoret- public transportation systems, for example, or
ical model to underline how a society’s shared bidding up the price of land. Such effects are
values can reduce the cost of transacting busi- particular to the sector and geographic area
ness, owing to higher trust and easier enforce- involved, so it is difficult to draw broad con-
ability of sanctions. Thus immigration, which clusions. However, skilled immigrants have
increases diversity, may lower productivity by made significant contributions to high-tech-
raising transaction costs. Finally, Alesina and nology sectors that are subject to increasing
Ferrara (2004) find that increases in ethnic returns to scale.


These qualifications to the scenario results also discussing the Green Paper on an EU Approach to
illustrate how model exercises must abstract Managing Economic Migration (EU 2005).
8. These numbers will be moderated to the extent
from reality to provide quantitative measures
that labor-force participation rates in the 65 cohort
of the impact of migration. Some of the is-
are positive, if small. Moreover, labor-force participa-
sues that the model does not consider would tion rates for the elderly are likely to increase as pen-
likely be small in the medium term (adjust- sions and benefits stagnate or decline with fiscal pres-
ment costs, transportation costs) or would sures and as life expectancy rates continue to increase.
tend to increase the economic benefits of mi- We may also witness an increase in labor-force partici-
gration (improved productivity of migrants pation rates among people of working age.
9. These global gains are comparable to the recent
over time, greater labor-market flexibility and
findings by Walmsley and Winters (2003), when ad-
supply of labor). Other issues would increase
justed for the size of the economy in 2001 relative to
benefits to destination countries, while poten- the projected size of 2025.
tially harming origin countries (higher invest- 10. The model’s specification is described in van
ment, economies of scale). Still others may der Mensbrugghe (2005a).
have both economic and social effects, with- 11. “Migrants” refers to migrants from developing
out lending themselves to determinations of countries unless otherwise stated.
12. The phase-in period is somewhat arbitrary.
their direction and size (diversity, irregular
Because of its 10-year implementation, it minimizes
adjustment costs to some extent. The five-year period
between 2020 and 2025 enables an assessment of long-
run steady-state impacts.
Notes 13. This is by design. An alternative would be to
1. In keeping with the overall thrust of this report, increase the stock of migrants in proportion to their
we focus here on South-North migration, although it is current structure—by host region and skill level. In this
important to recognize that a large portion of migrants case, the largest proportional increase would be for un-
from developing countries move to other developing skilled workers in the United States.
countries. 14. A switch of 14 million workers from develop-
2. Some readers may also find the chapter too tech- ing to high-income countries has only a small impact
nical, as it necessarily deals with detailed specification (a decline of 0.4 percent) on aggregate employment in
issues—for example, the degree of differentiation be- developing countries, albeit with potentially greater
tween native and migrant workers, the fiscal impact of consequences among the relatively more scarce skilled
migrants, and how to take into account the change in workers.
prices between developed and developing countries 15. Many factors determine the level of remittances.
when evaluating gains to migrants. For example, new migrants may leave many dependents
3. Data on the stocks of migrants are generally in their home countries, which would tend to raise re-
taken from census reports in countries of destination and mittances. On the other hand, at least in the short-term,
thus include both regular and irregular migrants. How- moving and start-up costs could lower remittances.
ever, irregular migrants tend to be less likely to report 16. For simplicity, migrants from other high-
their immigrant status, so the estimate of total mi- income countries are added to the true natives.
grants is probably low. 17. Again, for simplicity, all migrants in developing
4. The breakup of the Soviet Union and emergence countries—both from rich and developing countries—
of 15 new independent countries in 1991 created new are lumped together for the purposes of the aggregate
populations of “international” migrants without mi- analysis.
gration having taken place. 18. The impact on households other than the
5. The exceptions were Belgium, France, Ireland, “new” migrants is not affected by cost-of-living adjust-
Portugal, and the United Kingdom. ments. Since these households do not move, they face the
6. In their regression equation explaining immigra- same system of prices, and thus their change in real
tion to the United States, Hatton and Williamson (2002) income simply depends on the standard real income
calculate that IRCA doubled the Mexican immigration measure.
rate from 1989 to 1991. 19. This assumes a perhaps implausible Leontief
7. Germany, Ireland, and the Czech Republic are in utility function but the purpose is simply to illustrate
the process of establishing new immigration regimes, the point that corrections need to be made for differ-
with a major focus on economic migration. The EU is ences in the cost of living.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

20. Were the true prices available, one could do a 28. Results presented at the eighth annual confer-
standard equivalent variation calculation that would ence on Global Economic Analysis held in Lübeck,
take into consideration the change in prices. Germany. See
21. In the high-income countries, new migrants’ events/conferences/2005/program_day3.htm.
total real consumption is $562 billion, compared with 29. The full merchandise trade reform scenario is
$80 billion in the baseline, hence the real increase of with a standard model and ignores any beneficial im-
$481 billion. When the $562 billion is adjusted for the pact through higher trade-induced productivity or
difference in the cost of living, real consumption, as scale economy effects. Note that the gains from reform
perceived from the point of view of the new migrant, is of services trade could be multiples of merchandise
only $244 billion, taking the change in real income trade reform. See Anderson, Martin, and van der
down to $162 billion. The cost-of-living adjustment Mensbrugghe (2005).
averages 2.3, lower than the 3.1 GDP-weighted aver- 30. In the case of high-income countries, ‘similar’
age PPP of developing countries. This occurs because workers would be migrant workers from other high-
middle-income countries (with a relatively low PPP ad- income countries. Other migrant workers are bundled
justment) have a higher weight in migration than in de- together in a so-called ‘foreign-born’ aggregate.
veloping countries’ GDP. 31. One would expect the elasticity to increase as
22. High-income countries, on the other hand, see the proportion of migrants in the population increases
a substantial rise in exports, $211 billion (2.2 percent), (box 2.5).
and a more modest $113 billion rise in net imports 32. See, for example, Bchir and others (2002).
(1.2 percent), with imports from developing countries 33. Box 2.5 shows how wages—native and
declining by $23 billion. This implies that although a foreign—are related to an increase in the stock of mi-
large part of the increase in high-income exports can be grants. Two parameters are crucial—the substitution
attributed to the increase in remittances, a significant between native and foreign workers and the share of
portion is also coming through intraregional trade foreign workers in the labor force.
among high-income countries driven in part by chang- 34. The general equilibrium elasticity is only 0.27
ing comparative advantage. for unskilled workers; for roughly a 40 percent increase
23. A standard “Dutch disease” effect of foreign in supply, wages decline by around 10 percent.
inflows. 35. Migrants from other high-income countries
24. Studies of the impact of trade reform in (also assumed to have no nonwage income) see only a
developing and industrial countries tend to show that small change in their real incomes, as their wages are
wages in developing countries rise relatively more— closely linked to the wages of native workers.
particularly for unskilled workers—than in industrial 36. Observed wage differentials can arise from a
countries, but those changes are relatively minor com- combination of two effects—differences in productiv-
pared to the initial gap in wages. For example, unre- ity and differentiated labor demand. If labor is
ported results from Anderson, Martin, and van der perfectly substitutable, then the equilibrating condi-
Mensbrugghe (2005) show that full merchandise trade tion is the equality of efficiency wages, that is,
reform would increase unskilled real wages in develop- productivity-adjusted nominal wages. If labor is differ-
ing countries by 3.7 percent (unweighted average), but entiated, efficiency wages are no longer necessarily
by only 0.7 percent in industrial countries. This could equalized, and the equilibrium wage will be determined
induce a small reduction in the incentive to migrate, by supply and demand conditions for the differentiated
but it would not substantially alter the significant wage labor.
multiple of 4 to 5 (taking into account cost-of-living 37. The empirical evidence on “catch-up” is lim-
differentials). ited. In the case of migrants to the United States, Bor-
25. The cost-of-living adjustment for the new mi- jas (2003b) shows that migrants who arrived in the
grants treats their consumption of public goods and 1960s almost caught up with natives within a 10–15
services the same as their private consumption—that is, year period. Those who arrived in the 1970s made less
it is adjusted by the same PPP factor. progress in closing the gap with natives. However, the
26. The assumption is that the new migrant house- wages of migrants arriving in the 1980s actually fell
holds come with the dependency ratio of their home further behind those of natives after a 10-year period.
country. The scenarios described in this chapter assume no
27. There will be compositional impacts in trans- change in the relative productivity of migrants. Such an
lating gains from 2025 to 2001, since developing coun- assumption would require a more elaborate specifica-
tries are growing on average more rapidly than high- tion of migrants to capture their changing composition
income countries. over time, similar to modeling capital vintages. By


ignoring the catch-up process, our results may under- Auerbach, Alan J., and Philip Oreopoulos. 1999. “An-
estimate the longer-term gains from migration. alyzing the Fiscal Impact of U.S. Immigration.”
38. The changes in wages by region of origin are American Economic Review Papers and Proceed-
identical for all workers in each high-income region, ings 89(2): 176–80.
but due to aggregation effects, this will not necessarily Bchir, Mohamed Hedi, Yvan Decreux, Jean-Louis
be true when averaging across regions. Guérin, and Sébastien Jean. 2002. “MIRAGE, a
39. The model has a vintage structure with a lower Computable General Equilibrium Model for
substitution elasticity for “old” or installed capital and Trade Policy Analysis.” CEPII Working Paper
a higher substitution elasticity for “new” capital. The 2002-17. Paris. December.
actual substitution will be a weighted average of the Bonin, Holger, Bernd Raffelhuschen, and Jan Walliser.
old and the new vintages, with a higher average for 2000. “Can Immigration Alleviate the Demo-
countries with relatively high rates of investment. graphic Burden?” Finanz Archiv 57(1): 1–21.
40. This follows directly from the assumption that Borjas, George J. 1994. “The Economics of Immi-
the productivity level of migrants is initially 75 percent gration”, Journal of Economic Literature 32
that of natives. December (1994): 1667–1717.
41. The attrition rate will be a combination of ———. 2003a. “The Labor Demand Curve IS Down-
factors—return migration, retirement, and death. The ward Sloping: Reexamining the Impact of Immi-
first factor is probably most important the first year, gration on the Labor Market.” Quarterly Journal
whereas the other two factors will depend on the age of Economics. November.
of the migrant. ———. 2003b. “The Economic Integration of
42. The value of 0.4 percent was chosen because it Immigrants in the United States: Lessons for Pol-
corresponds to the change in the number of workers in icy.” WIDER Discussion Paper DP2003/78.
developing countries—though it should be noted that World Institute for Development Economics
the change in workers represents a change in the stock Research, United Nations University, Helskinki.
level, whereas the change in investment is a change in December.
flows. Borjas, George, Richard B. Freeman, and Lawrence
43. He notes that this trend may be changing, as Katz. 1997. “How Much Do Immigration and
technology and globalization encourages smaller-scale Trade Affect Labor Market Outcomes?” Brook-
production and more permanent immigration. ings Papers on Economic Activity 1: 1–90.
44. This may occur because “the productivity ben- Butcher, Kristin F. and David Card. 1991. “Immigra-
efits of skill complementarities are realized only when tion and Wages—Evidence from the 1980s.”
the production process is sufficiently diversified,” or American Economic Review 81(2): 292–6.
because high-income economies are able to develop in- Card, David. 1989. “The Impact of the Mariel Boat
stitutions that help them cope better with the potential Lift on the Miami Labor Market.” National
for conflict inherent in ethnic diversity (Alesina and Bureau of Economic Research Working Paper
Ferrara 2004). 3069. Cambridge, MA.
———. 2001. “Immigrant Inflows, Native Outflows,
and the Local Labor Market Impacts of Higher
References Immigration.” Journal of Labor Economics 19:
Alesina, Alberto, and Eliana La Ferrara. 2004. “Ethnic 22–64.
Diversity and Economic Performance.” Centro Carrington, W., and P. de Lima. 1996. “The Impact of
Studi Luca D’Agliano Development Studies 1970s Repatriates from Africa on the Portuguese
Working Papers 193. Milan and Turin. December. Labor Market.” Industrial and Labor Relations
Anderson, Kym, Will Martin, and Dominique van der Review 49(2): 330–47.
Mensbrugghe. 2005. “Market and Welfare Impli- Collado, M. Dolores, Inigo Iturbe-Ormaetxe, and
cations of Doha Reform Scenarios.” In Agricul- Guadalupe Valera. 2004. “Quantifying the Im-
tural Trade Reform and the Doha Development pact of Immigration on the Spanish Welfare
Agenda, ed. Kym Anderson and Will Martin. State.” International Tax and Public Finance 11:
New York: Palgrave Macmillan. 335–53.
Angrist, Joshua, and Adriana Kugler. 2002. “Protective Coppel, Jonathan, Jean-Christophe Dumont, and
or Counter-Productive? European Labour Market Ignazio Visco. 2001. “Trends in Immigration and
Institutions and the Effect of Immigrants on EU Economic Consequences.” Economics Depart-
Natives.” Centre for Economic Policy Research ment Working Paper 284. Organisation for Eco-
Discussion Paper 3196. London. nomic Co-operation and Development, Paris.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Davis, Donald R., and David E. Weinstein. 2002. Market.” Industrial and Labor Relations Review
“Technological Superiority and the Losses from 45: 556–72. April.
Migration.” National Bureau of Economic IOM (International Organization for Migration).
Research Working Paper 8971. Cambridge, 2005. World Migration 2005. Geneva.
MA. Jaeger, David A. 1996. “Local Labor Markets, Admis-
DeNew, John P., and Klaus F. Zimmermann. 1994. sion Categories, and Immigrant Location
Native Wage Impacts of Foreign Labor. Journal of Choice.” Unpublished paper. Hunter College,
Population Economics 7: 177–92. New York. June.
Dustmann, Christian, and Albrecht Glitz. 2005. “Im- LaLonde, Robert J., and Robert H. Topel. 1997. “Eco-
migration, Jobs and Wages: Theory, Evidence and nomic Impact of International Migration and the
Opinion.” Centre for Research and Analysis of Economic Performance of Migrants.” In Hand-
Migration, Department of Economics, University book of Population and Family Economics,
College London. ed. Mark Rosenzweig and Oded Stark. Amster-
Dustmann, Christian, Francesca Fabbri, Ian Preston, dam: North-Holland.
and Jonathan Wadsworth. 2003. “The Local Lee, Ronald, and Timothy Miller. 2000. “Immigration,
Labour Market Effects of Immigration in the Social Security, and Broader Fiscal Impacts.”
UK.” Home Office Online Report 06/03. American Economic Review Papers and Proceed-
London. ings 90(2): 350–54.
EU (European Union). 2005. “Green Paper on an EU Longhi, Simonetta, Peter Nijkamp, and Jacques Poot.
Approach to Managing Economic Migration.” 2004. “A Meta-Analytic Assessment of the Effect
COM(2004)811 final, Commission of the Euro- of Immigration on Wages.” Population Studies
pean Communities, Brussels. November. Centre Discussion Papers 47. University of
Fehr, Hans, Sabine Jokisch, and Laurence Kotlikoff. Waikato, Hamilton, New Zealand. December.
2004. “The Role of Immigration in Dealing with Massey, Douglas. 2000. “Higher Education and Social
the Developed World’s Demographic Transition.” Mobility in the United States, 1940–1998.” Paper
National Bureau of Economic Research Working delivered at Association of American Universities
Paper 10512. Cambridge, MA. Centennial Meeting, April 17, Washington, DC.
Florida, Richard. 2002. “Bohemia and Economic Ge- Mora, Jorge, and J. Edward Taylor. 2005. “Determi-
ography.” Journal of Economic Geography 2: nants of Migration, Destination and Sector
55–71. Choice: Disentangling Individual, Household and
Glasser, Edward L., Jed Kolko, and Albert Saiz. 2001. Community Effects.” In International Migration,
“Consumer City.” Journal of Economic Geogra- Remittances, and Development, ed. Caglar
phy 1: 27–50. Ozden and Maurice Schiff, Washington, DC:
Gott, Ceri, and Karl Johnston. 2002. “The Migrant World Bank.
Population in the UK: Fiscal Effects.” RDS Occa- Moses, Jonathon W., and Bjørn Letnes. 2004. “The
sional Paper 77. Home Office, London. Economic Costs to International Labor Restric-
Gross, Dominique. 1999. “Three Million Foreigners, tions: Revisiting the Empirical Discussion.”
Three Million Unemployed: Immigration and World Development 32(10): 1609–26.
the French Labor Market.” IMF Staff Working Nana, Ganesh, and Julian Williams. 1999. “Fiscal Im-
Paper 99/124. International Monetary Fund, pacts of Migrants to New Zealand.” Report to
Washington, DC. the New Zealand Immigration Service, Auckland.
Gustafsson, B., and T. Osterberg. 2001. “Immigrants Ottaviano, Gianmarco, and Giovanni Peri. 2004. “The
and the Public Sector Budget: Accounting Exer- Economic Value of Cultural Diversity.” National
cises for Sweden.” Journal of Population Eco- Bureau of Economic Research Working Paper
nomics 14(4): 689–708. 10904. Cambridge, MA.
Gustmann, Alan L., and Thomas L. Steinmeier. 1998. OECD (Organisation for Economic Co-operation and
“Social Security Benefits of Immigrants and U.S. Development). 1997. Trends in International
Born.” National Bureau of Economic Research Migration. Paris.
Working Paper 6478. Cambridge, MA. ———. 2005. Trends in International Migration. Paris.
Hamilton, Bob, and John Whalley. 1984. “Efficiency Parsons, Christopher R., Ronald Skeldon, Terrie L.
and Distributional Implications of Global Restric- Walmsley, and L. Alan Winters. 2005. “Quantify-
tions on Labour Mobility.” Journal of Develop- ing the International Bilateral Movements of
ment Economics 14: 61–75. Migrants.” Paper presented at the Eighth Annual
Hunt, Jennifer. 1992. “The Impact of the 1962 Repa- Conference on Global Economic Analysis, June
triates from Algeria on the French Labor 9–11, Lübeck, Germany.


Passel, Jeffrey S. 2005. “Estimates of the Size and Timmer, Hans, and Dominique van der Mensbrugghe.
Characteristics of the Undocumented Popula- 2005. “Migration, PPP and the Money Metric of
tion.” Pew Hispanic Center, Washington, DC. Welfare Gains.” Development Economics Depart-
Piore, Michael J. 1986. “Perspectives on Labor Market ment, World Bank, Washington, DC.
Flexibility.” Industrial Relations 25: 146–66. U.S. Binational Study on Migration. 1997. Binational
Spring. Study: Migration Between Mexico and the United
Pischke, Jorn-Steffen, and Johannes Velling. 1994. States.
“Wage and Employment Effects of Immigration U.S. Labor Survey. 2005. Bureau of Labor Statistics.
to Germany: An Analysis Based on Local Labor
Markets.” Centre for Economic Policy Research van der Mensbrugghe, Dominique. 2005a. “LINKAGE
Discussion Paper 935. London. March. Technical Reference Document: Version 6.0.” Un-
Poot, Jacques, and Bill Cochrane. 2004. “Measuring published paper. World Bank, Washington, DC.
the Economic Impact of Immigration: A Scoping ———. 2005b. “Derivation of Output and Wage Elas-
Paper.” Immigration Research Programme, ticities Relative to an Increase in Migrants.”
New Zealand Immigration Service, Auckland. Development Economics Department, World
December. Bank, Washington, DC.
Rowthorn, Robert. 2004. “The Economic Impact of Walmsley, Terrie L., S. Amer Ahmed, and Christopher
Immigration.” Civitas Online Report. Civitas: R. Parsons. 2005. “The GMig2 Data Base: A
The Institute for the Study of Civil Society, Data Base of Bilateral Labor Migration, Wages
London. and Remittances.” GTAP Research Memoran-
Schiff, Maurice. 1998. “Trade, Migration and Welfare: dum No. 6. Center for Global Trade Analysis,
The Impact of Social Capital.” In H. Singer, N. Purdue University. September.
Hatti, and R. Tandon (eds.) Globalization, Tech- Walmsley, Terrie Louise, and L. Alan Winters. 2003.
nology, and Trade in the 21st Century. Vol. 19, “Relaxing the Restrictions on the Temporary
New World Order, Delhi: B. R. Publishing. Movements of Natural Persons: A Simulation
Smith, James P., and Barry Edmonston. 1997. The Analysis.” Centre for Economic Policy Research
New Americans: Economic, Demographic and Discussion Paper Series 3719. London. January.
Fiscal Effects of Immigration. Washington, DC: Wickramasekera, Piyasiri. 2002. “Asian Labor Migra-
National Academy Press. tion: Issues and Challenges in an Era of Global-
Sriskandarajah, Dhananjayan, Laurence Cooley, and ization.” International Labor Office, Interna-
Howard Reed. 2005. “Paying Their Way: The tional Migration Program. Geneva.
Fiscal Contribution of Immigrants in the UK.” Winter-Ebmer, Rudolf, and Josef Zweimuller. 1999.
Institute for Public Policy Research, London. “Do Immigrants Displace Young Native Workers:
Storesletten, Kjetil. 2000. “Sustaining Fiscal Policy The Austrian Experience.” Journal of Population
through Immigration.” Journal of Political Econ- Economics 12: 327–40.
omy 108(21).

The Policy Challenges
of Migration: The Origin
Countries’ Perspective

In evaluating the impact of remittances, the cations, that may be as important as the eco-
main subject of this report, it is important to nomic analysis provided here. For all of these
take into account the implications of the reasons, the analysis and policy recommenda-
initial decision to emigrate. This chapter will tions in this chapter must remain heavily qual-
analyze the implications for migrants and ori- ified. Our purpose is to signal to policymakers
gin countries of migration for economic gain in developing countries, and to the develop-
from developing to high-income countries.1 ment community in general, the elements that
Focusing on this form of migration can should be considered in formulating migration
highlight some key policy dilemmas that gov- policy.
ernments face in improving the developmental
International migration often generates great
impact of migration.
benefits for migrants and their families,
Migration is an extremely diverse phenom-
although at some risk. Migration can greatly
enon. Its economic impact on each origin
increase incomes of both migrants and their
country, and the impact of policy, will depend
families and has helped countless households
on many circumstances—among them the
escape poverty. While most workers gain greatly
skills and former employment of migrants, the
from migration, the decision to migrate is
history of migration (the existence and loca-
sometimes made with inadequate information
tion of a large diaspora), the sectors affected,
and at high risk and cost, particularly if the
patterns of trade and production, the invest-
migration is irregular. By providing information
ment climate, and the size and geographical
on migration opportunities and risks,
location of the country. For example, migra-
governments could help avoid unfortunate,
tion policies appropriate for a large develop-
costly migration decisions. Governments should
ing country with substantial low-skilled emi-
also consider means to prevent and prosecute
gration and effective institutions will differ
trafficking and other abuse of migrants, and to
from the policies for a small island economy
strengthen migration-related partnerships
with substantial high-skilled emigration and
between origin and destination countries.
weak institutions.
Migration is as complex as it is diverse, so Increasing the emigration of low-skilled
predicting the impact of policy changes will be workers would significantly reduce poverty in
problematic until more research is done and developing countries. In addition to enabling
better data obtained. In particular, the gender emigrants to escape poverty and to reducing
implications of migration are poorly under- poverty in the country of origin through
stood and require more research. Migration remittances (discussed in chapter 5),
also has important social and political impli- low-skilled emigration can increase wages and

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

reduce unemployment and underemployment influence the impact, it is impossible to reli-

of poor workers in the country of origin. ably estimate the net benefit, or cost, to origin
Many of the poorest lack the financial countries of high-skilled emigration. We can
resources or the skills required for successful only offer some rough observations that reflect
emigration to high-income countries, but the wide variation in high-skilled emigration
available data indicate that a significant rates among countries:
number of them do emigrate, although at
• Very high rates of high-skilled emigra-
lower rates than the non-poor. Reducing the
tion affect a small share of developing
restrictions on low-skill emigration, while
countries’ population, and many coun-
remaining sensitive to concerns in desti-
tries with high rates of high-skilled emi-
nation countries over social tensions, job
gration have poor investment climates
opportunities for low-skilled natives and the
that likely limit the productive employ-
potential burden on public expenditures, may
ment of high-skilled workers. On the
best be achieved through managed migration
other hand, the lack of high-skilled
programs designed jointly by origin and
workers may contribute to the poor
destination countries. Such programs
investment climate and limit the supply
should provide for temporary, low-skilled
response to economic reform.
migration—with incentives for return.
• Some countries also find it difficult to
productively employ all high-skilled
Emigration of high-skilled workers may
workers because of small economic scale
reduce living standards of those left behind and
or misguided educational policies that re-
impair growth, but can also be beneficial. Like
sult in a large supply of college graduates
low-skilled migration, high-skilled migration
for whom no suitable jobs exist.
can greatly benefit migrants and their families
• High-skilled emigration has had a severe
and can help relieve labor market pressures.
impact on public services with positive
In addition, a well-educated diaspora can
externalities. The loss of skills through
improve access to capital, technology,
high-skilled emigration has particularly
information, foreign exchange, and business
impaired health services in several devel-
contacts for firms in the country of origin. At
oping countries.
the same time, high-skilled emigration may
reduce growth in the origin country because Origin countries harmed by high-skilled
(a) other workers lose the opportunity for emigration face difficulties in managing the
training and mutually beneficial exchanges of problem. Service requirements for access to
ideas; (b) opportunities to achieve economies publicly financed education can be evaded and
of scale in skill-intensive activities may be are likely to discourage return, and proposals
reduced; (c) society loses its return on high- for the taxation of emigrants to the benefit of
skilled workers trained at public expense; (d) the origin country have made little progress.
the price of technical services (where the Improved working conditions in public em-
potential for substitution of low-skill workers ployment and investments in the infrastruc-
is limited) may rise. Highly educated citizens ture for research and development may be
may also help to improve governance, improve effective in retaining key workers. Working
the quality of debate on public issues, conditions can also be improved by strength-
encourage the education of children, and ening governance, which may require political
strengthen the administrative capacity of the will rather than money. Origin countries can
state—all of which may be reduced through also encourage educated emigrants to return
emigration of the highly skilled. by identifying job opportunities, cooperating
Because of the lack of data and the myriad with destination countries that have programs
of individual country circumstances that can to promote return, permitting dual nationality,


and helping to facilitate the portability of so- adjusted for differences in purchasing power
cial insurance benefits. (Freeman and Oostendorp 2000) (figure 3.1).
These data may overstate the wages that
migrants expect, because their earnings, at least
The migration decision and its initially, tend to be lower than those of natives
impact on migrants and their (Lucas 2004a). Moreover, many poor workers
families who lack local language skills and have mini-
mal education may find limited employment
M aking the costly and sometimes risky
decision to move to another country
generally involves the expectation of large
prospects in high-income country job markets.
On the other hand, these data may understate
increases (or lower variability) in income, the benefit of migration from the perspective
described by economists as the net present of the household. In measuring differences in
value of lifetime earnings.2 The migrant’s ex- welfare between migrants and those who do
pected income gain from emigration also re- not migrate, migrants’ earnings in high-
flects his or her employment prospects at home income countries are reduced to reflect the
and the likelihood of employment overseas. higher cost of living in high-income countries—
or purchasing power parity (PPP). To the ex-
Better economic prospects drive migration tent that migrants send earnings back home in
Migrants from developing to high-income the form of remittances, however, this adjust-
countries generally enjoy large increases in ment is not relevant, so household gains may
earnings.3 A dataset compiled by the Interna- exceed the PPP-adjusted rise in earnings.4 Fur-
tional Labour Organization (ILO) shows that thermore, the data on income differences may
workers in high-income countries earn a me- influence expectations of future earnings for
dian wage that is almost five times the level migrants and their children, and would un-
of that of workers in low-income countries, doubtedly generate much larger migration, in
the absence of controls. Evidence of substan-
tial migration pressure includes long queues of
Figure 3.1 Median wage levels for workers applicants for immigration to high-income
in the same occupation, relative to high- countries, the rise in irregular immigration,
income economies, 1988–92a the increase in asylum seekers (Hatton and
Percent Williamson 2002), and the high fees paid to
45 smugglers who help migrants cross borders
40 illegally (Cornelius 2001).
The expectation of higher earnings is not
the only economic incentive for migration.
20 Households may decide to send some mem-
15 bers abroad to diversify the family’s source of
10 income and thus reduce risk, as shocks affect-
5 ing the level of wages and the probability of
Upper-middle- Lower-middle- Low-income
employment in the destination country may
income countries income countries countries not be correlated with the shocks affecting
Source: Freeman and Oostendorp (2000).
domestic workers (Daveri and Faini 1999).5
Note: Chart reports the median wage in each country/skill
group relative to the highest wage for that skill group, with Migration involves considerable costs
the ratio in high-income economies as the numeraire. Thus
the median wage in low-income countries (averaged across Despite clear gains for many, migration involves
skill groups) is 20 percent of the median wage in high-income costs and risks that, together with restrictions
a. Adjusted for purchasing power parity.
on migration, help explain why most people
prefer to stay at home. Migration can entail

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

substantial up-front costs—transportation, fees Table 3.1 Fees charged by recruitment

charged by recruitment agencies, fees to obtain agencies
a visa and work permit, maintenance while Country US$ Year
searching for work, forgone earnings (if the
Migration to the Middle East
migrant was or could be employed at home),
Bangladesh 1,727 1995
the reduction in value of location-specific skills India 900 1995
(for example, knowing one’s native language), Pakistan 768 1995
Sri Lanka 689 1995
and the pain of being separated from family and
Sri Lanka 893 1986
familiar surroundings. Obviously these costs
Migration to Japan
will vary enormously among migrants. Thailand ⬎8,000 1996
Lack of data makes it difficult to directly Migration to Malaysia
test the relationship between costs and the de- Thailand 666–1,000 1991
cision to migrate. However, distance can be Sources: Abella 2004; Lucas 2004a; Eelens and Speckmann
used as a proxy for costs, representing not 1990; Spaan 1994.
only transport costs, but also migrants’ lim-
ited familiarity with countries of destination. 2003b). Because many migrants lack informa-
Adams and Page (2003) find that distance is tion on foreign job markets, and some agen-
a significant determinant of the direction cies may have considerable market power,
of migration from developing countries to recruitment agencies have captured a substan-
the United States, European members of the tial share of the rents generated by limits on
Organisation for Economic Co-operation and immigration (Lucas 2004a). Available data
Development (OECD), and the Arab Gulf. (mostly from past years) indicate that recruit-
Long, Tucker, and Urton (1988) found that ment charges range from $689 (in 1995) for
distance was an important constraint on inter- Sri Lankan immigrants to the Middle East to
nal migration in the United States and the $8,000 (in 1996) for Thais seeking work in
United Kingdom; what holds for internal mi- Japan (table 3.1).
gration (without political barriers or language Recruitment agencies’ potential for earning
differences, and with limited cultural differ- rents raises the issue of whether governments
ences) probably holds even more strongly for should regulate their fees. As constraints on
international migration. Household surveys migration generate significant excess profits,
also provide indirect evidence that distance efforts to regulate recruitment fees would
has an important impact on migration: while appear to have merit. However, limitations on
some of the poorest do migrate internation- the terms of mutually agreed contracts can be
ally, they are more likely to remain at home or difficult to enforce, and excessive regulation
to migrate internally (see below). can drive recruiters underground or lead them
Among the largest quantifiable costs to to switch to other countries. A few govern-
migrants are fees paid to private recruitment ments have attempted to regulate fees paid to
agencies, whose role in the international labor recruiters, and require registration and mini-
market has increased substantially, in part be- mum capital requirements or financial guar-
cause of the rise of temporary labor migration antees to limit abuses (ILO 2003a). In general,
programs.6 A major conduit of job opportuni- successful regulation of recruitment agencies
ties for migrant workers, private recruitment involves a large pool of potential migrants (to
agencies often are instrumental in seeking out reduce the likelihood of recruitment agencies
new markets for job opportunities abroad. switching to alternative sources), effective
They also provide services such as language government institutions, and regulations that
training and assistance with settlement (see focus on the most egregious abuses (instead
Xiang 2003 in the case of China). However, of just reducing the market rate). Even the
they also can be a source of abuse (ILO Philippines, considered to have a model


program, experiences difficulties in enforcing nancial incentives, such as special exemptions

fee limits, while most recruiters, including the from travel taxes for those who clear contracts
Union of Filipino Overseas Contract Workers, with government agencies, to facilitate govern-
argue that stringent regulation of recruiters ment review of contracts. Some countries with
impairs the competitive position of Philippine large numbers of emigrants offer a compre-
migrants relative to workers from other coun- hensive set of services, including predeparture
tries (Martin 2005). Destination countries training, information on labor markets in des-
are probably in a stronger position to regulate tination countries, legal services,8 reintegra-
recruitment agencies effectively. tion support, and welfare funds financed from
fees paid to origin-country governments by
Decisions to migrate are often made departing workers.9
with inadequate information
The distance and differences in language and A diaspora can reduce the costs
culture between countries of origin and desti- facing migrants
nation imply that migration is particularly The stock of emigrants in countries of destina-
affected by inadequate information. Migrants tion can reduce the costs facing new migrants
may have a distorted notion of the possibilities from the same origin country. The major
of employment and the likely wage in coun- countries of origin with significant diasporas
tries of destination, as well as insufficient in- in high-income countries (figure 3.2) are for
formation on the costs and potential risks of the United States: China, Cuba, El Salvador,
the trip. Smugglers, recruitment agencies, and India, Mexico, Philippines, and Vietnam; and
others with a financial stake in encouraging in other countries: Turkey (for Germany);
migration may present a biased picture of the Serbia and Montenegro, Morocco, and Algeria
migration experience, and poor information (for France); and China (for Japan and
increases the potential for migrants to suffer Canada). As migrant networks spread, private
from fraud and abuse.7 institutions and voluntary associations emerge
Some origin-country governments have to provide a range of services, including coun-
attempted to protect their emigrant workers seling, social services, and legal advice; lodg-
by regulating the terms of labor contracts. For ing, credit and job search assistance; and the
example, Indonesia, the Philippines, and Sri means to reduce the cost of undocumented mi-
Lanka have drafted model contracts for vari- gration, including smuggling and transport,
ous occupations and host countries that detail counterfeit documents, and arranged mar-
working terms and conditions to be specified riages (Massey and others 1993). The migrant
in advance (Abella 1997). Such efforts can be diaspora can also reduce the likelihood of, and
useful in articulating standards and informing fears concerning, abuse (see Gunatilleke 1998
migrants of their rights. However, enforcement for the example of Sri Lanka).
of such contracts in destination countries can Evidence of how the diaspora encourages
be problematic (IOM 2003). The Philippines, migration can be seen in the grouping of
India, and Bangladesh review contracts prior immigrants from the same country or local
to migrants’ departure, and Bangladesh veri- region in countries of destination. Bartel
fies the genuineness of overseas work visas. (1989) and Jaeger (2000) find that U.S. immi-
Some countries have entered into bilateral grants tend to move near former immigrants of
agreements that require destination countries the same ethnicity. Munshi (2003) finds that an
to issue visas only if the contract is approved individual is more likely to be employed and to
by the origin country. Restrictive policies of hold a higher paying nonagricultural job when
this kind run the risk of encouraging irregular a large number of migrants from his home
migration if workers cannot secure approval community are in the United States.10 Studies
of contracts in advance. Some countries use fi- of the Asia Pacific region have shown that

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 3.2 Major developing country diasporas in developed countries

Major diasporas in the United States Other major diasporas

Millions of people Millions of people

12 2


6 1


0 0


an ns

an ns

pa in

ad in
an ns
m in




Fr ia

Fr ria

Ja se

C se
Fr ca
er ks






in rb


G Tur


in roc

in lge






Source: OECD 2004.

Note: Data from the United States are based on country of birth, while most European migration data are based on nationality.

networks raise migration rates once a few ini-

Figure 3.3 Estimates of stock of irregular
tial migrants are established in destination
countries (Massey and others 1998).
% of reported migrant stock
Irregular migration is often subject
to substantial costs and risks
Irregular migration appears to have increased
significantly in major countries of destination, 40

although the estimates are unreliable. The 30

estimates are based typically on the differences 20

between census reports and other immigrant 10
registries, arrests at the border or internally, 0
and regularization programs (Jandl 2004).11




do ed




ng it

Ki Un

Irregular migration may have doubled in the





United States between 1990 and 2000, and

may now account for some nine million peo- Source: Medium estimates from Jandl 2003.
ple (Passel, Capps, and Fix 2004)—about Note: See note 11 at the end of this chapter for a definition of
“irregular” in this context.
25 percent of the total stock of migrants. The
scale of irregular migration in Europe may
also be high. Just under 700,000 irregular mi- countries provide high-skilled workers with
grants applied for regularization under the re- greater opportunities for legal entry and
cent amnesty drive in Spain. Mid-range esti- residence, and (b) it is more difficult for high-
mates provided by Jandl (2003) indicate that skilled workers to practice their professions
irregular migrants range from less than 10 without adequate documentation, such as edu-
percent of the total reported stock of migrants cational credentials (Chiswick 2000). Also,
in France to 60 percent in Greece (figure 3.3). skilled migrants often have more to lose at
Most irregular migrants are low-skilled home and may be less inclined to run the risks
because (a) immigration laws in high-income involved in irregular migration. Irregular


migrants also tend to be temporary, rather than on what is generally viewed as a basic right to
permanent, immigrants (Carter 1999).12 emigrate, are likely to be counterproductive:
Irregular migration imposes substantial they may compel many women to move as
costs on migrants, compared with permanent undocumented migrants, thus increasing their
migration. It can be more expensive: the aver- vulnerability (Misra and Rosenberg 2003).
age price in 1991 for smuggling an illegal More comprehensive, cooperative policies by
migrant from China to the United States was governments are likely to have a more positive
estimated at $30,000 and from $3,750 to effect, including the dissemination of informa-
$12,000 for migrants smuggled to Lithuania tion on the risks of migration, strengthened
(Salt and Stein 1997). Irregular migrants can protection for women in destination coun-
also be paid low wages, have poor working tries, and stepped-up identification and prose-
conditions, and be subject to violations of the cution of traffickers. Migration agreements
protections afforded under industrial-country between countries of origin, transit, and desti-
labor laws (Vayrynen 2003). Employers may nation can help achieve such policy coherence
be able to pay irregular migrants less than (as in the bilateral agreements between some
legal migrants and natives because only cer- EU states and Morocco and Tunisia, for
tain employers will hire irregular migrants, or example).
because the migrants are reluctant to move
away from support networks. Lower pay and There are costs for those left behind
higher costs of migration also make irregular Finally, migration may impose costs on family
migration less desirable for the origin country, members left behind, particularly children.
because they cut into remittances. Remittances For example, Battistella and Conaco (1996)
can be reduced by the relatively expensive find that the children of migrant parents from
money transfer operations used by irregular Luzon, Philippines, performed worse in school
migrants who lack access to bank accounts and tended to be less socially adjusted (partic-
(see chapter 6). ularly if the mother had emigrated) than chil-
Irregular migrants can also be exposed dren with both parents at home. On the other
to physical danger. Since 1994 an estimated hand, Bryant (2005) found that the improve-
2,600 undocumented migrants have died ment in the children’s health and schooling (fi-
crossing the United States–Mexico border nanced by remittances), coupled with strong
(Meek 2003). Entrapment into prostitution involvement of the extended family, tended to
is a danger for women and children (Wickra- mitigate the social costs of a parent’s migra-
masekera 2002). Trafficking in persons is tion. In general, emigration does impose hard-
estimated globally to involve some 600,000 to ships on family members left behind, but it
800,000 men, women, and children each year also improves household income and im-
(U.S. Department of State 2004). Different proves families’ ability to make compensating
national policies toward migration control adjustments that mitigate those hardships.14
make it difficult to combat trafficking and
smuggling, although the international proto- The impact of international migration
cols against these activities provide a common on countries of origin varies
instrument to criminalize them.13 Several The impact of migration on countries of origin
governments, notably in Southeast Asia, have varies greatly, depending on the size of emi-
instituted restrictions on the emigration of grant flows, the kinds of migrants, and labor
women, fearing their exploitation. Unskilled and product market conditions in the country.
and semi-skilled women are allowed to emi- In describing these effects, it is useful to dis-
grate from Bangladesh only when accompa- tinguish between skill levels, given the differ-
nied by a male partner (Siddiqui 2003). Such ences in the labor markets for low- and high-
outright bans, in addition to being limitations skilled workers.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Low-skilled migration improvement, despite massive emigration

(Lucas 2004a). The wage response to emigra-
Low-skilled migration can improve labor tion depends on the institutional setting in
market conditions for other poor workers the labor market in the home country (such as
The stock of low-skilled emigrants who moved the role of unions, public sector employment,
from developing to industrial countries in 2000 and minimum wage laws); the extent of emi-
averaged about 0.8 percent of developing gration relative to the domestic labor force;
countries’ low-skilled, working-age residents— and the degree to which emigrants were pro-
about the same as in 1990. The regions with ductively employed before migrating.
countries close to the major destination coun- The impact of international migration may
tries had relatively high rates of low-skilled differ considerably among regions within
emigration (figure 3.4).15 countries of origin, depending on the degree of
The effects of South–North migration on geographic concentration of emigration and
working conditions for low-skilled workers in the links with other regions through internal
the developing world as a whole must be migration. People in regions lying close to a
small. In individual countries, however, large- common border with the destination country
scale emigration can place increased pressure or with easier access to overseas markets (such
on wages or reduce unemployment of low- as metropolitan centers or coastal areas) have
skilled workers at the margin. For example, a tendency to migrate (Long, Tucker, and
real wages in Pakistan’s construction sector Urton 1988; Malmberg 1997). These effects
and the Philippines’ manufacturing sector can be greatly magnified through the influence
closely trace the deployment of overseas of migrant networks, once initiated from a
workers (Majid 2000; Gazdar 2003).16 Low- specific location (Gunatilleke 1998; Shah
skill emigration also may reduce underem- 1998). Internal migration to areas of high de-
ployment or raise labor-market participation parture can be quite important to the trickle-
without significant wage increases. Wage down benefits of international migration, and
trends in Albania, Bangladesh, and Sri Lanka, internal migration also can have an important
for example, display no obvious signs of poverty-reducing impact (box 3.1).

Migration of low-skilled workers

is usually beneficial
Figure 3.4 Emigration rates for low-skilled Whether emigration results in reduced under-
workers employment, increased labor-market partici-
% of low-skilled workers in home region pation, or higher wages, low-skilled workers
4 in the home labor market gain, either directly
3.5 or indirectly, from additional remittance
spending. Emigration of low-skilled workers
thus can act as a safety valve for the failure to
create appropriate employment at home.
There have been cases, however, usually in the
context of South–South migration, where
large outflows of temporary or irregular work-
ers have resulted in massive return flows due
be ca


ric n
Ea fric d

ci d

Af ara
A an

Pa an
l A an

ib ri

to economic or political shocks in destination


So ic
an ati sia
ar e

C m

th t


tra e

or as

en p

d nA
C uro

countries.17 Also, reliance on large-scale emi-






gration may retard efforts to address the issue


Source: Docquier and Marfouk 2004.

of employment expansion over the long
term, as a result of either the remittance-driven


Box 3.1 Internal versus international migration

ESCAP 2003). In Ghana, internal remittances are esti-
A lthough internal migration is much larger than
international migration, they are in some re-
spects similar phenomena. Both are largely driven by
mated to reduce the level of poverty by 14 percent,
compared with only 5 percent for international
economic disparities among regions, although con- remittances (Adams 2005).
flict and natural disasters (not discussed here) can Internal migration can have large costs on receiving
also catalyze large movements of people. Both inter- areas. The proliferation of HIV/AIDs in Ghana has
nal and international migration can be permanent or been linked to the movement of women from rural to
temporary, voluntary or forced. Both can subject mi- urban areas, where unemployment and poverty often
grants to substantial risks. Both can result in im- force them into the unprotected sex trade (Black
proved labor conditions in the regions of origin. 2004b). In some countries, internal migration to the
Internal migration in Bangladesh, for example, cities has been so massive as to increase crowding and
makes more rural land available for tenancy
place inordinate burdens on public services, which
(Afsar 2003). While the risks are often higher
lowers the quality of urban life.a
with international migration (particularly for
Internal migration often responds to substantial
irregular migrants), internal migration can also be
plagued by trafficking of persons, particularly of emigration abroad. In the Philippines, there are indica-
women and children, as evident in parts of Sub- tions of large movements from rural areas of the
Saharan Africa (Black 2004) and in India and China Philippines into the Manila region from which most
(Lee 2005). overseas workers are drawn, although this movement
Internal and international migration also have appears to have done little to help sustain wages in
their differences. Wage differentials are often lower rural areas (Saith 1997). Bangladesh has seen rapid re-
within countries than between countries, reflecting sponses of intra-village migration to replace departing
smaller differences in economic conditions within workers (Mahmud 1989). At the same time, as inter-
countries than between countries. nal migrants gain skills, resources, information, and
Internal migration may have a larger role in network contacts, they often emigrate internationally.
reducing poverty than does international migration. For example, workers displaced by falling agricultural
While the expected wage gain is lower in internal prices in southern Mexico often moved to northern
migration, poor workers may have a better chance towns to work in maquiladoras, later moving to the
of finding work domestically than in economies with United States.
higher wages, where workers’ lack of skills and lan- The links between internal and international migra-
guage ability restrict job opportunities. Moreover, tion are inadequately understood, in part because
international migration tends to be more costly, so basic data are lacking. To gather more data, it has
the poorest workers may not be able to afford it. been recommended that a migration module in demo-
While data are sparse, household surveys in a few graphic and health surveys, censuses, household in-
countries indicate the important impact that internal come and expenditure surveys, and labor force surveys
migration has on poverty. In Sierra Leone, internal be included (Afsar 2005).
remittances helped reduce income inequality in poor
areas in the 1980s (Black and others 2004). In the
aIn China, for example, rapid urbanization has been accom-
Asia Pacific region, international remittances accrue
disproportionately to richer regions, while domestic panied by the emergence of urban enclaves of landless, unem-
ployed migrants from rural areas (Pan 2004).
transfers are directed mostly to poorer regions (UN-

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

exchange-rate appreciation or the reduced remittances from abroad (used here as a proxy
pressure for policy reform. In general, how- for having a household member who emi-
ever, the opportunity to send low-skilled grated). Adams (2005) shows that less than
workers abroad provides substantial benefits 8 percent of Ghanaian households that re-
to origin countries because of the impact on ceived international remittances had estimated
labor markets and remittances. incomes (excluding remittances) that fell
within the first to fourth deciles of households
Low-skilled migration has contributed by per capita expenditures; 55 percent had ex-
to poverty alleviation penditures in the top three deciles. Lucas
The reduced supply of low-skilled workers (2004a) quotes studies of Kerala (India), Pak-
may help to alleviate poverty, if as a result of istan, the Philippines, and Thailand to support
emigration, poor people receive higher wages a conclusion that most emigrants were not
or find new opportunities to work or receive from the lowest income levels, although the
remittances (see chapter 5). Low-skilled emi- poorest did participate to some extent.
gration also alleviates poverty to the extent
that the people emigrating are poor.18 It is
unlikely, however, that a large proportion of
High-skilled emigration
migrants to industrial countries are poor
according to the World Bank’s definition There is a sharp increase in high-skilled
of poverty as living on less than $2 a day— migration
although certainly a very large share is poor The emigration of high-skilled workers from
compared to even the poorest in high-income developing countries has increased since the
countries. Most migrants from Mexico to the 1970s.20 By 1990, the stock of high-skilled
United States come from households located at South–North migrants in the United States
the middle and upper-middle levels of the in- alone was more than eight times the total
come distribution (Rivera 2005). Individuals number of high-skilled migrants from devel-
with very low incomes are unlikely to be able oping to industrial countries over the 1961–72
to obtain the financial resources necessary for period, not counting foreign students
migration (see, for example, Mahmud 1989 (Docquier and Rapoport 2004). The number
for Bangladesh). Most of the world’s poor peo- of highly educated emigrants from developing
ple live in countries that are far away from in- countries residing in OECD countries doubled
dustrial countries (Bangladesh, Brazil, China, from 1990 to 2000, compared to an approxi-
India, Indonesia, and most of the countries of mate 50 percent rise in the number of devel-
Sub-Saharan Africa), so transportation is ex- oping-country emigrants with only a primary
pensive. Moreover, many poor people lack the education (Docquier and Marfouk 2004).
rudimentary skills required to obtain a job in Rates of high-skill emigration vary enor-
industrial countries, as well as the social net- mously among developing countries, from less
works that would facilitate migration and pro- than 1 percent (Turkmenistan) to almost
vide assistance once in the destination country. 90 percent (Suriname) and by region, from
Nevertheless, the limited data indicate that 15 percent for Sub-Saharan Africa to 5 per-
the very poor do move abroad to some extent. cent for Europe and Central Asia (figure 3.5).
In Sri Lanka, returns from household surveys It is important to keep in mind this degree of
show that the share of households with a fam- diversity, as high-skill emigration can have
ily member abroad is approximately equal very different effects, depending on the size
across income groups.19 Adams (2004) pro- and economic conditions in origin countries.
vides model-based estimates implying that The increase in high-skilled migration is
about 5 percent of Guatemalan households partly due to the growing importance of
with incomes of less than $2 a day received selective immigration policies first introduced


graduate students and skilled immigrants were

Figure 3.5 Emigration rates for those with
found to be positively correlated with patent
a tertiary education, 2000
applications (Chellaraj, Maskus, and Mattoo
% of total tertiary-educated population 2005). Highly educated citizens may also
make contributions to public goods—for ex-
ample, in improving governance and strength-
ening the administrative capacity of the
state—which may be lost through high-skilled
emigration (McMahon 1999).
Second, the productivity of firms may
increase with size. If large firms require net-
works of professionals with specialized skills,
then overall productivity will be higher with
c ia

be ca

ric n

ric d

Af ara
Pa s

l A an

Af an
ib ri
Eu ific


d st A

an ati sia

So a

many professionals. For example, the value of

ar e

C m

tra e

th t
or as

en p
an Ea

C ro



telephone networks increases with the number




of people connected. Expanding networks

Source: Docquier and Marfouk 2004. efficiently may require highly technical skills.
Third, emigration of high-skilled workers
may impose a fiscal cost. In most developing
in Australia and Canada in the 1980s and countries education is heavily subsidized by
later in other OECD countries.21 Major re- the state, so that the permanent emigration of
cruiting countries have increased their intake educated workers represents a loss of fiscal
of skilled migrants and relaxed their criteria revenues.22
relating to labor-market testing and job offers. Finally, emigration of high-skilled workers
Some countries (for example, Germany, will increase the price of services that require
Norway, and the United Kingdom) have intro- technical skills. It is difficult to provide com-
duced new programs; others (such as Austria, parable levels of service with low-skilled work-
Republic of Korea, the Netherlands, and ers, and greater resources devoted to training
Sweden) offer fiscal incentives to attract talent may be lost through further emigration.
to specific sectors (OECD 2005). These pro-
grams, and the migrants themselves, are re- But high-skilled migration is often
sponding to rising skill premiums in industrial beneficial for origin countries
countries that have tightened global competi- The costs of high-skilled emigration should
tion for skilled workers. be evaluated against the beneficial effects of
migration, skilled and unskilled: increased re-
In some instances, high-skilled emigration mittances, higher wages (for migrants and
has a negative impact on living standards workers who stay home), and benefits to des-
of those left behind and on growth tination countries (see chapter 2). Moreover,
There are several reasons that migration of high-skilled emigration will have a limited im-
high-skilled workers may decrease living stan- pact if it is difficult for high-skilled workers to
dards and growth. First, the total return to find productive employment in the country of
education may be greater than the private origin. This may be the case for three reasons.
return, because highly educated workers may First, the investment climate may be so poor,
be more productive when interacting with because of political instability or other rea-
similar workers, and they may help train other sons, that many high-skilled workers cannot
workers. One statistical measure of the bene- pursue their professions. Even under such con-
ficial impact of high-skilled immigrants is ditions, however, high-skilled emigration may
that in the United States, both international be harmful if it deprives the government of

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

competent administrators and limits the Table 3.2 Emigration rates of skilled
prospects for growth once the investment workers, 2000
climate improves. Second, a significant pro- Percentage of workers with tertiary education living abroad

portion of high-skilled workers may not be Less than 10% to 20% to More than
10% 20% 30% 30%
trained in professions required by the econ-
omy, perhaps because of government subsidy Number of countries 62 33 16 28
policies. And third, some of the smallest de- Share of developing 75 19 3 3
veloping countries lack the economic scale to country population (%)

productively employ a large number of spe- Source: Docquier and Marfouk 2004.
cialized professionals.23 These issues serve to
underscore concerns over the appropriateness
of state subsidization of university education The available data indicate that high rates
in many countries.24 of high-skill emigration affect only a small
Some recent articles have claimed that share of developing countries’ population. A
high-skilled emigration, even of productively data set developed by Docquier and Marfouk
employed workers, may benefit development. (2004) indicates that the 77 countries with
The opportunity to emigrate increases the re- high-skilled emigration rates (to industrial
turns to education, leading more individuals to countries) in excess of 10 percent account for
invest in education with a view to emigrating. only one-quarter of developing-country popu-
However, only some of the educated people lation (table 3.2).26 Moreover, about half of
will actually emigrate. If the increase in human these people live in countries with very poor
capital of those unable to emigrate exceeds the investment climates (included in the bottom
loss from those who do emigrate, then soci- 25 percent of developing countries, as mea-
ety’s human capital rises following the opening sured by the United Nations’ Human Devel-
of emigration opportunities (a phenomenon opment Index), which may indicate that many
known as the “brain gain”).25 The effect will high-skilled workers face limited opportunities
be largest in countries with large stocks of to practice their professions. It is important to
emigrants (so that the probability of emigra- note that these data do not distinguish by pro-
tion is high). These models have been fession (even though high emigration rates for
questioned, however, because they assume that literature professors and physicians would
foreign firms are not able to discriminate have different economic impacts) or by quality
among educated workers (otherwise they (the emigration of a Nobel laureate physicist
would take the best qualified, and so destroy would represent a greater loss than the emi-
incentives for education by marginal candi- gration of an average university graduate).
dates), and because these models do not apply Some countries encourage skilled migra-
where family reunification programs, unre- tion. China, Cuba, India, the Philippines, Sri
lated to the skills, predominate (Schiff 2005). Lanka, and Vietnam all have programs to
facilitate training for migration, suggesting
Findings on the impact of high-skilled that some policymakers see the benefits of
emigration are mixed skilled migration—among them remittances,
It is difficult to generalize about the impact of relieving job market pressures, development
skilled migration. The dispute over gains and of an extensive diaspora, and expectations
losses has remained largely conjectural and that many migrants will eventually return
has not been settled by the available empirical with improved skills (as discussed below).
studies. On balance, it is not possible at pre- Direct, cross-country tests of the relation-
sent to provide an aggregate, reliable estimate ship between high-skilled emigration and
of the true impact of high-skilled emigration. growth have been mixed. The preponderance
Some partial conclusions follow. of evidence supports the view that education


makes an important contribution to growth.27 abroad, but also by poor working conditions
Beine, Docquier, and Rapoport (2001) de- and public sector services in origin countries.
tected a positive and significant impact on
human capital formation from the opportu- Origin countries face considerable
nity for emigration, whereas Faini (2003) difficulties in limiting high-skilled
found that a higher probability of migration emigration
for workers with secondary education had no Even if high-skilled emigration were found to
visible impact on secondary educational be detrimental to living standards and growth,
achievement in the home country. countries of origin would face serious ob-
stacles in reducing it. Some countries have
High-skilled emigration has had enormous required that graduates of publicly funded
impact on some sectors, especially health education work for a period of time in public
The sectoral distribution of high-skilled emi- sector jobs. But such requirements can be
grants is important for assessing the implica- evaded, and their existence is likely to discour-
tions for countries of origin. Meyer and age return of migrants to the country of origin.
Brown (1999) estimate that about 12 percent Several proposals have been made for interna-
of developing-country nationals trained in sci- tional schemes to tax high-skilled emigrants,
ence and technology live in the United States. with the funds earmarked for developing coun-
If accurate, these estimates suggest that high- tries. Such schemes have made no progress, as
skilled emigration may be much more serious they would be hard to enforce. Calculating the
for production than shown by the data from welfare loss from high-skilled emigration and
Docquier and Marfouk (2004) given above, thus setting an appropriate level of tax would
where total high-skilled emigration to indus- be difficult. Moreover, the schemes would
trial countries was estimated at about 8 per- require the cooperation of migrants and coun-
cent of the stock of high-skilled developing- tries of destination—something not likely to be
country nationals. achieved. Bhaghwati (1976) advocates that
High-skilled emigration may have a partic- developing countries should subject their na-
ularly severe impact on the health sector, and tionals working abroad to local taxes, as does
the emigration of doctors and nurses may re- the United States. However, many developing
duce the likelihood of some countries meeting countries would find such a system of taxation
the Millennium Development Goals for re- difficult to administer.
ducing child mortality, improving maternal Some governments encourage skilled work-
health, and combating HIV/AIDS and tubercu- ers to stay by improving working conditions,
losis. Chanda (2001) estimates that at least 12 providing research facilities, and giving incen-
percent of the doctors trained in India live in tives for research (see the discussion of incen-
the United Kingdom, that Ethiopia lost half of tives for return, below). China has reported
its pathology graduates from 1984 to 1996, a nine-fold increase from 1995 to 2003 in
that Pakistan loses half of its medical school foreign programs offered in cooperation with
graduates every year; and that in Ghana only local institutions, which has resulted in lower
about one-third of medical school graduates numbers of students going abroad (Vincent-
remain in the country. Perhaps one-half of the Lancrin 2004). Such programs may require
graduates of South African medical schools substantial resources, and poorer countries
emigrate to high-income countries (Pang, will face difficulties in creating the conditions
Lansang, and Haines 2002), and Jamaica required to retain their most-skilled workers.
had to train five doctors, and Grenada 22, to In some cases improvements in governance,
keep just one (Stalker 1994).28 Of course, the which may require political determination
incentive for migration is often conditioned rather than large expenditures, may help to
not only by the opportunity for higher earnings retain workers.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Cooperation from destination countries generating possibilities for codevelopment be-

may help tween firms in the countries of origin and desti-
Effective means of limiting high-skilled emi- nation, and expanding technical cooperation.
gration, or increasing its benefits, will require With the growth of outsourcing of manufactur-
the cooperation of countries of destination. ing components and telecommunications and
One example has been the United Kingdom’s other services, diaspora networks may be of in-
code of practice for the international recruit- creasing importance. However, despite the
ment of health-care workers, which restricts broad agreement on the importance of diaspo-
the active solicitation of health care profes- ras and the many anecdotal comments on how
sionals from developing countries. However, they have assisted development, it is difficult to
implementation has been difficult because the quantify these benefits.
code does not apply to private sector recruiters The diaspora can be a significant source of
(House of Commons 2004). A recent proposal foreign-exchange earnings (beyond remit-
to require countries of destination to provide tances) for countries with sizeable emigration.
Caribbean countries with subsidies for train- Orozco (2003) documents diaspora-related
ing health professionals could increase the increases in exports and tourism receipts for
supply of health workers in both origin and Central America. Gould (1994) and Head
destination countries (Commonwealth Secre- and Ries (1998) found that increased emigra-
tariat 2005).29 More broadly, contributions by tion to the United States and Canada raised
destination countries to education in origin exports from countries of origin.
countries could both compensate origin- There is some evidence that the diaspora
country governments for their training of emi- plays an important role in the transfer of
grants and improve the qualifications of work- knowledge between destination and origin
ers coming to destination countries. Another countries. Agrawal, Cockburn, and McHale
approach, which would improve the coherence (2003) find that patent applications are likely
of development policies, would be for destina- to be filed in both the country of residence and
tion countries to increase their investment in the country of origin. Meyer and Brown (1999)
sectors in which they lack skills, rather than and Brown (2000) identify Internet-based ex-
“raiding” those skills from poor countries. patriate networks of skilled professionals and
students that facilitate the transfer of knowl-
edge. However, the effectiveness of these net-
Diasporas works is open to question: less than half of the
A large diaspora can expand market 61 Internet-based networks examined in 2004
access for origin countries were updated regularly, and only 56 percent
A potent benefit of high-skilled emigration is were updated within the past year.30 Origin-
the creation of a large, well-educated diaspora, country governments can help maintain ties to
which improves access to capital, information, the diaspora by supporting professional net-
and contacts for firms in countries of origin. Im- works, promoting dialogue with government,
migrants play a role in facilitating trade by pro- and funding educational, linguistic, and cul-
viding information and helping to enforce con- tural programs.
tracts (Rauch and Trindade 1999) and by acting
as intermediaries that can match buyers with re-
liable local suppliers (Yusuf 2001). Johnson and
Sedaca (2004) emphasize that diasporas can act The return of expatriates can
as “first movers” who catalyze growth oppor- benefit development
tunities and make connections between markets
that otherwise would not exist. Barré and oth-
ers (2003) cite the importance of diasporas in
T he return of expatriates to their home
country is widely perceived as benefitting
development (Ellerman 2003). Expatriates may


be more effective than foreigners in transfer- permanent basis, through changes in regu-
ring knowledge back home because of their laion. Both can allow dual citizenship, an
understanding of local culture. However, re- increasingly common practice (Aleinikoff and
turnees may also represent retirees, or the less- Klusmeyer 2002). Origin countries can elimi-
skilled of the skilled cohort of emigrants nate rules that prevent emigrants from owning
(Borjas and Bratsberg 1994; Lowell and or investing in property back home. Permanent
Findlay 2001), or may have difficulties in residents can be protected from losing their
readapting to the home country (Faini 2003), status if they leave for a relatively limited pe-
or their skills may have deteriorated while riod of time, as this discourages productive
abroad (Ghosh 1996). Returnees may be those temporary returns to the origin country.
disappointed by the wages or working condi- Destination countries can also allow returning
tions or may have more difficulty in finding or migrants to benefit from the rights they acquire
retaining jobs.31 during their work abroad, such as pensions,
A range of programs have been established health insurance, and disability programs
to encourage return of highly educated na- (Holzmann, Koettl, and Chernetsky 2005).34
tionals living abroad, with mixed results. Such arrangements, however, require effective
Among developing-country governments, for institutions in the origin country to provide
example, China, the Philippines, Taiwan such services and are best achieved through ne-
(China), Thailand, and Tunisia have offered a gotiations between origin and destination
wide range of incentives, including research countries.
funding, access to foreign exchange, expanded Destination countries have provided various
real estate investment options, and study incentives for the return of migrants. For exam-
opportunities.32 ple, France has provided loans and technical
The domestic policy environment is critical assistance to migrants from Mali and Senegal
to productive return. Cervantes and Guellec to establish businesses in their home countries.
(2002) cite the favorable impact of returning However, few of the businesses appear to have
expatriates in the Republic of Korea, attracted been successful, either because of the inade-
by strong research and development (R&D) quate investment climate in the recipient com-
environments and infrastructure investments. munities (Gubert 2005) or because participants
Industrial parks helped to lure entrepreneurs had worked in low-level jobs in France and
back to China. In Taiwan (China), the Hsinchu lacked entrepreneurial skills (Magoni 2004).
Industrial Park attracted more than 5,000 Many of these programs are quite small.35
returning scientists in 2000 alone (Saxenian International organizations, too, have man-
2002). Conversely, a poor investment climate aged programs to promote return, although
will inhibit return. In Armenia, barriers to they tend to cover few emigrants. The IOM’s
foreign direct investment (FDI) and inade- Return of Qualified African Nationals pro-
quate enforcement of contracts have prevented gram successfully attracted more than 2,000
a more active involvement of the Armenian di- highly skilled persons back to 41 African
aspora in local development (Gevorkyan and countries from 1974 to 1990, and the pro-
Grigorian 2003). Saxenian (2000) cites the re- gram was later expanded to the Migration for
luctance of Indian entrepreneurs to return be- Development in Africa program (MIDA).36
cause of government regulations that increase Similar programs have been run for Latin
the administrative cost of operating a American countries, Afghanistan, and
business—although the Indian diaspora has Bosnia and Herzegovina.37 The United Na-
contributed to the development of information tions Development Programme’s TOKTEN
technology in Bangalore.33 project promotes temporary return (three-
Both origin and destination countries can week to three-month development assign-
help facilitate return, on both a temporary and ments), which is often easier to achieve.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Temporary migration and Temporary migration may facilitate

international agreements greater migration flows
While temporary migration programs can gen-
T he number of temporary workers admit-
ted to high-income countries under skill-
based programs rose substantially in the
erate benefits similar to those of permanent mi-
gration (such as higher incomes for migrants,
improved efficiency in destination countries,
1990s, doubling in the United Kingdom and
remittances for origin countries), in some re-
almost quadrupling in the United States (fig-
spects those benefits differ between permanent
ure 3.6). Foreign student programs are also
and temporary migration. For the legal mi-
proliferating, in part due to competitive mar-
grant, being admitted on a temporary basis
keting and entry policies by Australia,
may be less desirable than permanent
Canada, Germany, and the United Kingdom.
residence, which provides for free movement
Inflows of unskilled, seasonal workers also
between destination and origin countries.
have increased since the early 1990s in most
However, the existence of temporary migration
high-income countries that have such pro-
programs (reflecting resistance to permanent
grams (UN 2004). Greater emphasis on tem-
migration) may facilitate larger legal flows.
porary migration may reflect opposition to
From the point of view of the destination
expanding permanent migration.
country, temporary migration offers the flexi-
Temporary labor migration schemes vary
bility to target required skills and to adjust
greatly from country to country. Skilled mi-
entry in response to changes in labor demand.
grants tend to have more opportunities under
Temporary migration can reduce fiscal pres-
unilateral visa programs (such as the U.S.
sures that may be associated with low-skilled
H-1B visa and the temporary skilled migration
immigration, in that temporary migrants tend
programs of Australia and Canada), while
not to bring their dependents (who may re-
unskilled migrants often must rely on bilateral
quire public services). At the same time, pro-
or regional agreements, such as the seasonal
grams can be made conditional on employ-
work programs between Germany and coun-
ment, thereby limiting social tensions from
tries in central and eastern Europe.
immigration. Provisions for temporary migra-
tion also can be viewed as offering an alterna-
tive to irregular migration.
The record of destination governments in
Figure 3.6 Number of temporary workers ensuring that temporary migration programs
admitted under skill-based programs
are indeed temporary has been mixed. For
Thousands example, over 22 years, most workers in the
600 Mexico–U.S. guest workers program returned
500 at the end of their seasonal jobs, although some
Australia found ways to obtain permanent status (Martin
Canada 2003). Similarly, from 1960 to 1973, three-
300 United Statesa
quarters of the 18.5 million foreigners who
United Kingdom
came to work in Germany left. But 25 percent
100 remained. Coupled with rules that allowed
many to eventually bring their dependents, this
1992 2000 resulted in significant permanent settlement.
There is some evidence that the recent guest
Source: United Nations.
Note: Double counting occurs if a person enters the United worker programs in Europe, particularly those
States more than once during the year. involving subcontracting for short-term pro-
a. Reflects the number of admissions under H-1B visas,
not the number of persons. jects and some return incentives, have managed
higher return rates (Lucas 2004a).


There are advantages and disadvantages eastern European states, and Saudi Arabia and
to temporary migration Egypt and Libya. There are several hundred
Reliance on repeated, temporary migration such agreements worldwide, including some
for workers also has some economic draw- 168 signed in the last 50 years in Latin America
backs. Hiring a temporary immigrant may alone, half in the past 10 years (IOM 2005a).38
mean a shorter duration of employment com- Bilateral agreements could improve the bene-
pared with hiring a permanent immigrant, and fits of temporary migration for origin countries
thus higher costs for training. Temporary mi- through greater certainty of access and condi-
grants are also less likely than permanent ones tions. This may be particularly important in
to invest in skills specific to the destination markets where increased competition from
country (such as language proficiency and other suppliers might lead to a reduction in ac-
licensing requirements), because the returns cess (as occurred in Saudi Arabia and the Mid-
are enjoyed over a shorter period of time dle East in the 1990s). Bilateral agreements can
(Chiswick 2000). Nevertheless, for emigrants help build the confidence in both origin and
from developing countries, the wage differen- destination countries that a particular channel
tials are so large that they may justify sub- of migration will generate real benefits and
stantial investments in acquiring such skills, minimize costs—for example, that migrants
even for temporary stays. will be treated well and will return at the end of
For the origin country, remittances (and their contract.
repatriation of assets) may be higher with Several factors impede the maximization
temporary migration, because temporary mi- of gains from bilateral agreements, however.
grants are less likely to bring their dependents Some origin countries may lack sufficiently
and more likely to maintain close ties with reliable information on demand for their
the home country. Perhaps most importantly, workers in destination countries, and in which
temporary migration programs can provide an sectors, to negotiate appropriate agreements.
opening to increase legal, unskilled migration, Destination countries may likewise have dif-
which generates the greatest developmental ficulty reliably estimating labor shortages in
impact for origin countries, as already noted. particular sectors. And origin countries may
On the other hand, temporary migration may face resource constraints in implementing
provide a less reliable means of exporting obligations with regard to prescreening of mi-
large labor surpluses, as cancellation of future grants or monitoring of their return, although
access is easier for destination countries than these may be covered by the destination
expelling existing migrants. However, it is this country (as in a nurses program between
flexibility (coupled with less long-run popula- Romania and Italy). Origin countries may also
tion pressure, fewer concerns over integration, lack bargaining power to conclude terms
and fewer pension commitments) that makes favorable to them or to conclude agreements
temporary migration desirable for destination at all. For example, of 18 bilateral agreements
countries, thus facilitating agreements for proposed by the Philippines with countries
larger unskilled migration (Winters 2005). in Africa, Asia, Europe, and the Middle
East, five countries refused to enter into
Bilateral agreements can play an agreements, and others have remained inac-
important role in low-skilled, temporary tive (Go 2004).
migration Nevertheless, there is scattered evidence
Bilateral labor agreements have become a that countries like the Philippines have been
major vehicle for low-skilled, seasonal workers able to use bilateral agreements to gain
in agriculture, tourism, and construction, as favorable employment conditions for their
evidenced by agreements between the United migrants—and in some cases to support their
States and Mexico, Germany and central and return and reintegration (Lucas 2004b).

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Bilateral agreements can help ensure that Agreement on Trade in Services (GATS) has
origin-country credentials are accepted in the potential to improve cooperation on labor
destination countries, for example.39 They can services between countries of origin and desti-
help ensure that temporary migration is indeed nation, but so far it has not facilitated a signif-
temporary, and that returning migrants are icant rise in cross-border labor movements
reintegrated, by supporting the transfer of (box 3.2). The ILO has pioneered the develop-
technology and human resource development ment of international instruments for protect-
in the origin country, as under Spain’s agree- ing the rights of migrant workers, and the UN
ments with Colombia and Ecuador (IOM General Assembly adopted the International
2005a). Bilateral agreements can also ensure Convention on the Protection of the Rights of
that the origin country cooperates in monitor- All Migrant Workers and Members of Their
ing and managing migration, for example, by Families, which clearly defines the rights of
incorporating a readmission provision (as in the migrant workers, including irregular workers
1997 agreement between Italy and Albania). (Wickramasekera 2002). The convention en-
They also can limit the effects of brain tered into force in 2004. However, none of the
drain. For example, a pilot scheme between major destination countries have ratified it yet,
the Dutch and Polish ministries of health and its means of enforcement are limited.
prepared Polish nurses for employment in the
Dutch health care system for a maximum International agreements governing
period of two years and to facilitate their migration contrast sharply with those
subsequent return and reintegration into the for trade
Polish health care system.40 Other proposals A major impulse behind the General Agree-
take a development-cooperation approach, ment on Tariffs and Trade and its successor,
under which destination countries fund the the World Trade Organization, was that mul-
training (to their standards) of a given number tilateral agreements that provide for nondis-
of nurses in excess of origin country demand, crimination among countries would maximize
with the surplus nurses granted temporary the gains from trade. By contrast, there is lit-
visas to work in destination countries for a tle support for multilateral, nondiscriminatory
specified period, with guaranteed return. approaches to migration, at least in destina-
tion countries. In part this is because the
Except in the EU, regional and economic implications of nondiscrimination
international agreements have had differ between trade and migration. In trade,
little impact on migration nondiscrimination maximizes economic effi-
At the regional level, there has been some ciency by allowing the lowest-cost supplier to
progress on removing technical and adminis- compete, thus reducing prices and forcing
trative barriers to the cross-border exchange high-cost producers to improve efficiency or
of skilled personnel for business purposes in exit the market. But labor markets in high-
Africa, Europe, Latin America, and parts of income countries are generally not permitted
Asia. Also, several consultative processes on (through minimum-wage laws and social-
migration have emerged at the regional and insurance schemes) to adjust fully to the
global levels.41 However, with the major ex- lowest-cost supplier. Thus the benefits of
ception of the EU, most regional arrangements nondiscrimination are weaker in migration
have had little impact on the free movement of than in trade. U.S. consumers benefit if Indian
less-skilled foreign workers or on permanent shirts are cheaper than Mexican shirts, but
migration (World Bank 2005). U.S. employers benefit little if Indians are will-
International treaties have had only limited ing to work for less than Mexicans—the
impact on migration. Mode 4 of the General decline in wages is limited.


Box 3.2 Mode 4 and international migration

T he WTO General Agreement on Trade in Ser-
vices (GATS) does not cover labor migration per
se, but rather the narrower concept of movement of
treatment, while migration regimes seek to retain
flexibility to make adjustments in line with labor-
market conditions. Second, GATS commitments
people across borders as one of four modes of deliv- follow the most-favored-nation (MFN) principle—
ering services. Mode 4 covers the temporary move- that is, treatment offered to one country must be
ment of persons across borders for the purpose of extended to all WTO members—whereas migration
supplying a service. “Temporary” movement is not regimes can offer special treatment to countries with
defined, but permanent migration is explicitly which regulatory trust or other special relationships
excluded, as are workers in most nonservice sectors, have been built (through visa waiver programs, for
such as agriculture or manufacturing. example). Third, GATS Mode 4 covers only a rela-
Mode 4 service suppliers can be viewed in terms tively limited subset of the workers moving around
of both duration and purpose of stay: they enter a the globe. Agricultural workers, for example, are not
country for a specific purpose (to fulfill a service generally viewed as falling under the GATS. Fourth,
contract), for a limited and (generally) specified pe- multilateral trade negotiations have a 50-year history,
riod of time, and are usually confined to one sector while migration has largely remained a national pol-
or one job (they do not enter the labor market and icy prerogative characterized by limited international
are not free to search for employment). Mode 4 is dialogue. Finally, movement of people, especially the
normally understood to include business visitors lower-skilled, raises a raft of issues related to social
(persons who come for three months or less to nego- and cultural integration, exploitation, impact on
tiate a contract), intracorporate transferees (persons local labor markets, and, more recently, security that
transferred within a company from one country to trade agreements are ill-equipped to address.
another), and suppliers of contractual services (indi- Against this background, what role might the
viduals or employees of foreign companies with a GATS play in managing labor mobility? The GATS
contract to supply a service to a client in the receiv- is a narrow, but sharp, instrument that can deliver a
ing country). While Mode 4 includes persons at all powerful guarantee of access, but only for certain
skill levels, to date market-opening commitments by types of workers. Beyond this, however, GATS nego-
WTO members have been limited to the highly tiations can be used to create a sense of urgency that
skilled. may serve to bring migration authorities to the table
Relatively few market-opening commitments on to discuss ways to manage mobility. Bilateral or re-
Mode 4 have been made by both developed and de- gional approaches could, for example, include low-
veloping countries. Those that have been made tend skilled workers and develop creative, cooperative ap-
to be subject to restrictions on number, type, and du- proaches to issues such as remittance transfer, brain
ration of stay of service suppliers. Countries’ actual drain, and loss of investment in education, prescreen-
regimes for temporary entry of workers tend to be ing of temporary workers, and return. These agree-
more liberal than their GATS commitments, how- ments could assist in building regulatory trust and
ever, and considerable movement on service issues is improving management schemes in receiving and
occurring in a range of sectors (such as health), sending countries. Over time, by creating a template
notwithstanding the near-absence of relevant GATS of basic requirements or criteria that could be ap-
commitments. plied to all countries on a nondiscriminatory basis,
Five issues arise regarding GATS Mode 4 as an they could be used to extend access to a wider group
instrument to manage labor mobility. First, GATS of countries and so approach the MFN principle of
commitments are fixed commitments of guaranteed GATS.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

But the largest reason that nondiscrimina- among other variables. Solimano (2002) found that
tory approaches are limited is that people are real per capita income differentials between Argentina
and source countries were the main determinant of net
not goods: migration has much broader impli-
migration flows in the twentieth century.
cations for society than does trade. Destina-
3. See also chapter 2, which points out that
tion countries tend to be concerned that migrants’ earnings (per worker) increase eleven-fold,
immigrants from countries with very different before adjusting for differences in purchasing power
cultures will not integrate easily into society, in high-income versus developing countries.
and high-income countries tend to limit low- 4. The basic idea is that the opportunity to earn
income migrants for fear of overburdening money based on developed country prices but spend it
(through remittances) based on developing country
public services (see chapter 2). Thus even
prices is a major benefit from migration. The same
those countries that have immigration regimes
adjustment from the perspective of the migrant is
that do not discriminate by country tend to discussed in the modeling exercise in chapter 2.
discriminate by level of skill. 5. See the discussion of remittances and smoothing
A final important distinction between trade of household consumption in chapter 5.
and migration is that trade is subject to 6. By the late 1990s, public employment services
relatively effective regulation, while many already played an insignificant role in the recruitment
of foreign workers, except where migration was
countries of destination face considerable
covered by bilateral labor agreements (ILO 1997). For
difficulties (and internal disagreements) in reg-
example, nine out of ten workers sent from Asia
ulating immigration. The lack of effective have used private recruiters (Abella 1997); and for
regulation and incomplete efforts to control Romania, most jobs in countries with which the gov-
immigration encourages many low-skilled mi- ernment has not secured bilateral agreements are found
grants to run substantial risks that can lead to by private intermediaries (Diminescu 2004).
conditions akin to slavery, great physical dan- 7. Hugo (2004) describes how work contractors
are the primary source of information for potential
ger, and even death. On the other hand, the
migrants from Indonesia, and relates this to the high
same lack of control works to the advantage of
levels of exploitation of Indonesian contract workers
migrants by offering opportunities that might compared with workers from other countries.
not otherwise exist and by benefiting groups 8. Support services provided through Philippine
within destination countries. The evidence in labor attachés have provided critical legal counseling
this chapter suggests that cooperation between and protection (Moreno-Fontes Chammartin 2005).
origin and destination countries, through 9. The funds operated by the Philippines, Pakistan
and Sri Lanka provide scholarships, legal aid in
agreements that provide for temporary, low-
destination countries, insurance against death and
skilled migration, and through enforcement of
disability, and loans for predeparture costs, housing,
laws protecting migrants from exploitation and self-employment. The administration and delivery
and abuse, can improve the impact of migra- can often be difficult, particularly on insurance (Tan
tion for countries and for migrants. 2004), and some emigrants may resent the mandatory
nature of the schemes (Abella 1997).
10. Also regarding Mexico, Mora and Taylor
Notes (2005) find that the presence of a family member in the
1. Of course, migration may arise out of a combi- United States increases by 7 percent the probability
nation of economic, political, and social goals. Also, that an individual will migrate, while McKenzie (2005)
migration among developing countries is an increas- shows that larger migration networks increase the
ingly important phenomenon, but given data limita- probability of other community members migrating.
tions we focus here on migration from developing to 11. An irregular migrant in this context is defined
industrial countries. as any person entering, residing, and working in a coun-
2. Empirical work largely confirms the view that try without proper documentation of their legal status
income differentials are important determinants of in that country, or any person who has committed a
migration. Borjas (1987), Karemera, Oguledo, and crime or breach of immigration law in that country and
Davis (2000), and Hatton and Williamson (2002) therefore is not entitled to remain in that country.
found that migration to the United States was nega- 12. However, Cornelius (2001) notes that the share
tively related to source-country income per capita, of irregular migrants who settle permanently in the


United States has increased—a trend accelerated by 1.5 million totaled about 6 percent of gross national
tighter border enforcement adopted in the mid-1990s. income, compared with an average of 1.7 percent for
13. Protocol to Prevent, Suppress and Punish Traf- all developing countries.
ficking in Persons, Especially Women and Children; 24. Available data do not distinguish émigrés
and Protocol against the Smuggling of Migrants by educated at home from those educated abroad, an issue
Land, Sea and Air, 2000. These supplement the Con- of growing importance as education is increasingly
vention against Transnational Crime, 2000. marketed to the developing world by high-income
14. Dedicated government offices such as the countries.
Philippines Overseas Workers Administration, unions 25. This theory is developed in Mountford (1997),
such as the seamen’s union in the Philippines, and non- Chau and Stark (1999), Stark (2003), and Drinkwater
governmental organizations (NGOs) can help families and others (2002).
and communities make adjustments when family mem- 26. These data do not include high-skilled emi-
bers migrate. grants to other developing countries, which may be an
15. These data are described in Docquier and important issue for many developing countries.
Marfouk (2004), which relies on census data (plus 27. Microeconomic evidence tends to find that
extensive estimations), and thus undercounts irregular education is associated with higher earnings (Mincer
migrants, who are mostly low skilled. The data are 1991). After some considerable debate, recent articles
taken largely from industrial countries, so that low- find that years of schooling have a positive impact on
skilled migration to other developing countries, as well productivity growth (de la Fuente and Domenech
as high-income countries in the Middle East and Asia, 2002), and that the quality of education (as measured,
is not reflected (which, for example, reduces the ratio for example, by pupil-student ratios or the dropout
of low-skilled emigrants from South Asia). rate) may matter more than the quantity (Barro and
16. Similarly, mine labor recruiting in South Africa Lee 2000).
increased wages in the plantation sectors in both 28. Clemens (2005) presents an alternative view,
Malawi and Mozambique, which ultimately resulted in arguing that health systems in Africa are not greatly
the curtailed permission to recruit in Malawi in the weakened by emigration because the option to emi-
early 1970s (Lucas 1987). grate encourages entry into the medical field.
17. For example, the compulsory repatriation of 29. Institutions exist in countries of origin that
workers to Kerala following the Gulf conflict in train workers for external labor markets, for example
1990–91 threw Kerala into a fairly sharp recession some nursing schools in the Philippines and a medical
(Lucas 2004b). school in Budapest that teaches in German. However,
18. This is not invariably true, for example, if the these schools do not receive funds from potential coun-
departure of one household member leaves his or her tries of destination (World Bank 2004).
dependents impoverished. In general, family income is 30. Data are based on a survey carried out for a
likely to rise with emigration, but cases of real hardship background paper, available on request.
caused by emigration do exist. 31. Workers who stayed in Albania had higher-
19. This calculation is based on income that in- quality skills than returnees (De Coulon and Piracha
cludes remittances from the emigrant, so the number of 2002), and returnees from Sweden were found to be
Sri Lankan households with a family member abroad less successful economically than emigrants who stayed
that were poor prior to migration is probably larger. (Edin, LaLonde, and Åslund 2000). See also Hugo
20. Much of the data on high-skilled migration (2002) on re-emigration from Australia and Constant
refers to individuals who have some tertiary education, and Massey (2003) on Germany.
although other kinds of qualifications (electrician, 32. See Pang, Lansing, and Haines 2002 on
plumber, ability to handle sophisticated machinery) are Thailand, Lucas (2004a) on China, and the IOM office
of economic interest. in Tunis on Tunisia.
21. See Lowell (2001) for a list of programs to 33. The High-Level Committee on the Indian
attract high-skilled workers. Diaspora (2001) notes the role of expatriates in at-
22. For example, forgone income tax revenues tracting R&D investments from Intel, Oracle, Texas
associated with Indian-born residents of the United Instruments, Sun Microsystems, and IBM.
States may be equal to one-third of current individual 34. Recognizing benefits earned abroad may
income tax receipts in India (Desai, Kapur, and reduce costs to the destination-country government, as
McHale 2001), although this is a very low share of many such services are likely to be less expensive in de-
total government revenues. veloping countries. The cost implications have some
23. Such countries benefit highly from remittances. uncertainty, as some migrants will choose to return
Remittances to countries with populations of less than even without portability of benefits.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

35. For example, a program between Germany and Adams, Richard, and John Page. 2003. “International
Eritrea disbursed only 65 loans from 1993 to 1997, Migration, Remittances, and Poverty in Develop-
while a project in Italy’s Veneto region trained only 30 ing Countries.” Policy Research Working Paper
Albanian immigrants in setting up companies or 3179. World Bank, Washington, DC.
launching joint ventures with local companies in their Afsar, Rita. 2003. “Internal Migration and the Devel-
country of origin. opment Nexus: The Case of Bangladesh.”
36. The Return of Qualified Africans program was Bangladesh Institute of Development Studies,
evaluated by the European Commission as contribut- Dhaka.
ing to development at micro levels (IOM 2005b). ———. 2005. “Internal Migration and Pro-Poor
37. The Return for Qualified Afghans has been Policy.” In Migration, Development, and Poverty
criticized for offering low compensation packages and Reduction in Asia, ed. Gervais Appave and Frank
involving few individuals despite being expensive to Laczko. Geneva: International Organization for
run (Jazayery 2002). Migration.
38. Australia, Argentina, Canada, and the United Agrawal, Ajay, Iain Cockburn, and John McHale.
States entered into bilateral labor agreements with 2003. “Gone But Not Forgotten: Labor Flows,
countries of origin in the mid-twentieth century. The Knowledge Spillovers, and Enduring Social Capi-
bracero program admitted some five million Mexican tal.” Working Paper 9950. National Bureau of
farm workers to the United States between 1942 and Economic Research, Cambridge, MA.
1966. In Europe, Germany and France recruited guest Aleinikoff, T. A., and Klusmeyer, D. 2002. Citizenship
workers from southern Europe, Turkey, and North Policies for an Age of Migration. Washington,
Africa after the Second World War until the economic DC: Carnegie Endowment for International Peace.
downturn of the 1970s. Barré, Rémi, Valéria Hernandez, Jean-Baptiste Meyer,
39. An example is the agreement that provides for and Dominique Vinck. 2003. Diasporas scien-
the acceptance of Vietnamese information technology tifiques: Comment les pays en développement
credentials in Japan (Vietnam Trade 2005). peuvent-ils tirer parti de leurs chercheurs et de
40. The pilot ended in January 2005, and the out- leurs ingénieurs? (Scientific diasporas: how can
comes are being evaluated by the Dutch government. developing countries benefit from their expatriate
41. Regional examples include the Regional Con- scientists and engineers?) Paris: IRD Editions.
ference on Migration (Puebla Process) and Lima Process Barro, R., and J. W. Lee. 2000. “International Data on
in the Americas; MIDSA and MIDWA in Africa; and the Educational Attainments—Updates and Implica-
Manila, APC, and Bali Processes in Asia. Inter-regional tions.” Working Paper 7911. National Bureau of
processes include the “5 plus 5” (a migration dialogue Economic Research, Cambridge, MA.
established in 2002 between southern Europe—France, Bartel, A. P. 1989. “Where Do the New U.S. Immi-
Italy, Malta, Portugal, Spain—and the Maghreb grants Live?” Journal of Labor Economics 7(4):
group—Algeria, Libya, Mauritania, Morocco, and 371–91.
Tunisia. Global consultation forums include the UN Battistella, Graziano, and Ma. Cecilia G. Conaco.
Global Commission on International Migration, the 1996. “Impact of Migration on the Children Left
Berne Initiative, IOM’s International Dialogue on Mi- Behind.” Asian Migrant 9(3): 86–91.
gration, and ILO’s International Labor Conference. Beine, Michel, Frederic Docquier, and Hillel Rapoport.
2001. “Brain Drain and Economic Growth:
Theory and Evidence.” Journal of Development
References Economics 64: 275–89.
Abella, M. 1997. Sending Workers Abroad: A Manual Bhagwati, Jagdish. 1976. “The Brain Drain.”
for Low- and Middle-Income Countries. Geneva: International Social Science Journal 28: 691–729.
International Labour Office. Black, R., 2004. “Migration and Pro-Poor Policy in
———. 2004. “The Role of Recruiters in Labor Africa.” Working Paper C6. Sussex Centre for
Migration.” In International Migration: Migration Research, University of Sussex.
Prospects and Policies in a Global Market, ed. Black, Richard, Savinna Ammassari, Shannon Mouil-
Douglas S. Massey and J. Edward Taylor. Oxford: lesseaux, and Radha Rajkotia. 2004. “Migration
Oxford University Press. and Pro Poor Policy in West Africa.” Working
Adams, Richard. 2004. “Remittances and Poverty in Paper C8. Sussex Centre for Migration Research,
Guatemala.” Policy Research Working Paper University of Sussex.
3418. World Bank, Washington, DC. Borjas, George. J. 1987. “Self-Selection and Earnings
———. 2005. “Remittances and Poverty in Ghana.” of Immigrants.” American Economic Review
Mimeograph. World Bank, Washington, DC. 77(4): 531–53.


Borjas, George J., and Bernt Bratsberg. 1994. “Who Cornelius, Wayne A. 2001. “Death at the Border: The
Leaves? The Outmigration of the Foreign Born.” Efficacy and ‘Unintended’ Consequences of U.S.
National Bureau of Economic Research Working Immigration Control Policy, 1993–2000.”
Paper 4913. Cambridge, MA. Working Paper 27, Center for Comparative
Brown, Mercy. 2000. “Using the Intellectual Diaspora Immigration Studies, University of California–
to Reverse the Brain Drain: Some Useful Exam- San Diego.
ples.” Paper presented at the Regional Conference Daveri, Francesco, and Riccardo Faini. 1999. “Where
on Brain Drain and Capacity Building in Africa. Do Migrants Go?” Oxford Economic Papers 51:
Addis Ababa, February 22–24. 595–622.
Bryant, John. 2005. “Children of International Mi- abstract/51/4/595.
grants in Indonesia, Thailand, and the Philip- De Coulon, Augustin, and Matloob Piracha. 2002.
pines: A Review of Evidence and Policies.” Work- “Self-Selection and the Performance of Return
ing Paper 2005-05. UNICEF Innocenti Research Migrants: The Case of Albania.” Discussion
Center, Florence. Paper 0211. Department of Economics, Univer-
Carter, Thomas J. 1999. “Illegal Immigration in an sity of Kent, Canterbury.
Efficiency Wage Model.” Journal of International De la Fuente, A., and R. Domenech. 2002. “Human
Economics 49: 385–401. Capital in Growth Regressions: How Much Dif-
Cervantes, M., and D. Guellec. 2002. “The Brain ference Does Data Quality Make? An Update and
Drain: Old Myths, New Realities.” Observer, Further Results.” Unpublished paper. Universidad
May 7. Autonoma, Barcelona.
Chanda, Rupa. 2001. “Trade in Health Services.” Desai, Mihir A., Devesh Kapur, and John McHale.
Paper prepared for the Working Group on Health 2001. “The Fiscal Impact of the Brain Drain:
and International Economy of the Commission Indian Emigration to the U.S.” Prepared for the
on Macroeconomics and Health, World Health Third Annual NBER-NCAER conference, Dec-
Organization, Geneva. ember 17–18. Harvard University and National
Chau, Nancy H., and Oded Stark. 1999. “Migra- Bureau for Economic Research, Cambridge,
tion under Asymmetric Information and Human MA.
Capital Formation.” Review of International Diminescu, Dana. 2004. “Assessment and Evaluation
Economics 7(3): 455–83. of Bilateral Labour Agreements Signed by
Chellaraj, Gnanaraj, Keith Maskus, and Aaditya Romania.” In Migration for Employment: Bilat-
Mattoo. 2005. “The Contribution of Skilled Im- eral Agreements at a Crossroads. Paris: Organisa-
migration and International Graduate Students tion for Economic Co-operation and Develop-
to U.S. Innovation.” In International Migration, ment.
Remittances, and Development, ed. Caglar Docquier, Frederic, and Abdeslam Marfouk. 2004.
Ozden and Maurice Schiff. Washington, DC: “Measuring the International Mobility of Skilled
World Bank. Workers (1990–2000).” Policy Research Working
Chiswick, Barry R. 1988. “Illegal Immigration and Paper 3381. Development Research Group,
Immigration Control.” Journal of Economic Per- World Bank, Washington, DC.
spectives 2(3): 101–15. Docquier, Frederic, and Hillel Rapoport. 2004.
———. 2000. “The Economics of Illegal Migration “Skilled Migration: The Perspective of Develop-
for the Host Economy.” Paper presented at the ing Countries.” Policy Research Working Paper
National Association for Business Economics 3382. Development Research Group, World
Annual Meeting. September. Bank, Washington, DC.
Clemens, Michael. 2005. “Do No Harm—Is the Emi- Drinkwater, Stephen, Paul Levine, Emanuela Lotti, and
gration of Health Professionals Bad for Africa?” Joseph Pearlman. 2002. “The Economic Impact
Prepared for the G-20 Workshop on Demo- of Migration: A Survey.” Paper prepared for
graphic Challenges and Migration. Sydney, Second Workshop of the Fifth Framework Pro-
Australia, August. gramme Project, “European Enlargement: The
Commonwealth Secretariat. 2005. “A Managed Impact of East-West Migration on Growth and
Migration Program for Teachers and Nurses.” Employment,” December 6–7, Vienna.
London. Edin, Per-Anders, Robert J. LaLonde, and Olof
Constant, Amelie, and Douglas S. Massey. 2003. “Self- Åslund. 2000. “Emigration of Immigrants and
Selection, Earnings, and Out-Migration: A Longi- Measures of Immigrant Assimilation: Evidence
tudinal Study of Immigrants to Germany.” Popu- from Sweden.” Swedish Economic Policy Review
lation Economics 16: 631–53. November. 7: 163–204. Fall.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Eelens, Frank, and J. D. Speckmann. 1990. “Recruit- Head, K., and J. Ries. 1998. “Immigration and Trade
ment of Labor Migrants for the Middle East: Creation: Econometric Evidence from Canada.”
The Sri Lankan Case.” International Migration Canadian Journal of Economics 31(1).
Review 24(2): 297–322. High-Level Committee on the Indian Diaspora. 2001.
Ellerman, David. 2003. “Policy Research on Migration “The Indian Diaspora.” Ministry of External
and Development.” World Bank Policy Research Affairs, Government of India. December.
Working Paper 3117. Washington, DC.
Faini, Riccardo. 2005. “Does the Brain Drain Boost Holzmann, Robert, Johannes Koettl, Taras Chernetsky.
Growth?” Research Program on International 2005. “Portability Regimes of Pension and
Migration and Development. DECRG. Mimeo. Health Care Benefits for International Migrants:
World Bank. Washington, DC. An Analysis of Issues and Good Practices.” Social
Freeman, Richard B., and Remco H. Oostendorp. Protection Discussion Paper Series 0519. World
2000. “Wages Around the World: Pay across Bank, Washington, DC.
Occupations and Countries.” Working Paper House of Commons. 2004. “Migration and Develop-
8058. National Bureau of Economic Research, ment: How to Make Migration Work for Poverty
Cambridge, MA. Reduction.” Sixth Report of Session 2003–04,
Gazdar, Haris. 2003, “A Review of Migration Issues in HC 79-1. London. July.
Pakistan.” Paper presented at the Regional Con- Hugo, Graeme J. 2002, “Migration Policies Designed
ference on Migration, Development and Pro-Poor to Facilitate the Recruitment of Skilled Workers
Choices in Asia, June 22–24, Dhaka. in Australia.” In International Mobility of the
Gevorkyan, Alexandr V., and David A. Grigorian. Highly Skilled. Paris: Organisation for Economic
2003. “Armenia and Its Diaspora: Is There Scope Co-operation and Development.
for a Stronger Economic Link?” Armenian Forum ———. 2004. “Information, Exploitation, and Em-
3(2): 1–35. powerment: The Case of Indonesian Contract
Ghosh, Bimal. 1996. “Economic Migration and the Workers.” Unpublished paper. University of
Sending Countries.” In The Economics of Labor Adelaide.
Migration, ed. Julien van den Broek. Cheltenham, ILO (International Labour Office). 1997. “Protecting
UK: Edward Elgar. the Most Vulnerable of Today’s Workers.” Inter-
Go, S. 2004. “Fighting for the Rights of Migrant national Migration Branch. Geneva.
Workers: The Case of the Philippines.” In Migra- ———. 2003a. “Booklet 2: Decision-Making and
tion for Employment: Bilateral Agreeements at a Preparing for Employment Abroad.” In Prevent-
Crossroads. Paris. Organisation of Economic Co- ing Discrimination, Exploitation and Abuse of
operation and Development. Women Migrant Workers: An Information
Gould, David. 1994. “Immigrant Links to the Home Guide. Geneva: International Labour Office.
Country: Empirical Implications for U.S. Bilateral ———. 2003b. “Booklet 4: Working and Living
Trade Flows.” Review of Economics and Statis- Abroad.” In Preventing Discrimination, Ex-
tics 76: 302–16. May. ploitation and Abuse of Women Migrant Work-
Gubert, Flore. 2005. “Migrant Remittances and Their ers: An Information Guide. Geneva: International
Impact on the Economic Development of Sending Labour Office.
Countries: The Case of Africa.” Paper presented IOM (International Organization for Migration).
at OECD International Conference on Migra- 2003. Labour Migration in Asia. Geneva.
tion, Remittances, and the Economic Develop- ———. 2005a. Migration, Development, and Poverty
ment of Sending Countries, February 23–25, Reduction in Asia. Geneva.
Marrakech. ———. 2005b. World Migration 2005. Geneva.
Gunatilleke, Godfrey. 1998, “The Role of Social Jandl, Michael. 2003. “Estimates on the Numbers of
Networks and Community Structures in Interna- Illegal and Smuggled Immigrants in Europe.” Pre-
tional Migration from Sri Lanka.” In Emigration sentation at the Eighth International Metropolis
Dynamics in Developing Countries II: South Conference of the Inernational Centre for Migra-
Asia, ed. Reginald Appleyard. Aldershot, tion Policy Development, September 17, Vienna.
England: Ashgate. ———. 2004. “The Estimation of Illegal Migration in
Hatton, Timothy J., and Jeffrey G. Williamson. 2002. Europe.” Studi Emigrazione/Migration Studies
“What Fundamentals Drive World Migration?” 61(153): 141–55. March.
Discussion Paper 3559. Centre for Economic Jaeger, David A. 2000. “Local Labor Markets, Admis-
Policy Research, London. sion Categories, and Immigrant Location


Choice.” Unpublished paper. Hunter College, Mahmud, Wahiduddin. 1989. “The Impact of Over-
New York. seas Labour Migration on the Bangladesh Econ-
Jazayery, Leila. 2002. “The Migration-Development omy: A Macro-Economic Perspective.” In To the
Nexus: Afghanistan Case Study.” International Gulf and Back: Studies on the Economic Impact
Migration 40(5): 231–52. of Asian Labour Migration, ed. Rashid Amjad.
Johnson, Brett, and Santiago Sedaca. 2004. “Diaspo- New Delhi: ILO-ARTEP.
ras, Émigrés, and Development: Economic Majid, Nomaan. 2000, “Pakistan: Employment, Out-
Linkages and Programmatic Responses.” Study put and Productivity.” Issues in Development
conducted for the Trade Enhancement Service Discussion Paper 33. International Labour Office,
Sector (TESS) Project under contract with the U.S. Geneva.
Agency for International Development, Carana Malmberg, Gunnar. 1997, “Time and Space in Interna-
Corporation, Washington, DC. January. tional Migration.” In International Migration, Im-
Karemera, David, Victor I. Oguledo, and Bobby Davis. mobility and Development: Multidisciplinary Per-
2000. “A Gravity Model Analysis of Interna- spectives, ed. Tomas Hammar, Grete Brochmann,
tional Migration to North America.” Applied Kristof Tamas, and Thomas Faist. Oxford: Berg.
Economics 32(13): 1745–55. Martin, Philip. 2003. “Managing Labor Migration:
Kanbur, Ravi, and Hillel Rapoport. 2005. “Migration Temporary Worker Programs for the Twenty-First
Selectivity and the Evolution of Spatial Century.” International Institute for Labour
Inequality.” Journal of Economic Geography 5: Studies, Geneva.
1–15. ———. 2005. “Merchants of Labor: Agents of Evolv-
Lee, J. 2005. “Human Trafficking in East Asia: Cur- ing Migration Infrastructure.” Discussion Paper
rent Trends, Data Collection, and Knowledge 158/2005. International Institute for Labour
Gap.” In Data and Research on Human Traffick- Studies, Geneva.
ing: A Global Survey. Geneva. IOM. Massey, Douglas S., Joaquin Arango, Graeme Hugo,
Long, Larry, C. Jack Tucker, and William L. Urton. 1988. Ali Kouaouci, Adela Pellegrino, and J. Edward
“Migration Distances: An International Com- Taylor. 1993. “Theories of International Migra-
parison.” Demography 25: 633–60. November. tion: A Review and Appraisal.” Population and
Lowell, B. Lindsay. 2001. “Policy Responses to the In- Development Review 19(3): 431–66.
ternational Mobility of Skilled Labour.” Interna- ———. 1998. Worlds in Motion: Understanding Inter-
tional Labour Office, International Migration national Migration at the End of the Millennium.
Branch, Geneva. Oxford: Clarendon.
Lowell, B. Lindsay, and Allan Findlay. 2001. “Migra- McKenzie, David J. 2005. “Beyond Remittances: The
tion of Highly Skilled Persons from Developing Effects of Migration on Mexican Households.” In
Countries: Impact and Policy Responses. Synthe- International Migration, Remittances, and Devel-
sis Report.” International Labour Office, Interna- opment, ed. Caglar Ozden and Maurice Schiff.
tional Migration Branch, Geneva. New York: Palgrave Macmillan.
Lucas, Robert E.B. 1987. “Emigration to South McMahon, Walter M. 1999. Education and Develop-
Africa’s Mines.” American Economic Review 77: ment: Measuring the Social Benefits. Oxford:
313–30. June. Oxford University Press.
———. 2004a. International Migration Regimes Meek, Miki. 2003. “Life and Death on the Southwest
and Economic Development. Report from the Border.” National Geographic. November.
seminar of the Executive Group on Development Meyer, Jean-Baptiste, and Mercy Brown. 1999. “Scien-
Issues on International Migration Regimes and tific Diasporas: A New Approach to the Brain
Economic Development, May 13, Stockholm. Drain.” Paper prepared for the UNESCO-ICSU World Conference on Science, June 26–July 1,
———. 2004b. “International Migration to the High Budapest.
Income Countries: Some Consequences for the Mincer, Jacob. 1991. “Human Capital, Technology,
Sending Countries.” Unpublished paper. Boston and the Wage Structure: What Do Time Series
University. Economics Department. Show?” Working Paper 3581. National Bureau of
Magoni, Raphaele. 2004. “France.” In International Economic Research, Cambridge, MA.
Migration and Relations with Third Countries: Misra, Neha, and Ruth Rosenberg. 2003. “Migrant
European and U.S. Approaches, eds. Jan Niessen Workers.” In Trafficking of Women and Children
and Yongmi Schibel. Brussels: Migration Policy in Indonesia, ed. Ruth Rosenberg. Jakarta: Inter-
Group. national Catholic Migration Commission

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

(ICMC) and American Center for International Salt, John, and Jeremy Stein. 1997. “Migration as a
Labor Solidarity (Solidarity Center). Business: The Case of Trafficking.” International
Mora, Jorge, and J. Edward Taylor. 2005. “Determi- Migration 35(4).
nants of Migration, Destination and Sector Saxenian, AnnaLee. 2000. “Brain Drain or Brain Cir-
Choice: Disentangling Individual, Household, and culation? The Silicon Valley–Asia Connection.”
Community Effects.” In International Migration, Lecture in the Modern Asia Series, Harvard
Remittances, and Development, ed. Caglar Ozden University Asia Center, September 29.
and Maurice Schiff. New York: Palgrave ———. 2002. Local and Global Networks of Im-
Macmillan. migrant Professionals in Silicon Valley. San Fran-
Moreno-Fontes Chammartin, G. 2005. “Domestic cisco: Public Policy Institute of California.
Workers: Little Protection for the Underpaid.” Schiff, Maurice. 2005. “Brain Gain: Its Size and Im-
Migration Information Source. April. www. pact on Welfare and Growth Have Been Greatly Exaggerated.” In Caglar Ozden and Maurice
Mountford, Andrew. 1997. “Can a Brain Drain Be Schiff (eds.) International Migration, Remittances
Good for Growth in the Source Economy?” and the Brain Drain. New York: Palgrave
Journal of Development Economics 53(2): Macmillan.
287–303. Shah, Nasra M. 1998, “The Role of Social Networks
Munshi, Kaivan. 2003. “Networks in the Modern among South Asian Male Migrants in Kuwait.”
Economy: Mexican Migrants in the United States In Emigration Dynamics in Developing Countries
Labor Market.” Quarterly Journal of Economics II: South Asia, ed. Reginald Appleyard. Aldershot,
118(2): England: Ashgate.
OECD (Organisation for Economic Co-operation and Siddiqui, T. 2003. “Migration as a Livelihood Strategy
Development). 2004. Trends in International of the Poor: The Bangladesh Case.” Paper pre-
Migration. Paris. sented at the Regional Conference on Migration,
———. 2005. Trends in International Migration. Development, and Pro-Poor Policy Choices in
Paris. Asia, June 22–24, Dhaka, Bangladesh.
Orozco, Manuel. 2003. “Worker Remittances, Solimano, Andres. 2002. “International Migration,
Transnationalism, and Development.” Inter- Globalization and Development: Main Issues.”
American Dialogue, Washington, DC. Unpublished paper. UN Economic Commission for
Pan, W. 2004. “Yikao nongmin gaosu tuijing cheng- Latin America and the Caribbean, Santiago, Chile.
shihua” (Making farmers accelerate urbaniza- Spaan, Ernst. 1994. “Taikongs and Calos: The Role of
tion). Zhanlue Yu Guanli (Strategy and Manage- Middlemen and Brokers in Javanese International
ment) 6(2). Migration.” International Migration Review
Pang, Tikki, Mary Ann Lansang, and Andy Haines. 28(1): 93–113.
2002. “Brain Drain and Health Professionals.” Stalker. 1994. “The Work of Strangers: A Survey of
British Medical Journal 324(7336): 499–500. International Labour Migration.” International
March 2. Labour Office, Geneva.
Passel, J. S., R. Capps, and M. E. Fix. 2004. Undocu- Stark, Oded. 2003. “Rethinking the Brain Drain.”
mented Immigrants: Facts and Figures. Urban World Development. 32(1): 15–22.
Institute, Washington, DC. Tan, E. 2004. “Welfare Funds for Migrant Workers:
url.cfm?ID⫽1000587. A Comparative Study of Pakistan, Philippines,
Rauch, James E., and Victor Trindade. 1999. “Ethnic and Sri Lanka.” International Organization for
Chinese Networks in International Trade.” Migration, Geneva.
NBER Working Paper 7189, National Bureau of UNESCAP (UN Economic and Social Commission for
Economic Research, Cambridge, MA. June. Asia and the Pacific). 2003. “Migration Patterns
Rivera, Jose Jorge Mora. 2005. “The Impact of Migra- and Policies in the Asian and Pacific Region.”
tion and Remittances on Distribution and Sources Asian Population Studies Series 160. United Na-
of Income: The Mexican Rural Case.” Paper tions, New York.
prepared for the meeting of the United Nations UN (United Nations). 2004. World Economic and
Expert Group on International Migration and Social Survey. New York.
Development, July 6–8, New York. U.S. Department of State. 2004. Trafficking in Persons
Saith, Ashwani. 1997. “Emigration Pressures and Report. Washington, DC.
Structural Change: Case of the Philippines.” In- Vayrynen, Raimo. 2003. “Illegal Immigration, Human
ternational Migration Papers 19. International Trafficking, and Organized Crime.” Discussion
Labour Office, Geneva. Paper 2003/72. World Institute for Development


Economics Research, United Nations University, Globalization.” International Labour Office, In-
Helsinki. ternational Migration Program, Geneva.
Vietnam Trade. 2005. “Japan Accepts Credentials of Winters, Alan. 2005. “Demographic Transition and
Vietnam’s IT Engineers.” Vietnam Trade. January the Temporary Mobility of Labor.” Paper pre-
20.⫽1&id⫽9& pared for the G-20 Workshop on Demographic
nid⫽5783&lang⫽en&start\⫽20&archive⫽. Challenges and Migration, August, Sydney.
Vincent-Lancrin, S. 2004. “Building Capacity through World Bank. 2005. Global Economic Prospects.
Cross-Border Tertiary Education.” Paper pre- Washington, DC.
pared for the conference of the UNESCO/OECD Xiang, B. 2003. “Emigration from China: A Sending
Australia Forum on Trade in Educational Ser- Country Perspective.” International Organization
vices, “Bridging the Divide: Building Capacity for for Migration, Geneva.
Post-Secondary Education through Cross-Border Yusuf, Shahid. 2001. “Globalization and the Challenge
Provision,” October 11–12, Sydney. www.oecd. for Developing Countries.” Background paper for
org/dataoecd/43/25/33784331.pdf. World Development Report 1999/2000. World
Wickramasekera, Piyasiri. 2002. “Asian Labour Bank, Washington, DC.
Migration: Issues and Challenges in an Era of

Trends, Determinants, and
Macroeconomic Effects
of Remittances

Chapter 3 reviewed the trends, opportunities, them; and (3) designing policies to reduce the
and policy challenges associated with inter- transaction costs of remittances, strengthen
national migration. It also introduced the the formal financial infrastructure supporting
economic importance of the funds that inter- remittances, and leverage remittances to im-
national migrants send back to their country prove access to financial services in recipient
of origin. In recent years, those funds have economies.
emerged as a major source of external financ- Officially recorded remittance estimates
ing in developing countries. Although there is may significantly underestimate the real mag-
no universal agreement yet on how to measure nitude of remittances. Model-based estimates
international migrants’ remittances to devel- and household surveys suggest that informal
oping countries, a comprehensive measure of flows could add at least 50 percent to the offi-
certain officially recorded flows—workers’ cial estimate, with significant regional and
remittances, compensation of employees, and country variation. The true size of remittance
migrant transfers—produced an estimate of flows could be even larger, in view of substan-
$167 billion for 2005, up from $160 billion in tial underrecording of flows through formal
2004. Given measurement uncertainties, channels.
notably the unknown extent of unrecorded Despite the prominence given to remit-
flows through formal and informal channels, tances from developed countries, South–South
the true size of remittance flows may be much remittance flows make up 30–45 percent of
higher—perhaps 50 percent or more. Because total remittances received by developing coun-
of their volume and their potential to reduce tries, reflecting the fact that over half of
poverty, remittances are attracting growing migrants from developing countries migrate to
attention from policymakers at the highest other developing countries. Remittance flows
levels in both developed and developing to poor countries originate largely in the
countries.1 middle-income developing countries.
This chapter and chapters 5 and 6 consider Recorded remittance flows have surged in
remittances from several angles. The organiz- recent years, driven by a combination of
ing framework is driven by three items on the factors—among them better data collection,
international policy agenda: (1) understanding reflecting greater awareness of the develop-
the true size and trends in remittance flows to ment potential of remittances, as well as con-
developing countries, as well as their macro- cerns about money laundering and terrorist
economic impact; (2) evaluating the impact of financing; lower costs and wider networks in
remittances on the households that receive the industry that supports remittance; and

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

growth in the number of migrants and their The plan of this chapter is as follows. In the
incomes. Government policies to improve next section, trends in remittance flows to de-
banking access and the technology of money veloping countries are presented along with a
transfers have also helped increase the flow of range of estimates for their true size—that is,
remittances and promote their transfer with informal flows included. We identify the
through formal channels. major sending and receiving countries, includ-
Efforts to encourage remittances, however, ing those in the South. In the following sec-
sometimes generate unwanted effects. Tax tion, we examine the factors affecting remit-
incentives may attract remittance inflows, for tance flows, including the prospects for future
example, but they also create opportunities remittance growth, and policies and regula-
for tax evasion. Likewise, matching-fund tions in source and destination countries that
programs for migrant associations may chan- affect the cost of remittances. In the final sec-
nel collective remittances to development tion, we consider the macroeconomic effects
projects, but in so doing they may divert funds of remittances, including the effects on sta-
from other local funding priorities. bility, country creditworthiness, international
For some recipient countries, remittances capital-market access, the real exchange rate,
are large enough to have broader macroeco- and competitiveness.
nomic implications. By generating a steady
stream of foreign-exchange earnings, they
can improve a country’s creditworthiness for Remittance data and trends
external borrowing, and through innovative
financing mechanisms (such as securitization),
they can expand access to capital and lower
T he quality and coverage of data on remit-
tances leave much to be desired. First, there
is no consensus on the boundaries of the phe-
borrowing costs. While large and sustained nomenon under study. Should only workers’ re-
remittance inflows can contribute to currency mittances be counted, or should we include
appreciation and so affect the production of compensation of employees and migrant trans-
cost-sensitive tradables (such as labor- fers? (See annex 4A.1 for more details on these
intensive manufactures), this outcome may nomenclatural disputes.) Second, in several
be less severe than it is in the case of natural-re- countries, many types of formal remittance
source earnings (since remittances are distrib- flows go unrecorded, due to weaknesses in data
uted more widely and may avoid exacerbating collection (related to both definitions and cov-
the strains on institutional capacity that are erage).2 Reporting of “small” remittance trans-
often associated with natural-resource actions made through formal channels is not
booms). Furthermore, the “Dutch disease” mandatory in most countries,3 and remittances
effects of remittances are of relatively minor sent through post offices, exchange bureaus,
concern insofar as remittances grow gradually and other agents of money transfer operators
over long periods. Remittances have a large (MTOs) are often not reflected in official statis-
positive effect on national income in many de- tics (de Luna Martinez 2005). Third, flows
veloping countries, and there is compelling through informal channels (such as unregulated
evidence that they contribute significantly to money transfer firms or family and friends who
poverty reduction (see chapter 5). Although carry remittances) are rarely captured. Finally,
the evidence on the effect of remittances on remittances are often misclassified as export
long-term growth remains inconclusive, in revenue, tourism receipts, nonresident deposits,
economies where the financial system is under- or even foreign direct investment (FDI). Im-
developed, remittances appear to alleviate proving the quality of remittance statistics is the
credit constraints and may stimulate economic focus of ongoing cooperative international
growth. efforts (see box 4.1).

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Box 4.1 International working group on improving

data on remittances
A t its meeting in Sea Island in April 2004, the G-8
called upon the international financial institu-
tions (IFIs) to lead a global effort to improve remit-
resident households to or from nonresident house-
holds, and “net” compensation of employees from
persons working abroad for short periods of time
tance statistics. In January 2005, the World Bank (less than one year).
and International Monetary Fund (IMF) held an The TSG also recommended that institutional
international meeting of data users and compilers remittances—those involving government, cor-
who agreed that balance-of-payments statistics were porations, and nonprofit institutions serving
the appropriate framework for collecting, reporting, households—should also be reported as a new mem-
and improving official statistics on remittances; that orandum item in the standard presentation of
balance-of-payments concepts and definitions relat- balance-of-payments statistics. That item would lead
ing to remittances should be reviewed; and that to a further memorandum item, “total remittances,”
improved guidance for collecting and compiling the sum of personal and institutional remittances.
remittance statistics, including through the use of Because the concepts of personal transfers and
household surveys, was needed. Participants at the remittances are based on the concept of residence
international meeting also agreed that improvements rather than migration status, the TSG recommended
to relevant statistical concepts and definitions should that the concept of “migrant” be replaced by the
be discussed in a Technical Sub-Group on the Move- concept of “residence” in the balance-of payments-
ment of Persons (TSG), chaired by the UN Statistics framework. Reporting of bilateral remittance flows
Division with membership from central banks and is not currently required in the balance of payments,
national and international statistical agencies. but the recommendation of the TSG is that flows to
The TSG recommended that the “workers’ remit- and from major partner countries be identified.
tances” item in the balance of payments be replaced The TSG is expected to produce a final report in
by “personal transfers.” The new term would cover spring 2006.a
all current transfers in cash or in kind made or
received by resident households to or from other
nonresident households. It went on to recommend
aA concurrent effort to improve remittance statistics is
that a new aggregate, “personal remittances,” be
under way at the Center for Latin America Monetary Studies
reported in the standard balance-of-payments presen- (CEMLA) with support from the Multilateral Investment Fund
tation as a memorandum item. It was proposed that of the Inter-American Development Bank and technical advice
personal remittances comprise current and capital from an international advisory council that includes the IMF
transfers in cash or in kind, made or received, by and World Bank.

Officially recorded remittance flows using officially recorded data, rose to

are surging $167 billion in 2005, up 73 percent from
In this report (as in past editions of the World 2001 (table 4.1). More than half of that in-
Bank’s annual Global Development Finance crease occurred in China, India, and Mexico.
and the IMF’s 2005 World Economic Out- Low-income countries, led by India, registered
look), migrant remittances are calculated as an increase of $18 billion during this period
the sum of workers’ remittances, compensa- (box 4.2). Of 34 developing countries that re-
tion of employees, and migrant transfers (see ceived remittances in excess of $1 billion in
annex 4A.1). Thus defined, remittances re- 2004, 26 countries registered more than
ceived by developing countries, estimated 30 percent growth during 2001–4: Algeria

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 4.1 Workers’ remittances to developing countries, 1990–2005

$ billions
Change (%)
1990 1995 2000 2001 2002 2003 2004e 2005e 2005–2001

Developing countries 31.2 57.8 85.6 96.5 113.4 142.1 160.4 166.9 73
Lower middle income 13.9 30.0 42.6 47.4 57.3 72.5 83.5 88.0 86
Upper middle income 9.1 14.5 20.0 22.3 23.0 27.8 33.0 33.8 52
Low income 8.1 13.3 22.8 26.8 33.1 41.8 43.9 45 68
Latin America and the 5.8 13.4 20.1 24.4 28.1 34.8 40.7 42.4 74
South Asia 5.6 10.0 17.2 19.2 24.2 31.1 31.4 32.0 67
East Asia and the 3.3 9.7 16.7 20.1 27.2 35.8 40.9 43.1 114
Middle East and North 11.4 13.4 13.2 15.1 15.6 18.6 20.3 21.3 41
Europe and Central 3.2 8.1 13.4 13.0 13.3 15.1 19.4 19.9 53
Sub-Saharan Africa 1.9 3.2 4.9 4.7 5.2 6.8 7.7 8.1 72
World (developing & 68.6 101.6 131.5 147.1 166.2 200.2 225.8 232.3 58
Outward remittances from 6.1 12.5 12.1 14.3 18.7 20.2 24.1 – –
developing countries
Outward remittances from 11.2 16.6 15.4 15.1 15.9 14.8 13.6 – –
Saudi Arabia

Source: World Bank staff estimates based on IMF BoP Yearbook 2004 and country sources.
Note: Remittances are defined as the sum of workers’ remittances, compensation of employees, and migrant transfers (see
annex 4A.1). e  estimate.
– Data not available.

and Guatemala reported more than a tripling of Table 4.2 Recorded remittances have
remittance inflows; Brazil, China, Honduras, grown faster than private capital flows
Nigeria, Pakistan, and Serbia and Montenegro and ODA
$ billions
reported growth in the range of 101–170 per-
cent. (Also, five high-income countries— 1995 2004

Austria, Australia, Belgium, Germany, and

Workers’ remittances 58 160
Spain—reported 45–79 percent growth in Foreign direct investment 107 166
remittance inflows during 2001–4.) Private debt and portfolio equity 170 136
The growing importance of remittances as Official development assistance 59 79

a source of foreign exchange is reflected in the Source: World Bank (2005).

fact that remittance growth has outpaced
private capital flows and official develop-
ment assistance (ODA) over the last decade
(table 4.2). Recorded remittance receipts were Herzegovina, Cape Verde, Gaza, Haiti,
equivalent to about 6.7 percent of developing Jamaica, Kiribati, Lebanon, Nepal, Samoa,
countries’ imports and 7.5 percent of domes- Serbia and Montenegro, and Tonga. In an-
tic investment. They also were larger than of- other 28 countries, they were larger than the
ficial flows and private equity (non-FDI) flows earnings from the most important commodity
in 2004. Remittances were larger than public export; for example, in Mexico, remittances
and private capital inflows in 36 developing are larger than FDI; in Sri Lanka, they are
countries in 2004 and larger than total larger than tea exports; and in Morocco, they
merchandise exports in Albania, Bosnia and are larger than tourism receipts.

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Box 4.2 The recent surge in remittance flows to India

I ndia has reported a spectacular increase in remit-
tance inflows—from $13 billion in 2001 to more
than $20 billion in 2003 (see figure). Several factors
in India’s growing economy and (b) an easing of the
regulations and controls, more flexible exchange
rates, and gradual opening of the capital account.
account for this remarkable increase. First, the num- The elimination of the black-market premium on the
ber of migrants has grown sharply. During the oil rupee and convenient remittance services provided by
boom in the 1970s and 1980s, thousands of low- Indian and international banks have no doubt shifted
skilled Indian workers migrated to the Persian Gulf some remittance flows from informal hawala chan-
countries. In the 1990s, migration to Australia, nels to formal channels.
Canada, and the United States, increased signifi- Third, nonresident Indians have also responded
cantly, particularly among information technology to several attractive deposit schemes and bonds
(IT) workers on temporary work permits.a offered by the government of India. These offer
Second, the swelling of migrants’ ranks coincided attractive interest rates and an appreciating rupee.
with (a) better incentives to send and invest money While nonresident deposits are conceptually differ-
ent from remittances (they are a liability item in
the capital account), evidence suggests that a large
Remittance flows to India
part of such deposits is converted to local currency.
$ billions For example, for the Resurgent India Bond that
25 matured in 2003, most of the redemption value
stayed in India to meet various local currency
20 needs of the nonresident depositors and their fami-
lies. Nevertheless, remittances in the form of for-
eign-currency deposits can be speculative and may
reverse in the event of deterioration in the
investment sentiment.

aIn particular, migration to United States doubled during

the 1990s. Remittances from United States as a share of total
87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03
19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 remittances to India grew from 37 percent in 1997 to 51 per-
cent in 2003.

Figure 4.1 identifies the top 20 remittance High-income countries are the dominant
recipients in 2004. Among developing coun- source of global remittance flows (figure 4.2).
tries, China, India, Mexico, and the Philippines The United States was the largest source
were among the top recipients. Several indus- country with nearly $39 billion in outward
trial countries appear in this list as well. remittances in 2004. However, outward
When remittances are calculated in per remittances from developing countries
capita terms or as a share of GDP, a different amounted to $24 billion in the same year.4
picture emerges. The top 20 recipients in When expressed in terms of GDP shares, out-
shares of GDP are all developing countries; ward remittances play the largest role in the
all receive more than 10 percent of GDP as upper-middle-income developing countries
remittance flows (figure 4.1). Small countries (0.7 percent of GDP in these countries
(Bosnia and Herzegovina, Haiti, Lesotho, compared to 0.2–0.4 percent in other devel-
Moldova, and Tonga) are among the most de- oping countries and in high-income countries;
pendent on remittances. figure 4.2).

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 4.1 Top 20 remittance-recipient countries, 2004

Billions of dollars Share of GDP

India 21.7 Tonga 31.1

China 21.3 Moldova 27.1
Mexico 18.1 Lesotho 25.8
France 12.7 Haiti 24.8
Philippines 11.6 Bosnia and Herzegovina 22.5
Spain 6.9 Jordan 20.4
Belgium 6.8 Jamaica 17.4
Germany 6.5 Serbia and Montenegro 17.2
United Kingdom 6.4 EI Salvador 16.2
Morocco 4.2 Honduras 15.5
Serbia 4.1 Philippines 13.5
Pakistan 3.9 Dominican Rep. 13.2
Brazil 3.6 Lebanon 12.4
Bangladesh 3.4 Samoa 12.4
Egypt, Arab Rep. 3.3 Tajikistan 12.1
Portugal 3.2 Nicaragua 11.9
Vietnam 3.2 Albania 11.7
Colombia 3.2 Nepal 11.7
United States 3.0 Kiribati 11.3
Nigeria 2.8 Yemen, Rep. 10.0

0 5 10 15 20 25 0 15 30 45

Source: IMF BoP Yearbook, 2004, and World Bank staff estimates.

South–South remittance flows sending country. Comprehensive global data

are considerable are not available,5 but estimates are that in
Official data show that several developing poor countries of East Asia, South Asia, and
countries (China, Malaysia, and the Russian Sub-Saharan Africa, more than two-thirds of
Federation) are among the top 20 sources of emigrants migrate to a country in the same re-
remittances. Anecdotally, outward remittances gion. In South Asia and Sub-Saharan Africa,
from India and South Africa are also believed most migrate to another developing country.
to be large, although this is not reflected in the Second, even less is known about how bi-
official data (Genesis Analytics 2005). The lateral remittance patterns differ. We do not
World Bank (2005a) points out a strong asso- know, for example, how much, in total, is sent
ciation between remittance receipts and the from one country to another, or how remit-
length of the border shared with more pros- tance propensities differ across sending and
perous neighbors. Harrison and others (2004) receiving countries. But by making plausible
also report that most remittance flows occur assumptions about these flows (in particular,
within the same region. that bilateral remittances are a function of the
These factors all point to the conclusion stock of migrants in the sending country), it
that South–South remittance flows are sub- is possible to estimate bilateral remittance
stantial. But placing more precision on these flows and to calculate what proportion
flows is hard to do. First, relatively little is comes through South–South links.6 Using this
known about bilateral migration flows—that method, we estimate that nearly 30 percent of
is, about how many migrants (or what share) total remittance flows to developing countries
in each receiving country come from each originate in other developing countries. This

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Eastern Europe, Turkey, and North Africa;

Figure 4.2 Estimated remittance payment, and the Persian Gulf to South and Southeast
by country group, 2004
Billions of dollars

125.3 Informal remittances are large

Remittances transferred through informal
operators or hand carried by travelers are
unlikely to be captured in official statistics,
although they may represent a substantial
addition to remittances sent through official
15.4 channels. While it is extremely difficult to es-
3.0 5.7 timate the flows through informal channels,
Low-income Lower- Upper- High-income they appear to be large. First, the fact that
countries middle- middle- countries recorded remittances to several countries
income income
countries countries through formal channels doubled, tripled, or
quadrupled between 2001 and 2003 suggests
Share of GDP
that a significant part of the increase is likely
0.7 to reflect a shift from informal to formal chan-
nels in response to the tightened regulatory
scrutiny that has occurred since September 11,
0.4 2001.
Second, evidence from household surveys
suggests widespread use of informal remit-
tance channels (table 4.3).8 Household sur-
veys also help identify factors affecting the use
of remittance channels. In the presence of a
Low-income Lower- Upper- High-income well-developed formal sector, regular remitters
countries middle- middle- countries
income income and large remitters are unlikely to use the
countries countries
informal sector. Trust in the financial system is
an important prerequisite for a growing bank
presence in the (formal) remittance market.
estimate is consistent with the fact that nearly High remittance costs and the presence of
half of the migrant stock from the South mi- dual exchange rates are two key factors
grate to another country in the South.7 affecting the choice of informal remittance
One of the challenges of understanding
remittance flows is that their characteristics,
costs, and channels can vary widely from one Table 4.3 Choice of remittance channel
bilateral corridor to another (and also widely in selected countries
% remittances
from different locations within each country).
Understanding how remittance corridors Formal Informal

differ in the kinds of migrants they serve and

Dominican Republic 96 4
their means of transferring money is useful for Guatemala 95 5
providers of remittance services as well as pol- El Salvador 85 15
Armenia 62 38
icymakers (Hernandez-Coss 2004; Terry
Moldova 53 47
2005; and chapter 6 of this volume). Some of Bangladesh 46 54
the major remittance corridors are those that Uganda 20 80
connect Canada and the United States to Latin Source: World Bank staff estimates based on household
America and Asia; the European Union to surveys. See also Freund and Spatafora 2005.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

channels.9 If there were no cost advantages to official sector.10 They also show significant
using informal channels, there would be little regional variation. Informal remittances ap-
incentive to use them, and remittances could pear to be larger in Sub-Saharan Africa, the
arguably shift entirely to formal channels. Middle East and North Africa, and Europe
Thus if the costs of formal transfers were and Central Asia than in other regions.11
reduced to the range reported in the informal While the magnitude of the regional estimates
sector (2–5 percent), and if official and paral- varies across methods, the relative ranking of
lel exchange rates were unified, the resultant regional effects is more robust.
increase in recorded remittance flows could
be interpreted as an estimate of the size of
informal flows. Factors affecting remittance flows
Table 4.4 reports the results of an exercise
to estimate the size of the informal remittance
sector (see annex 4A.2 for a fuller explana-
T he surge in remittance flows over the past
few years reflects a mix of factors, as
noted. In some areas, there have been signifi-
tion). Cross-country regression analysis
cant reductions in remittance costs—60 per-
shows that reported remittances are lower,
cent in the United States–Mexico corridor
and informal flows higher, in corridors where
since 1999. On the measurement side, the size-
remittance costs are higher and where there
able depreciation of the dollar against most
are significant black-market premiums over
other major currencies (the euro in particular)
the official exchange rate. Using the estimated
since 2002 has increased the dollar value of
coefficients from these regressions, the
nondollar remittances over time.12 Improve-
predicted increase in officially recorded remit-
ments in data recording by central banks—in
tances is calculated in response to a 2–5 per-
response to growing recognition of the impor-
cent decline in remittance costs and elimina-
tance of remittances by national authorities,
tion of the exchange-rate premium. These
and as a result of broader efforts to improve
calculations suggest that the informal remit-
data quality—have generated sharp increases
tance sector is at least 50 percent of the
in remittance flows in some cases. In addition,
heightened security and scrutiny by immigra-
tion and finance authorities in many high-
Table 4.4 Estimated increase in formal income countries may have encouraged out-
remittances if transaction costs were ward surges in remittances, as undocumented
reduced to 2 to 5 percent and dual migrants responded to increased uncertainty
exchange rates were eliminated and risk of deportation or other legal action by
Percent remitting a larger share of their savings or
Cross-sectional Panel income. This factor has reportedly been im-
Region estimates estimates portant in Pakistan, which recorded a tripling
of remittance receipts from 2001 to 2003.
All developing countries 69 54
Sub-Saharan Africa 201 122
Eastern Europe and Central Asia 151 73
East Asia and the Pacific 56 ..
The surge in remittances is likely to
South Asia 25 55 continue in the medium term
Middle East and North Africa 165 .. In addition to these special factors, powerful
Latin America and the Caribbean 51 99
economic factors also influence the growth
Source: Freund and Spatafora 2005. of remittances. Increases in the number of
Note: Results averaged over 1995–2003. See annex 4A.2 for
a fuller explanation of the procedures used. In column 3,
migrants will have the greatest and most direct
a reduced form equation is estimated on the basis of the impact, of course, along with compositional
explanatory variables used in the cost regression reported in
table 4A.2.2.
features, such as the mix between temporary
.. Negligible. and permanent workers (temporary workers

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

are believed to remit a larger share of their Government policies can affect
income) and the skill mix (low-skilled workers remittance flows
tend to send a higher proportion of their lower Many sending and receiving governments
incomes). Employment opportunities in the are only now beginning to think about policies
host country affect income, and therefore to increase remittance flows and promote
remittances, while changes in the cost of living transfers through formal channels. In the
may affect the size of the surplus that remitters remittance-receiving countries, these policies
are able to send. include tax exemptions for remittance in-
The complex interplay of these factors come; improved access to banking services by
makes assessing the future growth potential of recipients; incentives to attract investments
remittance flows quite difficult. It is plausible by the diaspora; access to foreign exchange
that in the coming years, official remittance or lower duties on imports; support for the
flows will continue to rise at the 7–8 percent projects of migrant associations; and help for
annual rate seen during the 1990s. With both migrants in accessing financial systems. In the
the supply and demand for migrants grow- remittance-source countries, they include poli-
ing, migration flows—especially temporary cies affecting access to banks, access to foreign
migration—are likely to continue to be strong. exchange, support to migrant groups, types of
Growing income levels in source countries and immigration regimes, and cooperation with
rising costs of living in receiving countries, receiving countries.14
together with the falling costs of remittances,
would also imply larger remittances, espe- Policies in remittance-receiving countries
cially through recorded channels. Taxes on incoming remittances. Most
It is unlikely, however, that the surge in remittance-receiving countries today do not
remittance flows seen in some countries since impose taxes on incoming remittances. There
2001 will continue much longer. The shift in may be some implicit tax on remittances, how-
flows from informal to formal channels, to the ever, in the form of a general financial services
extent that it occurred in response to tightened tax15 or on remittances in kind (for example,
scrutiny, is likely to dwindle. (In Pakistan, food, clothing, electronic items, or vehicles).
for example, remittance flows have flattened When Vietnam removed its 5 percent tax on
since 2003.) In the more mature United remittances in 1997, it found that the flow of
States–Mexico corridor, where remittance remittances through formal channels in-
costs have already fallen drastically (by 60 per- creased. Such tax exemptions may well in-
cent since 1999), the effect of further cost crease remittance inflows,16 but they also raise
reduction will not be as large as it was five the possibility of misuse for tax evasion.
years ago. Travel and customs privileges for returns and
Some analysts argue that in the more mature imported goods. Many remittance-receiving
markets, “remittance decay” may set in, espe- countries give preferential treatment to
cially if temporary or undocumented workers migrants sending home or bringing with them
are allowed permanent and legal residence. goods and equipment. For example, once a
While it is true that the marginal propensity to year Tunisians are entitled to import goods
remit tends to decline with the length of a mi- and/or services up to a customs value of
grant’s stay in a host country, and ties with the TD1,000 without paying tax, and a private
home country weaken over time, there is no vehicle, home equipment, and furniture are
empirical evidence that the dollar amount of tax free when they return; Guatemala permits
remittances actually declines in these circum- a once-a-year tax-free remittance of any com-
stances.13 On the contrary, the effect of rising modity valued up to $500. Pakistan, Turkey,
incomes of the migrant sender may show up as Vietnam, and many other countries also offer
an increase in remittances over time. such import privileges.17

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Relaxation of exchange and capital controls. Bank of New Jersey. In Bangladesh the dra-
Unification of exchange rates and allowing matic increase in formal remittances since
more banks and financial institutions to un- 2001 is, in part, the result of the improved ser-
dertake foreign exchange transactions have vices of the banking sector (Siddiqui 2004).
been among the most successful ways of at-
tracting remittances to formal channels and ID cards for migrants. Providing identification
expanding remittance services in many coun- cards to migrants (regardless of their legal mi-
tries. Also, allowing residents to hold foreign gration status) to access banking facilities has
currency deposits using remittances from also opened up more opportunities for formal
abroad is believed to have resulted in a large remittance transfer. Mexican immigrants, for
increase in formal remittances in many coun- example, can obtain a photo-identification
tries in South Asia and Africa (Siddiqui 2004). card in the form of a matricula consular from
India’s liberalization of the exchange rate in the Mexican consulates abroad. This card is
1991 has been linked to a decrease in the use widely accepted by commercial banks in the
of illegal transfer channels to the state of United States to open bank accounts (and in
Kerala; and the Philippines found that by many states, for issuing driving licenses, see
abolishing exchange controls it quadrupled its box 6.1). Other Latin American governments
formal inward remittances in the same year are discussing similar arrangements for their
(Buencamino and Gorbunov 2002). Allowing nationals in the United States. Most sending
the market to decide exchange rates in 2002 countries require legal documentation for any
also helped the Bangladesh Bank to curb the bank transaction. Some receiving countries
informal hundi business significantly (Siddiqui issue ID cards to expedite domestic services
2004). In 2004, an increase in foreign cur- for their emigrants, for example, the Tunisian
rency reserves in Zimbabwe was ascribed, in carte consulaire for special customs clearance,
part, to the introduction of a new money reduced airfares, and foreign currency bank
transfer system (Homelink) set up by the gov- accounts in Tunisia.18
ernment to facilitate formal transfers. Support to hometown associations (HTAs)
and matching grants. Providing funds to
Allowing domestic banks to operate overseas. supplement or match collective remittances
Governments have allowed more of their made by emigrant groups is another means to
domestic financial institutions (including engage migrants in the development of home
microfinance institutions in some countries) to communities. With enhanced institutional
open branches and provide services to their capacities, HTAs could be valuable develop-
migrants working in other countries. These ment partners for governments, the private
domestic banks bring trust and offer remit- sector, and communities, but importantly as a
tance services at competitive prices. For exam- complement to, not a substitute for, strength-
ple, the Groupe Banques Populaires has ened financial and investment systems on the
picked up 66 percent of total remittances to ground (Gubert 2005). A careful evaluation of
Morocco by offering low fees, simple proce- support to HTAs through matching grant
dures, and other nonfinancial services to schemes and other means is yet to be under-
Moroccans abroad (Amin and Freund 2005). taken (see box 4.3).
Two small Armenian banks specializing in re-
mittance transfers, Anelik and Unibank, have Loans/pension schemes and bonds targeted at
come to dominate the formal transfer system the diasporas. These measures can expand
for Armenians in parts of Europe; and opportunities for investment and provide in-
Fonkoze in Haiti has expanded its U.S.–based centives for the formal transfer of money from
clientele in partnership with the City National abroad (see also chapter 6). While investments

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

in the form of nonresident deposits or dias- in the country. Such schemes were a major fac-
pora bonds are not, strictly speaking, remit- tor behind the doubling of remittance flows to
tances (because they involve the purchase of India between 2002 and 2003 (box 4.2).
assets, rather than transfers to households),
they may indirectly encourage remittances. Active policies and institutional arrangements
Many countries have successfully issued pre- to support the diaspora. Countries like
mium bonds to their diaspora (for Mexico and the Philippines with more suc-
Bangladesh, China, Eritrea, India, Israel, cessful remittance programs tend to have well
Lebanon, Pakistan and the Philippines, see established institutional frameworks to train,
Carling 2005). Even when investments in support, and ensure the welfare of their expa-
these bonds are in foreign currency terms, triates abroad. There is also a broad range of
after maturity some portion is likely to remain outreach activities to assist migrant welfare

Box 4.3 Collective remittances through hometown

associations and matching schemes
M any migrants are increasingly pooling their
resources and investing collectively in
development-related activities in home communities,
distribution of consumer goods (as in general stores
or grain banks) or the purchase of farming equip-
ment (Gubert 2005). In Latin America, it is observed
either through hometown associations (HTAs) or that when at least 30 percent of households in a
other migrant group schemes.a HTAs are the most town receive remittances, HTAs can help improve
prominent, because of their proliferation among the the quality of life of households (IOM 2005). But the
Latin American and Caribbean diaspora in Canada focus of HTAs is expanding to include more invest-
and the United States since the late 1990s. Similar ment in economic infrastructure and income-
associations exist in France (some 1,000 organiza- generating projects managed by the community and
tions de solidarite internationale issues de migrations local NGOs or banks (Orozco 2003).
or OSIMs), the United Kingdom, and Africa.b The Governments have, on occasion, offered
activities of HTAs are mixed and poorly docu- matching grants for remittances from diaspora
mented, but they range from diaspora support in groups or HTAs to attract funding for specific
the host country to community investment projects community projects.c The best known of these
in villages in the home country. matching schemes is Mexico’s 3-for-1 program,
Collective remittances via HTAs currently account started in 1997, under which the local, state, and
for only 1 percent of all remittances in Central federal governments all contribute $1 for every $1
America, but it is estimated that they could rise to of remittances sent to a community for a designated
3–5 percent in ten years if their management and development project. By 2002, the 3-for-1 program
institutional capacity improves (IFAD 2005). had established projects totaling $43.5 million, two-
Most HTAs tend to be small scale and philan- thirds of which benefited labor-intensive agricultural
thropic in orientation, and they invest in projects of economies in four high emigration states (IOM
no more than $10,000. They have traditionally 2005). In the period 2002–4, more than 3,000 such
focused on infrastructure and social projects projects benefited some 1 million inhabitants in
(schools, churches, recreational parks, medical out- 23 Mexican states.d
reach clinics, and household support) and on chan- Evidence from Mexico suggests, however, that
neling post-disaster humanitarian aid (for example, HTAs have not been very successful. But in some
in El Salvador). In Africa, there is evidence that the cases (for example, Zacatecas) where HTAs have
more sustainable projects tend to facilitate household exchanged or debated project ideas and investment

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Box 4.3 (continued)

climate issues with the local and state governments, projects, or from other regions with a greater need
they are believed to have been successful.e for assistance.
On the positive side, HTA involvement in projects
is argued to ensure that programs are focused on aHTAs are grassroots migrant organizations, usually formed
community needs. Resources have gone primarily to around the interests and needs of a mutual hometown. The
rural areas, where they have increased the supply of term has been coined in the United States, where many thou-
essential services (health, education, roads, and elec- sands of Latin American and Caribbean HTAs have sprung up
tricity). Donations by HTAs are often as much as or in the past 15 years or so (Orozco and Welle 2004).
bMigrant associations exist in many countries, but are
more than the municipal budget for public works,
mostly concerned with the conditions of the diaspora and net-
particularly in towns with small populations (Orozco
working abroad. Some, like the Sierra Leonean Women’s Forum
2003). HTAs can promote higher standards of trans- in the United Kingdom, are concerned with immediate survival
parency and accountability among local authorities, needs (food, clothing) back home (Black and others 2004).
and higher labor standards. cIn addition to Mexico, the Salvadoran government partners

There are obviously limitations on the potential with HTAs in rural development projects in El Salvador. In
for HTAs to serve as conduits for broader develop- 2001, the federation of HTAs (COMUNIDADES) and the Na-
tional Corporation of Municipalities created the Social Invest-
ment projects. They may not have the best informa-
ment for Local Development Fund (FISDL) to provide
tion about the needs of the local community, or they matching project funding. In France, the Osims can also
may have different priorities. The capacity of HTAs receive institutional and financial subsidies from the govern-
to scale up or form partnerships is limited by the fact ment for social and economic development projects back home
that their members are volunteers, and their (Magoni 2004).
dSee “3 por 1. Proyectos Compartidos,” prepared for the
fundraising ability finite. They can also become
divided and weaken their own advocacy potential seminar “Migracion, remesas y el Programa 3 por 1 para Mi-
grantes,” Secretaria de Desarollo Social, Mexico and IADB,
(Newland and Patrick 2004). When matching funds
Washington DC, June 2005.
come from fiscally constrained governments, there is eSee Gubert 2005, Iskander 2005, and Orozco 2004. The
also the problem that they may be diverted from literature is not clear on what “success” means in these cases
other—perhaps higher priority—development (beyond mere survival of the HTAs).

and promote remittances and investment in ments on their émigrés, but with little success.
the home country, from pre-migration infor- Also, restrictive emigration policies have dri-
mation and orientation (Philippines), IDs for ven migrants into using clandestine remittance
customs and other purposes (Colombia, channels.19
Tunisia), finance for study (Tunisia), support in
legal and administrative disputes (Morocco), Policies in remittance-source countries
fairs and re-orientation visits for émigrés and Only a handful of remittance-sending coun-
their families (Colombia, Tunisia), shortened tries have proactive remittance-supporting
military service (and payment of fee in lieu, policies. Most are noninterventionist or have
Turkey), hotline for migrant investors had little engagement to date, but this is
(Tunisia), and a diaspora trust fund (Nigeria). changing with the growing appreciation of the
Some countries like Bangladesh, Egypt, significance of remittances for development in
Eritrea, Pakistan, Philippines, and Thailand countries such as Australia, Canada, the
(and Mexico and Turkey in the 1960s) have United States, and most West European states
tried to impose mandatory remittance require- (Ellerman 2003, Carling 2005). USAID has

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

undertaken extensive research on remittances, migrants is likely to lead to larger remittances.

as has the United Kingdom’s Department for Also, as discussed above, the ties of migrants
International Development (DFID) and the to their home country weaken with the pas-
Norwegian International Peace Institute sage of time, causing remittances to decline.
(PRIO). All propose ways forward for more Given the personalized nature of remit-
proactive policies by sending countries—for tances, governments are unlikely to have much
example, to support migrant associations, fa- success in using remittance policies to steer
cilitate low cost, reduce bureaucratic remit- migration differently. Some countries, like
tance transfer, greater competition in the re- Canada, France, and Germany, have tried to
mittance market, and inform decision making direct remittance flows to investments in the
by migrants and affected communities. home country to encourage return migration,
but these efforts have met with little success.
Immigration policies. Policies that affect the There are also some examples of “forced” re-
size, type, and tenure of migration flows also mittance transfer programs between sending
affect remittance patterns. A larger migration and receiving countries, although these raise
stock would in general imply larger remittance vexing legal issues and do not appear to be ef-
flows to the country of origin. Given the mi- fective either in encouraging migrant return or
gration stock, a larger share of temporary mobilizing resources (box 4.4).

Box 4.4 Forced remittances

W hile it is generally assumed that migrant
workers are free to choose how much, when,
and to whom to send money, there have been cases
$8 million in forced savings had been paid out to ex-
braceros and only $6 million was unaccounted for;
but the LA Times reported (on March 30, 2001) that
when sending or receiving governments, banks in a total of $34 million in forced savings was collected
the home country, or employers have decided to re- during 1942–6. The loss of the money was explained
tain a certain proportion of pay for remittances. The by successive bank consolidations and restructuring,
rationale for such “forced” remittances is to ensure and as a result, records of accounts had disappeared
that temporary migrant workers do not stay on, but (Migration News,
return home after the end of their contract. Some- The braceros were mostly poorly educated peasants,
times, the objective of such measures is to steer the who did not even know about the deductions and
use of remittances to investment in the country of who later were intimidated by the forms and corre-
origin. spondence needed to claim their money (LA Times).
For example, from 1942 to 1964 the “Bracero op.cit.). In March 2001, a class action suit was filed
program” regulated migration of 4.6 million farm on behalf of former braceros at a San Francisco dis-
workers between Mexico and the United States. trict court claiming $30 million–$50 million in sav-
From 1942–9, a tenth of the wages earned by these ings not returned and additional punitive damages.
braceros was deducted from their pay by the U.S. This claim was rejected because of the statute of lim-
employers and paid into accounts held by the Bank itations (San Francisco Chronicle, August 29, 2002).
of Mexico at two commercial banks in San Fran- In 2003, the Mexican government agreed to reim-
cisco. From there it was transferred to the Bank of burse, within six months, an upfront sum of $150
Mexico and then on to the Banco de Credito Agri- per person and then monthly rates of pesos 200 for
cola. Alternatively, the employers gave the worker a up to pesos 60,000, provided the ex-braceros could
check for the deducted amount at the end of the con- produce identification (the Bracero Net program).
tract to be cashed back home in Mexico. A 1946 re- Forced remittances may also be used by a gov-
port by the Mexican government claimed that ernment to encourage the use of remittances for

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Box 4.4 (continued)

investment in the domestic economy. Under the the farmer-employer to pay directly into their savings
Deferred Pay Scheme, mine workers from Lesotho account via payroll deduction.
have 30 percent (initially 60 percent, until 1990) of Forced savings of this type raise legal issues in
their pay deposited at a Lesotho bank and the bal- that they violate an accepted principle of wage pro-
ance into a savings account at TEBA (The Employ- tection, that is, the idea that “wages shall be paid di-
ment Bureau of Africa) Bank. The certificate con- rectly to the worker concerned” (article 5 ILO Con-
firming the identity of the account holder is handed vention 95 of 1949). The only exception provided
out by the TEBA Bank at the end of the contract, for is that the “worker concerned has agreed to the
before the mine worker goes home to collect the contrary.” It is not clear whether that has been the
balance from the deferred pay scheme. case with the braceros or with the other examples
A similar arrangement is foreseen in the memo- cited here. Convention 95 states that “employers
randum of understanding between the Governments shall be prohibited from limiting in any manner the
of Thailand and Laos on employment cooperation. freedom of the worker to dispose of his/her wages.”
All Laotian guest workers are obliged to pay 15 per- Article 8.2 further spells out that “workers shall be
cent of their earnings into a “deportation fund” set informed of the conditions under which such
up by the host country, Thailand. Workers who wish deductions may be made.” (Mexico ratified this
to return home can claim their contribution in full convention in 1955. The United States has not
with interest. The request must be filed three months ratified it.)
before the return date, and the money is to be paid Forced remittances are also probably not the most
to the workers within 45 days after the last day of effective measure to ensure that temporary migrant
employment (articles 11 and 12). workers return home. If they return, it is likely not
A milder form of induced remittances has been driven by their desire to reclaim their savings. When
introduced for temporary Mexican farm workers in offered a choice, migrants avoid such systems. In
the United States and Canada. Before their departure South Africa a considerable number of mine workers
visas and work permits are issued, the temporary from Lesotho did not participate in the deferred pay
farm workers register with the Ministry of Labor in scheme, often in complicity with the mining compa-
Mexico. After the papers are delivered, migrants nies (Sparreboom 1996, p. 13). If the Lesotho de-
open a savings account with the subsidiary or an ferred pay scheme was voluntary, then the volume of
associated institution of a North American bank in savings would drop to a level of the voluntary
Mexico. Once they arrive in the United States or schemes of workers from Botswana and Swaziland,
Canada, the temporary workers either make the namely 1 percent of the levels of the obligatory
remittance transaction themselves or arrange with scheme (TEBA 1995).

Banking and financial markets. Greater relax- of Mexican immigrants in the United States
ation and competition in money transfer mar- and low-cost electronic transfers through the
kets leads to reduced prices and more money Federal Reserve Bank’s automated clearing-
reaching the beneficiaries. This process is fa- house system for Mexico (see chapter 6).
cilitated further by improving access of remit- Spain has initiated agreements between
tance service providers to national payment Spanish and Latin American financial institu-
and settlement systems. This seems to have tions to reduce transfer fees and foster the
worked well within framed agreements such entry of new agents into the financial market,
as the United States–Mexican Partnership for particularly in rural areas. In the past,
Prosperity program of 2001, involving the Germany worked closely with Turkey to en-
matricula consular to improve banking access courage remittances into formal channels

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

(UN 2005). In some remittance-source coun- High levels (or large increases) in remit-
tries, outward remittance flows are affected by tance flows can be expected to have direct
exchange controls. For example, South repercussions on foreign exchange rates, do-
Africa’s policy of limiting foreign exchange mestic interest rates, and the balance of pay-
dealings only to banks has prompted (un- ments, and indirect repercussions on macro-
banked) remitters to use informal channels— variables. Because of their relative stability
only 5 percent of remittances to other South- and targeting (directly to households), they
ern African Development Community (SADC) may bring some additional benefits. However,
countries are being sent via formal channels, as the experience with and analysis of natural
according to Genesis Analytics (2005).20 resource booms have shown, large inflows can
also have some undesirable side effects (see
ID arrangements for migrants. The U.S. facil-
also box 4.5). And to the extent that remit-
itation of banking for both regular and irreg-
tance flows may naturally just go to countries
ular migrants from Mexico through the ma-
that are doing poorly or respond anticyclically
tricula consular mechanism has been highly
(increase during downturns, due to a drought,
successful in drawing more migrants into
for example), it may be hard to disentangle
safer and cheaper remittance modes. The Fed-
how remittances affect macro-performance. In
eral Deposit Insurance Corporation (FDIC)
this section, we consider some of the macro-
through its New Alliance Task Force initiative,
economic channels through which remittances
in collaboration with the Mexican consulates
affect recipient countries.
and commercial banks, has been successful in
improving banking access as well as the finan-
Remittances are stable and
cial literacy of immigrants.21
may be countercyclical
Support to HTAs or migrant associations. Remittances may move countercyclically rela-
HTAs and similar entities receive some support tive to the economic cycle of the recipient
from host governments in the United States, country. Remittances may rise when the recip-
France, and parts of Africa in recognition of ient economy suffers a downturn in activity or
their development assistance potential. While macroeconomic shocks due to financial crisis,
HTAs could potentially play a useful role in natural disaster, or political conflict, because
community infrastructure and other collectively migrants may send more funds during hard
funded projects, their ability to effectively chan- times to help their families and friends. Re-
nel large amounts of aid remains untested.22 mittances may thus smooth consumption and
contribute to the stability of recipient eco-
nomies by compensating for foreign exchange
Macroeconomic effects losses due to macroeconomic shocks.
of remittances Many authors have observed an increase in

U ntil recently most of the discussion and

research on remittances was focused on
the (microeconomic) end use by the recipient
remittance inflows following a natural dis-
aster (Clarke and Wallsten 2004) or an eco-
nomic downturn (Kapur 2003). Yang (2004)
households, including the effects on poverty showed that remittance receipts by Filipino
(see chapter 5). But as outlined earlier in this households increased following the 1997 fi-
chapter, the large size of remittances relative to nancial crisis. A 10 percent appreciation of a
other external flows and to the GDP in many migrant’s currency against the Philippine peso
countries suggests that the macroeconomic led to increases in household remittance re-
effects of remittances may be of critical impor- ceipts and a 0.6 percentage point decline in the
tance in many countries (recall that the top 19 poverty rate in migrant households. He also
remittance recipients receive more than 10 per- found evidence of positive spillover effects on
cent of their GDP in remittances). households without migrant members due to

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Figure 4.3 Remittances as percent of private consumption, two years before and two years
after natural disasters
16 15.5

12 11.3 10.8 11.3
10 9.6
8.9 8.4 8.4
7.9 7.7
5.7 5.2 5.5 5.6
6 5.1
4.4 4.7
3.8 4.2
2 1 0 1 2 2 1 0 1 2 2 1 0 1 2 2 1 0 1 2
Bangladesh Dominican Republic Haiti Honduras

Note: The bar marked 0 is the year of the natural disaster.

increases in remittance-driven economic activ- within four years of a hurricane. In contrast,

ity as well as by direct transfers from the mi- increases in ODA and FDI make up for
grant’s origin household. Mishra (2005) finds roughly 26 and 21 percent, respectively,
that a 1 percent decrease in real GDP was as- within four years.
sociated with a 3 percent increase in remit-
tances after a two-year lag in 13 Caribbean Remittances can improve country
countries during 1980–2002. To the extent that creditworthiness
remittances are used for investment purposes, Remittances can improve a country’s credit-
however, they may behave procyclically just as worthiness and thereby enhance its access to
other investment flows do. In Turkey and the international capital markets. The ratio of
Philippines, remittances were more volatile and
procyclical in the 1990s than in the 1980s.23
Remittance flows (as a share of personal Figure 4.4 Remittances as a share of
consumption) continued to rise after natural personal consumption, two years before
disasters in Bangladesh, Dominican Republic, and two years after conflict
Haiti, and Honduras (figure 4.3). In Albania, Percent
after an initial disruption in remittance 20
inflows (as a share of personal consumption) 16.9 16.9 17.0

in the year of conflict, remittance flows recov- 15

12.7 12.3
ered quickly (figure 4.4). In Sierra Leone,
remittances increased in the year of the con- 10

flict.24 Remittances as a share of personal

consumption rose in response to the financial 5
2.9 3.2
crisis in Mexico in 1995 and in Indonesia and 0.7 1.0 0.8

Thailand in 1997 (figure 4.5). 0

2 1 0 1 2 2 1 0 1 2
Yang (2005) found that the increase in re- Albania Sierra Leone
mittances makes up for 13 percent of income Note: The bar marked 0 is the year of the conflict.
losses in the current year and 28 percent

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

credit ratings for Lebanon and Haiti by two

Figure 4.5 Remittances as a share of notches; these would result in implied sover-
personal consumption, two years before
eign spread reductions ranging from 130 to
and two years after financial crises
334 basis points (table 4.5).26
2.1 Remittance securitization can help
2.0 2.0 2.0
2 1.9
1.8 countries raise external financing
1.5 Another way in which remittances affect
1.1 1.1
international capital market access is through
1 the use of structured finance techniques.
0.5 0.5 Several banks in developing countries (for
instance, Brazil) have been able to raise rela-
0 tively cheap and long-term financing from
2 1 0 1 2 2 1 0 1 2 2 1 0 1 2
international capital markets via securitization
Mexico Indonesia Thailand
of future remittance flows.
Note: The bar marked 0 is the year of the financial crisis.
Remittance securitization typically involves
the borrowing entity (such as a bank) pledging
its future remittance receivables to an offshore
debt to exports of goods and services, a key special purpose vehicle (SPV). The SPV issues
indebtedness indicator, would increase signifi- the debt (figure 4.7). Designated correspon-
cantly if remittances were excluded from dent banks are directed to channel remittance
the denominator (figure 4.6). Country credit flows of the borrowing bank through an off-
ratings by major international rating agencies shore collection account managed by a
often fail to account for remittances.25 Model- trustee. The collection agent makes principal
based calculations using debt-to-export and interest payments to the investors and
ratios that include remittances in the denomi- sends excess collections to the borrowing
nator indicate that including remittances in bank. Since remittances do not enter the is-
creditworthiness assessments would improve suer’s home country, the rating agencies

Figure 4.6 Indebtedness classification including and excluding remittances, 2003

Debt as percent of exportsa
700 Excluding remittances


Including remittances
400 355

300 257 245

208 193 190
200 167 178
151 145
127 121 114 126 116 124
100 82

Lebanon Ecuador Pakistan Philippines Jamaica Morocco Jordan El Salvador Guatemala

Source: World Development Indicators, and World Bank staff calculations.

Note: a. Present value of external debt as percent of exports of goods and services, and remittances.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 4.5 Impact of remittances on country credit rating and sovereign spread
Remittances as % Rating excluding Rating including Spread saving
of GDP, 2004 remittances remittancesa (basis points)

Serbia and Montenegro 7 B BB 150

Lebanon 14 B B 130
Haitia 28 CCC B 334
Nicaraguaa 11 CCC B 209
Ugandaa 5 B B 161

Sources: Standard and Poors and World Bank staff calculations.

Note: a. Calculated using a model similar to Cantor and Packer (1995); see Ratha and De (2005).

Figure 4.7 Remittance securitization structure

Remittance senders

Remittance Issuing bank credits

payments beneficiary’s account
(foreign currency) in domestic currency

Correspondent banks Issuing bank

Remittance flows (foreign currency)

Excess cash Issuing bank’s

Trustee collateral
account at
account (foreign currency) correspondent bank

Debt service payment

Offshore Domestic
International investors

believe that the structure mitigates the usual ing makes these transactions attractive to a
sovereign transfer and convertibility risks. wider range of “buy-and-hold” investors (for
Such transactions also often resort to excess example, insurance companies) that face limi-
coverage to mitigate the risk of volatility and tations on buying sub-investment grade. As a
seasonality in remittances. result, the issuer can access international cap-
By mitigating currency convertibility risk, ital markets at a lower interest rate spread and
a key component of sovereign risk, the future longer maturity. Moreover, by establishing a
flow securitization structure allows securities credit history for the borrower, these deals
to be rated better than the sovereign credit rat- enhance the ability and reduce the costs of
ing. These securities are typically structured to accessing capital markets in the future.
obtain an investment grade rating. In the case The first major securitization deal in-
of El Salvador, for example, the remittance- volving international migrant remittances
backed securities were rated investment grade, occurred in 1994 in Mexico. The volume of
two to four notches above the sub-investment remittance securitization has grown rapidly
grade sovereign rating. Investment-grade rat- since then (figure 4.8a). Using this instrument,

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

versified payment rights. These bonds resulted

Figure 4.8 Securitization of remittances, in a spread saving of more than 700 basis
points compared to Brazil’s sovereign spread.
a. Millions of dollars As experience with this instrument broad-
2,837 ens, and investors become more comfortable
with its characteristics, it is possible that it
could be used by a wider range of countries (in-
1,6001,690 cluding poor countries) and for a broader range
1,1751,170 of external flows (remittances, tourism re-
ceipts, and commodity earnings). It is not easy
540 to estimate the potential size of such future-
65 115 flow securitization. But preliminary calcula-
tions, assuming an over-collateralization ratio











of 5:1 and using migrant remittance figures for












2003, show that developing countries could

b. By country
potentially issue nearly $9 billion and low-
El Salvador Kazakhstan
6% 1%
income countries could raise up to $3 billion
annually from international capital markets.
Several policy hurdles need to be crossed
Brazil 31% before securitization deals can proceed. High
Mexico fixed costs of legal, investment banking, and
24% credit-rating services and long lead times can
pose difficulties for developing countries with
Panama few large entities and high borrowing needs. A
1% Peru master trust arrangement can permit issuers to
structure a large deal but to tap the market in
several tranches. Pooling receivables of several
35% branches (or even several borrowers) could
Note: Applies to the first half of 2004 only.
also help increase the deal size to justify large
fixed costs. While the absence of an appro-
priate legal infrastructure can also constrain
issuance, this need not require an overhaul
Mexico, El Salvador, and Turkey raised about of the entire legal system. A more focused
$2.3 billion during 1994–2000. As electronic approach that concentrates on bankruptcy
transfers became more widespread, it was law may suffice, by making sure that pledged
easier to track complex transactions, and assets remain pledged in the event of default.
remittances securitization gave way to So far, only the top-rated (in local currency
securitization of diversified payment rights terms) financial institutions have issued future
(DPRs), including migrant remittances, but remittance-backed bonds in an effort to pierce
also payments related to exports and FDI. the sovereign foreign currency rating ceiling
During 2000–4, a total of $10.4 billion was (that is, to obtain a higher rating for these
raised through securitization of DPRs by bonds than the sovereign foreign currency rat-
Brazil ($5.3 billion), Turkey ($4.1 billion), El ing). The securitization transactions typically
Salvador, Kazakhstan, Mexico, and Peru (fig- do not affect financial institutions’ ability to
ure 4.8b). Following a sharp increase in bor- deliver remittances to the ultimate beneficia-
rowing costs in 2002 (in part because of ries. Loosely speaking, the financial institu-
election-year uncertainties), Brazil has raised tions that undertake a securitization transac-
over $4 billion by issuing bonds backed by di- tion are pledging their rights to foreign

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

currency, but not their obligations to deliver larger portion of government expenditures on
remittances (typically in local currency terms). infrastructure and also practicing more liberal
Potential issuers should be reminded, how- trade policies; both these measures would
ever, of significant risks—currency devalua- tend to increase exports and also contribute
tion and, in the case of flexible rate debt, un- to improved labor productivity and
expected increases in interest rates—that are competitiveness.
associated with market-based foreign currency A related concern is whether reliance on un-
debt. Moreover, securitized debt is inflexible earned income in the form of remittances has
debt. Securitization of remittances (and other adverse effects on the incentives to work, as
future flows) by public sector entities reduces well as on the quality of economic policies and
the government’s flexibility in managing its governance, similar to the well-documented
external payments and can conflict with the effects of windfall gains from natural re-
negative pledge provision included in multilat- sources such as oil. While oil exports are al-
eral agencies’ loan and guarantee agreements, most always found to have a strong negative
which prohibit the establishment of a priority impact on various governance indicators, such
for other debts over the multilateral debts. as control of corruption and rule of law, pre-
liminary cross-country analysis suggests that
Large remittance inflows can lead remittance flows may not have such negative
to exchange rate appreciation and effects (box 4.5).29
lower export competitiveness
Large and sustained remittance inflows can The evidence on the effect of remittances
cause an appreciation of the real exchange on long-term growth is inconclusive
rate and make the production of cost-sensitive To the extent that they finance education and
tradables, including cash crops and manufac- health and increase investment, remittances
turing less profitable. Although empirical evi- could have a positive effect on economic
dence on the adverse effect of large inflows of growth. Remittances may relieve credit con-
foreign exchange in terms of trade and growth straints in the recipient community and spur
is limited,27 it is plausible that this effect exists entrepreneurial activity (Funkhouser 1992,
and is significant for some small economies Yang 2004, Woodruff and Zenteno 2004).
where remittances are very high. Amuedo- Faini (2002) finds that the impact of remit-
Dorantes and Pozo (2004) found that a dou- tances on growth is positive. He argues that
bling of workers’ remittances resulted in real remittances overcome capital market imper-
exchange rate appreciation of about 22 per- fections and allow migrant households to ac-
cent in a panel of 13 LAC countries (see also cumulate positive assets, as claimed by Stark
Winters and Martins 2004). Rajan and and Lucas (1988) and Taylor (1994). Mishra
Subramanian (2005), however, did not find (2005) found that a 1 percentage point in-
any evidence that remittance flows slow down crease in remittance inflows in 13 Caribbean
growth by affecting competitiveness.28 More- countries increased private investment by
over, as remittances tend to be relatively stable 0.6 percentage point (all measured relative to
and persistent over long periods, the “Dutch GDP). To the extent that they increase
disease” effects of remittances are less of a consumption, remittances may increase per
concern than similar effects of natural re- capita income levels and reduce poverty and
source windfalls and other cyclical flows, and income inequality, even if they do not directly
the real exchange rate level achieved through impact growth (see chapter 5).
sensible policies may be sustainable (IMF On the other hand, large outflow of work-
2005). Governments in countries receiving ers, especially skilled workers, can reduce
large remittances can mitigate the effects of growth in labor-sending countries. Remit-
real exchange rate appreciation by allocating a tances may also indirectly affect labor supply,

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Box 4.5 Unlike oil windfalls, remittance inflows do

not weaken institutional capacity
T he economic performance of most mineral ex-
porters, in particular oil exporters, has been far
less impressive than that of resource-poor countries
inequality, which results in nepotism and a weak
civil society. Resource rents are also believed to be
associated with civil conflict (Collier and Hoeffler
(Gelb and others 1988; Auty 2001; Gelb, Eifert, and 2002).
Tallroth 2002). To a large extent this outcome seems In contrast, remittances are widely dispersed, the
driven by mismanagement of the economy and weak great bulk of them is allocated in small amounts, and
institutions. Sala-i-Martin and Subramanian (2003) for the most part, remittances avoid the government
show empirically that concentration of resource “middleman.” Hence the expectation is that they can
flows has deleterious effects on the institutional avoid the negative effects of natural resource wind-
framework and capacity of a country. Natural re- falls on poverty, growth, and institutional capacity.
source windfalls—oil rents, for example—often fos- This is similar to an argument by Birdsall and Subra-
ter weak institutions because they allow the authori- manian (2004) that countries would be better off if
ties to pursue arbitrary, costly, and inefficient policies they distributed the bulk of the returns from resource
(Ross 2001). States that control such resources may flows to the general population, who would use the
resist secular modernization pressures because they funds more effectively than a highly centralized gov-
create alternative sources of power (Isham and ernment, and also greatly reduce the incentives for
others 2003). These rents also perpetuate economic corruption.

by encouraging some remittance-recipient developed financial systems and sound institu-

households to choose more leisure than tions is likely to imply that a higher share of
labor. Chami, Fullenkamp, and Jahjah (2005) remittances is invested in physical and human
argue that remittances may slow down growth capital (IMF 2005). Indeed, Giuliano and
by reducing work efforts by remittance Ruiz-Arranz (2005) show that in the
recipients.30 economies where the financial system is un-
One recent study of the impact of remit- derdeveloped, remittances alleviate credit
tances on growth over an extended period constraints and work as a substitute for finan-
(1970–2003) for 101 developing countries cial development, improving the allocation of
found no significant link between remittances capital and therefore accelerating economic
and per capita output growth, or between growth. Recent research also shows that
remittances and other variables such as educa- remittances may promote financial develop-
tion or investment rates (IMF 2005). This ment (Aggarwal and others 2005), which in
study, however, attributed this inconclusive re- turn can enhance growth and reduce poverty
sult to measurement difficulties arising from (Beck and others 2004).
the fact that remittances may behave counter-
cyclically with respect to growth.31 Also, em-
pirically it is difficult to measure the effects of Annex 4A.1 World Bank data
remittances on human capital formation, on remittances
which may occur over a very long period of
Remittances, like aid, may be more effec-
U sing the definition in chapter 7 of Global
Development Finance 2003, migrant re-
mittances are considered the sum of workers’
tive in a good policy environment. For remittances, compensation of employees, and
instance, a good investment climate with well- migrants’ transfers. Data for these variables

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 4A.1.1 Countries with alternative 2004 (the latest year for which BoP data are
estimates in 2004 currently available). The gap-filling methods
$ millions followed, and the reasons for making the
Algeria 2,460 adjustments are documented below.
China 21,283 Workers’ remittances, as defined in the
Gambia 8
Iran 1,032 IMF Balance of Payments manual, published
Kenya 464 in 1993 (fifth edition), are current private
Lebanon 2,700 transfers from migrant workers who are con-
Malaysia 987
Mauritious 215 sidered residents of the host country to recipi-
Nigeria 2,751 ents in their country of origin. If the migrants
Serbia and Montenegro 4,129 live in the host country for a year or longer,
Vietnam 3,200
they are considered residents, regardless of
Total $39,259 their immigration status. If the migrants have
lived in the host country for less than a year,
their entire income in the host country should
be classified as compensation of employees.
are taken mostly from the balance of payments Workers’ remittances are transfers, whereas
(BoP) data file of the IMF (see also Ratha compensation of employees is considered fac-
2003). However, many countries do not report tor income. In the earlier, fourth edition of the
data on remittances in the IMF BoP statistics, BoP manual, compensation of employees was
even though it is known that emigration from called labor income and was classified as non-
those countries took place (see table 4A.1.1 for factor services (referred to just as services in
a list of these countries). In 2003 about 87 the fifth edition).
countries did not report any remittances’ data. Although the residence guideline in the
Further, there was no consistency in reporting manual is clear, this rule is often not followed
the data. For example, only 28 countries re- for various reasons. Many countries compile
port workers’ remittances, compensation of data based on the citizenship of the migrant
employees, and migrants’ transfers. Forty-five worker rather than on their residency status.
countries report both workers’ remittances Further, data are shown entirely as either
and compensation of employees; 11 countries compensation of employees or as worker re-
report compensation of employees and mi- mittances, although they should be split be-
grants’ transfers; and 3 countries report work- tween the two categories if the guidelines
ers’ remittances and migrants’ transfers. There were correctly followed; for example, Saudi
are 14 countries that report only workers’ re- Arabia and Israel record only compensation
mittances and 19 countries that report only of employees. India shows very little com-
compensation of employees. pensation of employees, but large workers’
Reported data for developing countries remittances, although it is well known that
show only $113.4 billion in total remittances India supplies a large number of temporary
for the year 2003 (workers’ remittances IT workers to the United States and Euro-
$97.3 billion, compensation of employees pean countries. On the other hand, the
$14.8 billion, and migrants’ transfers $1.3 bil- Philippines shows large compensation of
lion), and 83.8 billion in 2004 (workers’ re- employees and very few migrants’ transfers.
mittances $68.7 billion, compensation of em- The distinction between these two categories
ployees $13.5 billion, and migrants’ transfers appears to be entirely arbitrary, depending
$1.5 billion). By filling in gaps for some devel- on country preference, convenience, and tax
oping countries for which remittance data laws or data availability. This fact has been
were missing, we arrived at an estimate of recognized at the World Bank since the
$142 billion in 2003, and $160 billion in 1980s, and worker remittances have been

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

treated as part of labor income and added to from the country desks are used for The Gam-
exports of goods and services in calculating bia, Iran, and Serbia and Montenegro; and
debt service ratios. data from central banks were used for
Though small in comparison to compensa- Lebanon and Vietnam. Some high-income
tion of employees and workers’ remittances, countries (notably Canada, Singapore, United
migrants’ transfers have become another Arab Emirates) also do not report remittance
source of confusion. Migrants’ transfers are data.
the net worth of migrants that are transferred
from one country to another at the time of Classification under other categories
migration (for a period of at least one year). Due to the difficulty in classifications, coun-
Migrants’ transfers are considered capital tries have often classified workers’ remittances
transfers in the BoP fifth edition manual, al- either as other current transfers or as transfers
though they were considered current private from other sectors. For example, in the case of
transfers in the fourth edition. As the number Haiti, before 1989 and after 1997, data were
of temporary workers increases, the impor- recorded as workers’ remittances, but during
tance of migrants’ transfers may increase. 1990–7, they were recorded as transfers from
Therefore, in order to get a complete picture other sectors. Kenya and Malaysia data have
of the resource flow, one has to consider these similar difficulties. For these countries, data
three items together. under “other sectors” from the IMF are
There are four main reasons for gaps in treated as worker remittances. In China a
remittance data: vintage, missing data, data large proportion of workers’ remittances are
recorded under other than the three categories classified as other private transfers in the IMF
mentioned above, and data collection practices. BoP file. Therefore, instead of the IMF’s work-
ers’ remittances, we have used workers’ remit-
Vintage tances data from the country desk. It is not
The Balance of Payments Yearbook publishes just the developing countries that follow this
data with a one-year lag. That is, the yearbook practice, many high-income OECD countries
published in December of the current year (for example, the United Kingdom) do the
should have data up to December of the pre- same.
vious year. However, this is not true for a There are also other problems in the data,
number of developing countries, for which the such as the difficulty in separating travel ex-
latest data available are two or even more penditure from remittances, which have not
than three years old. For about 28 countries in been addressed here. The increased acceptance
2003, and 59 countries in 2004, data have of credit and debit cards in developing coun-
been obtained from World Bank country desks tries further complicates the issue. In some
or extrapolated on the basis of earlier trends. countries, notably China, remittances may
In addition, two countries, Algeria and have been misclassified as FDI. The OECD
Nigeria, have not reported data to the IMF for definition of FDI (including the purchase of
a number of years. For Algeria the IMF data holiday or second homes by nonresidents)
stop in 1991, and for Nigeria data, stop in may be counted as FDI—a likely case in
1999. However, data for these countries are China. In the case of India and many other
available from the country and reported in the countries, remittances may have been classi-
country databases of the World Bank and fied as nonresident deposits, especially those
IMF. in local currency terms.

Missing data Data collection practices

Several developing countries (for example, A survey of central banks, based on responses
Lebanon) do not report to the IMF.32 Data from 40 central banks, reveals widespread

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

problems with remittance data collection estimates using a set of variables that are noted
methodology (de Luna Martinez 2005). Most in the literature to affect the choice of the re-
of the central banks use remittance data re- mittance channel. Empirically, this involves
ported by commercial banks, but leave out first estimating officially recorded remittances
flows through money transfer operators and as a function of fee, exchange commission, and
informal personal channels. the presence of a dual exchange rate (and other
Even when data are available and properly variables shown in the equation below). Next,
classified, in many cases they are not based on using the estimated coefficients on these
actual exchange records. In a number of cases, variables, we predict what remittances would
the preferred methodology of estimating the be if the values of these variables become
workers’ remittances is based on taking the closer to those prevailing in regions where
number of emigrants, and multiplying by an informal flows are small. We then interpret the
average amount sent. The sources for these difference between these predicted remittances
data are migration records, surveys of and the actual remittances as an estimate of in-
exchange and financial houses, and household formal flows.
surveys. However, these data are often weak For this purpose, we propose the following
or out of date. Also the methodology for model of remittances:
preparing estimates is not the same in all coun-
REMIT  ␤0  ␤1Host  ␤2Home
tries, and it is not always described in the
country notes in the publicly available balance-  ␤3 Migrant  ␤4Fee200
of-payments data. It is hoped that the increased
 ␤5Spread200  ␤6 Dual
awareness about the importance of remittances
and the shortcomings in both the remittance where REMIT is the log of remittances (or
and migrant workers’ data will result in efforts remittance per migrant or per capita); Host is
to improve the data transmission. the log of the host-country per capita output
Table 4A.1.1 shows the countries where we (trade or migration weighted across hosts);
have used alternative estimates of workers’ re- Home is the log of home-country per-capita
mittances’ using either country desk or the output; Migrant is the log of the stock of
central bank data. migrant workers in OECD countries; Fee200
Perhaps the most difficult aspect of remit- is the fixed fee for sending $200 from the
tance data is estimating informal flows. In United States to the source country; Spread200
annex 4A.2, we discuss different ways of esti- is the exchange commission for sending $200;
mating informal flows. One way to estimate and Dual is a dummy variable for dual ex-
the true size of remittances is to undertake change rates. The last three variables are likely
surveys of remittance senders and recipients. to have large impacts on the extent to which
Unless new, adequately randomized and rep- money is sent via formal channels. The data on
resentative surveys of recipients and senders remittances are available on a panel basis; data
are carried out, evidence from existing house- on transaction costs and on the number of mi-
hold surveys would only be indicative rather grant workers are only available for a cross-
than comprehensive. section. In table 4A.2.1 below, we report re-
gression results for a cross-section of countries
(using average figures for 1995–2003 for re-
Annex 4A.2 A model-based
mittances and other time series variables). In
estimation of informal remittance table 4A.2.2, we show results of remittance
flows cost functions estimated using cross-country

E stimating the size of unrecorded flows is

almost impossible. In what follows, we
make an effort to arrive at some crude
data. These equations estimate panel data on
remittance costs for use in panel data regres-
sions reported in table 4A.2.3. Reduced form

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Table 4A.2.1 Regression results: determinants of worker remittances

Dependent variable: Dependent variable: Dependent
Dependent variable: Ln (Remittances) Ln (Remittances per variable:
Ln (Remittances) IVa emigrant) Ln (Remittance
Explanatory variables (1) (2) (3) per capita) (4)

Dual exchange rate 0.42 0.18 0.31 0.74**

(1.32) (0.46) (0.90) (2.04)
Service fee 0.06** 0.12* 0.07** 0.04
(2.32) (1.94) (2.47) (1.00)
Exchange-rate spread 0.04 0.02 0.05
(0.50) (0.26) (0.52)
Stock of migrant 0.73** 0.64 0.22**
workers (7.66) (5.41) (2.45)
Main host per capita 0.10 0.05 0.22* 0.11
income (0.77) (0.31) (1.75) (0.69)
Home per capita 0.15 0.17 0.06 0.72**
income (1.03) (1.00) (0.40) (4.35)
Income 0.31** 0.31
(4.05) (3.75)
Number of 104 85 104 104
R2 0.70 0.69 0.08 0.35

Source: Freund and Spatafora (2005).

Note: Robust t-statistics appear in parentheses.
a. Instruments include financial development and dollarization. Hansen’s J-statistic is 2.49 (p-value 0.12), and the Shea partial
R-squared of the instruments is 0.37.
**significant at the 5 percent level; *significant at the 10 percent level.

Table 4A.2.2 Regression results: determinants of transaction costs

Dependent variable: remittance cost

Explanatory variables (1) (2) (3) (4) (5)

Bank concentration 0.05** 0.03 0.03

(2.29) (1.10) (1.35)
Financial development 0.05** 0.05** 0.05** 0.06** 0.05**
(2.41) (2.38) (2.42) (2.54) (2.53)
Financial risk 0.04 0.04 0.03 0.04 0.01
(0.32) (0.26) (0.24) (0.31) (0.02)
Dollarization 4.12** 3.92** 3.92** 4.10** 4.00**
(4.47) (3.87) (4.20) (4.14) (4.33)
Domestic output 0.22 0.56 0.32 0.75 0.46
(0.46) (1.02) (0.66) (1.47) (0.98)
Remittances 0.30 0.42*
(1.18) (1.85)
Emigrant stock 0.34 0.59**
(1.18) (2.40)
R2 0.52 0.56 0.53 0.52 0.52
Number of observations 76 69 76 70 78

Source: Freund and Spatafora (2005).

Note: The dependent variable is the percentage cost of remitting $200. Robust t-statistics appear in parentheses.
**significant at the 5 percent level; *significant at the 10 percent level.

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 4A.2.3 Panel regression results: determinants of remittances

Ln (Remittances Ln (Remittances
Ln (Remittances) Ln (Remittances) per capita) per capita)
Explanatory variables (1) (2) (3) (4)

Dual exchange rates 0.28** 0.22** 0.26** 0.21**

(2.69) (2.18) (2.48) (2.03)
Fitted cost 0.08* 0.08*
(1.70) (1.74)
Bank concentration 0.00 0.00
(1.00) (1.03)
Financial development 0.03** 0.02**
(5.13) (4.30)
Financial risk 0.00 0.01
(0.02) (1.12)
Dollarization 0.62* 0.58
(1.73) (1.60)
Net errors and omissions 0.02** 0.02** 0.02** 0.02*
(2.07) (2.03) (2.12) (1.97)
Home per capita income 2.10* 3.64** 0.43** 0.63**
(1.71) (2.92) (2.37) (3.31)
Host per capita income 1.37 1.28 1.33 1.41
(1.39) (1.34) (1.33) (1.46)
Home income 1.71 3.13**
(1.38) (2.46)
Number of observations 295 295 295 295
R2 0.27 0.35 0.21 0.29

Source: Freund and Spatafora (2005).

Note: All regressions include country and year fixed effects. Robust t-statistics appear in parentheses.
**significant at 5 percent level; *significant at 10 percent level.

equations for remittances using statistically 2. One market study estimates that global remit-
tances are about 2.5 times the size reported in the IMF
significant variables in the cost regressions are
balance of payments (Aite Group 2005). The recent
also shown in table 4A.2.3. upward revision of China’s remittances to $21 billion
in 2004, from an earlier estimate of $4.6 billion, lends
Notes some support to this notion, although there is no
1. In March 2004, the G-7 finance ministers indi- strong evidence that systematic misreporting is so
cated their intention to “continue to work on initia- large. The discrepancy for China is reportedly due to
tives to reduce barriers that raise the cost of sending re- the fact that the Chinese figures include compensation
mittances and to integrate remittance services in the only for state employees. Some authors believe that a
formal financial sector” and their commitment to portion of China’s FDI attributed to overseas Chinese
“work with governments, the private sector, and mul- may actually be a misclassification of migrant remit-
tilateral development banks to broaden the access for tances. Some also believe that the recent surge in re-
families and entrepreneurs to financial services.” At the mittances to China in part reflected speculative inflows
Sea Island Summit in June 2004, the G-8 heads of state in anticipation of a revaluation of the yuan.
called for “better coherence and coordination of inter- 3. For example, the reporting threshold (typically
national organizations working to enhance remittance per person per day) is $10,000 in the United States,
services and heighten the developmental impact of re- 12,500 euros in western European countries (on aver-
mittance receipts.” They indicated that “G-8 countries age), and 3 million yen in Japan.
will work with the World Bank, IMF, and other bodies 4. Saudi Arabia, the second largest source of re-
to improve data on remittance flows and to develop mittances at $13.6 billion (or 5.4 percent of GDP) in
standards for data collection in both sending and re- 2004, is now classified as a high-income country. Saudi
ceiving countries.” Arabia’s per capita income level has risen in response

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

to the current high oil prices. The authorities prefer 12. Between 2001 and 2004, the euro appreciated
that Saudia Arabia be treated as a developing country. by 28 percent relative to the U.S. dollar. During this
5. Efforts are under way through the GTAP con- period, outward remittances from France and
sortium to compile and estimate a comprehensive set of Germany actually declined by 5 percent in euro
bilateral migrant stocks, which are used here. See terms. Remittances from Italy and Spain increased
Walmsley, Ahmed, and Parsons 2005. nearly 40 percent in euro terms and 93 percent in
6. More precisely, bilateral remittance flows are U.S. dollar terms.
calculated by allocating reported remittance inflows in 13. A survey of Congolese, Senegalese, and Nigerian
each country according to weights constructed as diasporas in Belgium did not reveal any significant re-
follows: lationship between the propensity to remit and the
R(i,j)  [Remittance flows to country j] * number of years a migrant has lived in Belgium. On the
[M(j,i)/[sum over i of the nominator]] contrary, several migrants who had been in Belgium
R(i,j)  remittance flows from country i to country for more than two decades continued to send signifi-
j and M(j,i)  stock of migrants from country j in cant amounts of remittances. Evidence from the Pacific
country i. Data on migrant stocks M(.) are taken from Islands also do not support remittance decay (Connell
the GTAP database (Walmsley, Ahmed, and Parsons and Brown 2005; Simati and Gibson 2001). Grieco
2005). (2003), however, reported evidence of remittance
7. Including Saudi Arabia as a developing country decay in the case of Micronesian migrants in Guam
would raise South–South remittances to 45 percent and and Hawaii, caused by family reunification or death of
South–South migration stock to 60 percent. the beneficiaries.
8. A World Bank survey of the African diaspora in 14. The information presented here derives from a
Belgium conducted in spring 2005 revealed that survey of IOM country missions on the policies of their
42 percent of remittances from Belgium to Senegal, host countries, as well as studies by ADB, ECOSOC,
and 55 percent to Congo and Nigeria, go through in- USAID, DFID, and the World Bank.
formal channels. Anecodotal evidence suggests that 15. For example, Colombia has a 0.4 percent tax
nearly 70 percent of remittances in the France–Mali on transactions through money exchange bureaux and
corridor take place through informal channels. Hand- banks, a temporary arrangement in effect until 2007.
carriage is a popular yet informal channel of remit- Belarus also taxes remittances from nonimmediate
tances in many countries. In the Philippines, 40 percent family members.
of total flows are estimated to be remittances brought 16. For example, in Tajikistan the removal of the
home by migrants in person. Nearly 42 percent of out- state tax on cross-border bank transactions in 2003
ward remittances from South Africa are believed to reportedly helped raise remittances from $78 million in
move through informal channels (Genesis Analytics 2002 to $256 million in 2003 (Olimova and Bosc
2005). 2003).
9. In a calibration model, El Qorchi, Maimbo, and 17. For example, nonresident Pakistanis remitting
Wilson (2003) argue that the black-market premium is over $10,000 through banking channels can import
the key factor determining informal flows. Other fac- any personal item valued up to $1,200 duty-free per
tors affecting the choice of the channel are trust in the annum (World Bank 2005b).
intermediary and anonymity and convenience factors, 18. Extension of voting franchise to migrants over-
such as location, hours of operation, and language. seas and other policies of political inclusion may also
10. Results will underestimate the size of informal catalyze remittances and other financial flows to the
flows to the extent that they are affected by other fac- country of origin (Carey 2003; Yang 2003).
tors, such as a lack of legal documentation of migrants 19. For example, Pakistan does not permit women
and high tax rates. To the extent that there would still under 35 to emigrate as domestic workers and Vietnam
be some informal flows even at this lower remittance bans females from working overseas in the entertain-
cost level, the estimates are actually lower bounds on ment sector. Bangladesh recently abandoned similar re-
the true size of informal remittances. However, it is strictions recognizing that although such restrictions
possible that the increases estimated in table 4.4 repre- may protect migrants from exploitation, they may also
sent new remittance flows, and not just the shift from encourage more irregular migration, rendering them
the informal to the formal sector, in which case these even more vulnerable.
estimates would overstate informality. 20. This is in part responsible for Western Union’s
11. Page and Plaza (2005) use a similar methodol- withdrawal from the South African market (Genesis
ogy and find that the share of unrecorded remittances Analytics 2005).
relative to the total remittances averages 48 percent 21. Such activities complement government objec-
worldwide (and 73 percent in Sub-Saharan Africa). tives to improve banking access in poor neighborhoods

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

(e.g., through the Community Reinvestment Act; see typical developing country with a large pool of
Frias 2004). unemployed.
22. USAID also established an 18-month pilot pro- 31. It is difficult to disentangle the reverse-causality
gram in 2004 with the PanAmerican Development problem (that growth also affects remittances) while
Foundation to strengthen the capacity of U.S.-based measuring the effect of remittances on growth. Some
HTAs. The UK government is also looking at the pos- researchers argue that the empirical results showing
sibility of using HTAs as a conduit for development a negative association between remittances and growth
aid. may largely reflect the fact that remittances tend to
23. See also Global Development Finance 2003, rise when growth is weak in the remittance-recipient
chapter 7, and Sayan (2004). A separate study by country.
Sayan (2005) finds that in a sample of 12 low-income 32. The list of countries that do not report remit-
and lower-middle-income countries during 1976–2003, tance data also includes the following 29 countries:
real remittances responded to a fall in real GDP with a Afghanistan, Angola, Bahamas, Bahrain, Bhutan,
one-year lag. He also found evidence of countercycli- Burundi, Canada, Central African Republic, Chad,
cality due to consumption smoothing in India and Congo Democratic Republic, Equatorial Guinea, Iraq,
Bangladesh and procyclicality due to a stronger invest- Kuwait, Liberia, Singapore, Somalia, Taiwan (China),
ment motive in Jordan and Morocco. Turkmenistan, United Arab Emirates, Uzbekistan,
24. Black (2004, p. 12) reports that remittances re- Zambia, and Zimbabwe.
mained substantial during the civil war in Côte
25. This is likely to be the case in countries (such References
as the Philippines or Lebanon) where the headline Aggarwal, Reena, Asli Demirguc-Kunt, and Maria
worker remittance variable has underestimated or Soledad Martinez Peria. 2005. “Do Remittances
missing data. Promote Financial Development? Evidence from
26. Sovereign spread rises exponentially as credit a Panel of Developing Countries.” Unpublished
ratings worsen along the rating scale. A one-notch im- paper, World Bank, Washington, DC.
provement in credit ratings, therefore, results in higher Aite Group. 2005. “Consumer Money Transfers: Pow-
spread saving for countries at the bottom of the rating ering Global Remittances.” Unpublished paper.
scale. January.
27. See McMahon (1997) for a review of empirical Alesina, Alberto, Arnaud Devleeschauwer, William
studies on the so-called Dutch disease, a term coined by Easterly, Sergio Kurlat, and Romain Wacziarg.
The Economist in 1977. 2003. Fractionalization. Journal of Economic
28. They argue that migrants may lose interest in Growth 8(2): 155–94.
remitting money and prefer to send goods instead, if Amin, Mohammad, and Caroline Freund. 2005. “Mi-
the currency in the remittance recipient country is over- gration and Remittances in ESA Countries.”
valued. Thus controlling overvaluation through Washington, DC: The World Bank.
prudent macroeconomic policies can help attract Amuedo-Dorantes, Catalina, and Susan Pozo. 2004.
remittances. “Workers’ Remittances and the Real Exchange
29. Note that this result applies to cross-country Rate: A Paradox of Gifts.” World Development
comparison. It would be extremely difficult to empiri- 32(8): 1407–17.
cally estimate the effect of remittances on institutional Annual Finance & Accounting International
capacity over time in a given country, since institu- Conference—Managing Securitization for
tional changes take place over a very long time. Also Lebanon and the MENA Region, December 3-4,
such an exercise would require controlling for reverse 2004. Lebanese American University, School of
causality: remittances may respond to cyclical or Business, Beirut.
abrupt changes in economic growth and governance. A Auty, Richard. 2001. “Introduction and Overview.” In
priori, the effect of institutions on remittances can run Resource Abundance and Economic Develop-
either way: On the one hand, better institutional ment, ed. R. M. Auy. New York: Oxford Univer-
capacity may attract remittances meant for investment sity Press.
purposes. On the other hand, better institutional Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine.
capacity (if they also mean better performance) may 2004. “Finance, Inequality and Poverty: Cross-
mean less emigration and dependence on remittances. Country Evidence.” NBER Working Paper
30. However, reduced work effort by some individ- 10979. National Bureau of Economic Research,
uals may not reduce the aggregate work effort in a Cambridge, MA.

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Birdsall, Nancy, and Arvind Subramanian. 2004. “Sav- the Informal Hawala System.” IMF Occasional
ing Iraq from Its Oil.” Foreign Affairs (July/ Paper 222. International Monetary Fund,
August): 77–89. Washington, DC.
Black, Richard, Savina Ammassari, Shannon Mouil- Ellerman, David. 2003. “Policy Research on Migration
lesseaux, and Radha Rajkotia. 2004. “Migration and Development.” World Bank Policy Research
and Pro-Poor Policy in West Africa.” Working Working Paper 3117. Washington, DC.
Paper C8. Sussex Centre for Migration Research, Faini, Ricardo. 2002. “Migration, Remittances, and
University of Sussex. November. Growth.” Unpublished paper. University of Bres-
Buencamino, Leonidas, and Sergei Gorbunov. 2002. cia.
“Informal Money Transfer Systems: Opportuni- 2002-3/ conference%20papers/faini.pdf.
ties and Challenges for Development Finance.” Freund, Caroline, and Nikola Spatafora. 2005. “Re-
DESA Discussion Paper 26. mittances: Costs, Determinants, and Informality.”
Cantor, Richard, and Frank Packer. 1995. “Sovereign Background paper prepared for this report.
Credit Ratings.” Current Issues in Economics and World Bank.
Finance 1(3): 1–6. June. Frias, Michael. 2004. “Linking International Remit-
Carey, John M. 2003. “Political Institutions in El tance Flows to Financial Services: Tapping the
Salvador: Proposals for Reform to Improve Latino Immigrant Market.” Supervisory Insights.
Elections, Transparency, and Accountability.” FDIC. Winter 2004.
Unpublished paper. Dartmouth College. Funkhouser, E. 1992. “Migration from Nicaragua: percent20sal- Some Recent Evidence.” World Development
vador.pdf. 20(8): 1209–18.
Carling, Jorgen. 2005. “Migrant Remittances and Devel- Gelb, Alan, and others. 1988. Oil Windfalls: Blessings
opment Cooperation.” PRIO Report 1/2005, Oslo. or Curse? New York: Oxford University Press.
Chami, Ralph, Connel Fullenkamp, and Samir Jahjah. Gelb, Alan, Benn Eifert, and Nils Borje Tallroth. 2002.
2005. “Are Immigrant Remittance Flows a “The Political Economy of Fiscal Policy and Eco-
Source of Capital for Development?” IMF Staff nomic Management in Oil Exporting Countries.”
Papers 52(1). World Bank Policy Research Working Paper
Clarke, George, and Scott Wallsten. 2004. “Do Remit- 2899. Washington, DC.
tances Protect Households in Developing Coun- Genesis Analytics. 2005. “Supporting Remittances in
tries against Shocks? Evidence from a Natural Southern Africa: Estimating Market Potential and
Disaster in Jamaica.” Unpublished paper. World Assessing Regulatory Obstacles.” Johannesburg,
Bank, Washington, DC. South Africa.
Collier, Paul, and Anke Hoeffler. 2002. “Greed and Giuliano, Paola, and Marta Ruiz-Arranz. 2005. “Re-
Grievance in Civil War.” In Annual Bank Confer- mittances, Financial Development, and Growth.”
ence on Development Economics. Washington, IMF Working Paper. Washington, DC.
DC: World Bank. Grieco, Elizabeth. 2003. The Remittance Behavior of
Collyer, Michael. 2004, “The Development Impact of Immigrant Households: Micronesians in Hawaii
Temporary International Labour Migration on and Guam. New York: LFB Scholarly Publishing.
Southern Mediterranean Sending Countries: Con- Gubert, Flore. 2005. “Migrant Remittances and
trasting Examples of Morocco and Egypt,” Their Impact on the Economic Development of
Working Paper T6. Development Research Centre Sending Countries: The Case of Africa.” Paper pre-
on Migration, Globalisation and Poverty, Univer- sented at the OECD International Conference on
sity of Sussex, Brighton, UK. Migration, Remittances, and the Economic Devel-
Connell, John, and Richard P. C. Brown. 2005. “Re- opment of Sending Countries, February 23–25,
mittances in the Pacific: An Overview.” Asian Marrakech.
Development Bank, Manila. March. www.adb. Harrison, Anne, assisted by Tolani Britton and Annika
org/Documents/Reports/Remittances-Pacific/ Swanson. 2004. “Working Abroad—The Benefits
default.asp. Flowing from Nationals Working in Other
de Luna Martinez, Jose. 2005. “Workers’ Remittances Countries.” Paper prepared for the OECD Round
to Developing Countries: A Survey with Central Table on Sustainable Development, Paris. Sep-
Banks on Selected Public Policy Issues.” Policy tember.
Research Working Paper 3638. World Bank. 3aHarrison.pdf.
El Qorchi, M., S. Maimbo, and J. Wilson. 2003. “In- Hernandez-Coss, Raul E. 2004. “The U.S.-Mexico
formal Funds Transfer Systems: An Analysis of Remittance Corridor: Lessons on Shifting from

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Informal to Formal Remittance Systems.” Finan- Mishra, Prachi. 2005. “Macroeconomic Impact of
cial Sector, World Bank. Remittances in the Caribbean.” Unpublished
IFAD (International Fund for Agricultural Develop- paper. International Monetary Fund, Washington,
ment). 2004. “Remittances and Rural Develop- DC.
ment.” Discussion paper for the Governing Coun- Newland, Kathleen, and Erin Patrick. 2004. “Beyond
cil, 27th session, Rome, February 18–19, 2004. Remittances: The Role of Diaspora in Poverty Re-
IMF (International Monetary Fund). 2005. World duction in Their Countries of Origin.” Migration
Economic Outlook: Globalization and External Policy Institute, Washington, DC.
Imbalances (chapter 2). Washington, DC. April. Olimova, Saodat, and Igor Bosc. 2003. Labour Migra-
IOM. 2003b. World Migration 2003: Managing Mi- tion from Tajikistan. Dushanbe: Mission of
gration—Challenges and Responses for People on the International Organization for Migration.
the Move, Geneva.
IOM (International Organization for Migration). Orozco, Manuel. 2003. “Hometown Associations and
2005. World Migration Report. Geneva. Their Present and Future Partnerships: New De-
Isham, Jonathan, Michael Woolcock, Lant Pritchett, velopment Opportunities? Report commissioned
and Gwen Busby. Forthcoming. “The Varieties of by the U.S. Agency for International Develop-
Resource Experience: Natural Resource Export ment.
Structures and the Political Economy of Eco- Orozco, Manuel, and Katherine Welle. 2004. “Home-
nomic Growth.” In World Bank Economic town Associations and Development: A Look at
Review. Ownership, Sustainability, Correspondence, and
Iskander, Natasha. 2005. “Social Learning as a Pro- Replicability.” Inter-American Dialogue. Unpub-
ductive Project (Zacatecas and Guanajuato’s Cau- lished paper.
tionary Tales).” Paper presented at OECD Inter- Page, John, and Sonia Plaza. 2005. “Migration, Re-
national Conference on Migration, Remittances, mittances, and Development: A Review of Global
and the Economic Development of Sending Coun- Evidence.” Paper presented at the Plenary Session
tries, February 23–25, Marrakech. of the African Economic Research Consortium,
Kaufmann, Dani, Art Kraay, and M. Mastruzzi. 2005. May 29, Nairobi.
“Governance Matters IV: Governance Indicators Rajan, Raghuram, and Arvind Subramanian. 2005.
for 1996–2004.” World Bank Institute. Draft, “What Undermines Aid’s Impact on Growth?”
May 9, 2005. NBER Working Paper 11657. October.
Ketkar, Suhas and Dilip Ratha. 2001. “Securitization Ratha, Dilip. 2003. “Workers’ Remittances: An Im-
of Future Flow Receivables: A Useful Tool for De- portant and Stable Source of External Develop-
veloping Countries.” Finance and Development. ment Finance.” In Global Development Finance
International Monetary Fund. Washington, DC. 2003, Chapter 7. World Bank. Updated in Remit-
Ketkar, Suhas and Dilip Ratha. 2004. “Recent tances and Development: Development Impact
Advances in Future-Flow Securitization.” Paper and Future Prospects, ed. Samuel Maimbo and
presented at the OECD International Conference Dilip Ratha. Washington, DC: World Bank.
on Migration, Remittances and the Economic Ratha, Dilip. 2004. “Understanding the Importance of
Development of Sending Countries.” February Remittance Flows.” Migration Policy Institute.
23–25, Marrakech.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Ratha, Dilip, and Prabal De. 2005. “Predicting Sover-
Shleifer, and Robert Vishny. 1999. “The Quality eign Rating: International Capital Market Access
of Government,” Journal of Law, Economics, for Unrated Countries.” Background paper for
and Organization 15(1): 222–79. this report. World Bank.
Lam, Ricky and Leonard Wantchekon. 2003. Political Ravallion, Martin. 2003. “Measuring Aggregate
Dutch Disease. April 10. New York University. Welfare in Developing Countries: How Well
w w w. n y u . e d u / g s a s / d e p t / p o l i t i c s / Do National Accounts and Surveys Agree?” The
faculty/wantchekon/research/lr-04-10.pdf. Review of Economics and Statistics 85(3):
Magoni, Raphaele. 2004. “France.” In International 645–52.
Migration and Relations with Third Countries: Reuter, Peter, and Edwin M. Truman. 2004. Chasing
European and U.S. Approaches, ed. Jan Niessen Dirty Money: The Fight against Money Launder-
and Yongmi Schibel. Migration Policy Group. ing. Institute for International Economics.
McMahon, Gary. 1997. “The Natural Resource Curse: Washington.
Myth or Reality?” Unpublished paper. World Ross, Michael. 2001. “Does Oil Hinder Democracy?”
Bank Institute, Washington, DC. World Politics 53(3): 325–61.

T R E N D S , D E T E R M I N A N T S , A N D M A C R O E C O N O M I C E F F E C T S O F R E M I T T A N C E S

Sala-i-Martin, Xavier, and Arvind Subramanian. 2003. nent Representatives of Brazil, Chile, France,
“Addressing the Natural Resource Curse: An Germany, and Spain (A/59/719-E/2005/12) Tech-
Illustration from Nigeria.” IMF Working Paper nical Group on Innovative Financing Mecha-
WP/03/139. International Monetary Fund, nisms. Economic and Social Council. New York.
Washington, DC. Walmsley, Terrie, Syud Amer Ahmed, and Christopher
Sayan, Serdar. 2004. “Guest Workers’ Remittances and Parsons. 2005. “The GMig2 Data Base: A Data
Output Fluctuations in Host and Home Coun- Base of Bilateral Labor Migration, Wages and
tries.” Emerging Markets Finance and Trade Remittances.” GTAP Research Memorandum 6,
40(6): 68–81. Center for Global Trade Analysis, University of
———. 2005. “Business Cycles and Workers’ Remit- Sussex.
tances: How Do Migrant Workers Respond to download/2338.pdf.
Cyclical Movements of GDP at Home?” Unpub- Winters, L.A., and Pedro M.G. Martins. 2004. “When
lished paper. Bilkent University. Comparative Advantage Is Not Enough: Business
Siddiqui, Tasneem. 2004. “Efficiency of Migrant Costs in Small Remote Economies.” World Trade
Workers’ Remittances.” Prepared for the Asian Review 3(3): 1–37.
Development Bank. Woodruff, Christopher, and Rene Zenteno. (2004).
Simati, A., and J. Gibson. 2001. “Do Remittances “Remittances and Microenterprises in Mexico.”
Decay? Evidence from Tuvaluan Migrants in New IR/PS working paper. Graduate School of Inter-
Zealand.” Pacific Economic Bulletin 16(1): national Relations and Pacific Studies, University
55–63. of California–San Diego.
Sparreboom, P. 1996. “Migrant Worker Remittances World Bank. 2005a. Global Development Finance
in Lesotho: A Review of the Deferred Pay 2005. Washington, DC.
Scheme.” Social Finance Programme Working World Bank. 2005b. “Migrant Labor Remittances in
Paper 16. International Labour Office, Geneva. the South Asia Region.” Finance and Private Sec-
Stark, Oded, and Robert E.B. Lucas. 1988. “Migra- tor Development Unit/South Asia Region. Report
tion, Remittances, and the Family.” Economic No. 31577. February 2005.
Development and Cultural Change 36: 465–81. Yang, Dean. 2003. “Salvadorans Overseas: The Foun-
Taylor, Edward. 1994. “International Migration and dation of a Pro-Poor Growth Strategy.” Unpub-
Economic Development: A Micro Economy-wide lished paper. Gerald R. Ford School of Public Pol-
Analysis.” In Development Strategy, Employment icy, University of Michigan. www-personal.
and Migration, ed. J. Edward Taylor. Paris:
Organisation for Economic Co-operation and ———. 2004.
Development. ———. 2005. “Coping with Disaster: The Impact of
TEBA (The Employment Bureau of Africa). 2005. “A Hurricanes on International Financial Flows,
Presentation on the Existing and Potential Con- 1970–2001.” Research Program on International
tribution to Rural Southern Africa.” Marshall Migration and Development. DECRG. World
Town, Johannesburg. Bank.
Terry, Donald. 2005. “Sending Money Home: Remit- Yang, Dean, and Claudia Martinez. 2005. “Remit-
tances as a Tool in Latin America and the tances and Poverty in Migrants’ Home Areas:
Caribbean.” Development Bank, Manila, Joint Evidence from the Philippines.” In International
Conference on Remittances, September 12–13, Migration, Remittances, and the Brain Drain, ed.
2005. Caglar Ozden and Maurice Schiff. New York:
UN (United Nations). 2005. Letter of February 23, Palgrave Macmillan.
2005, to the Secretary General from the Perma-

Remittances, Households,
and Poverty

Chapter 4 presented evidence on the macro- (for example, crop failure, job loss, or a
economic dimensions of remittance flows— health crisis);
their overall size, determinants of their com- • Ease working capital constraints on
position (formal versus informal), the role of farms and small-scale entrepreneurs;
government policies in determining their • Lead to increased household expendi-
magnitude and use, and their macroeconomic tures in areas considered to be important
impacts—to developing countries. But as pre- for development, particularly education,
viously noted, these aggregate flows are com- entrepreneurship, and health.1
prised of millions of individual remittance
transfers among private households, all under- Our evaluation of the empirical analysis on
taken by senders and receivers striving to remittances and development is structured as
improve household welfare. This chapter con- follows. In the next section, we consider the ef-
siders the impact of remittance flows at the fects of remittances on poverty and inequality.
micro-level, in particular on the welfare and We then explore how remittances can alleviate
opportunities of the recipient households and the difficulties that households face in smooth-
their members. ing consumption. The next section considers
Evaluating household impact depends on the indirect effects of remittances on house-
data and analysis carried out at the household hold budgets in terms of induced labor supply
level, often through household surveys. effects, increased access to working capital,
Surveys are available for many countries and and multiplier effects. We then examine how
periods, and many of these have common or households allocate remittances to various
comparable structures, but substantial dif- categories of spending, with a particular em-
ferences in coverage and circumstances com- phasis on evidence of remittance-funded in-
plicate their interpretation. Such caveats vestments in human capital, micro-enterprises,
notwithstanding, the evidence presented in and property.
this chapter suggests that remittances can: Before continuing, two broad observations
on the scope and interpretation of the avail-
• Reduce poverty, even where they appear able analysis help to put the results in per-
to have little impact on measured in- spective. First, in evaluating the impact of
equality; remittances, it is important to consider the
• Help smooth household consumption by alternative (or counterfactual) situation that
responding positively to adverse shocks serves as a comparison. If a household

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

member migrates and sends back remittances, accounting for the counterfactual loss of
one could evaluate the net change in the mi- income from migration, as just mentioned).
grant’s contribution—that is, adding the re- But a growing body of evidence from poverty
mittances and subtracting the income of the simulation models, cross-country regressions,
migrant had he or she stayed and worked at and analysis of household survey data shows
home.2 That approach is the appropriate that remittances, in fact, do reduce poverty—
focus when the goal of migration is to gener- although the evidence of their effect on in-
ate remittances, or when we are interested in equality is mixed.
the overall effect of migration on remaining
household members. The alternative is to ig- Remittances reduce poverty
nore the lost domestic contribution—so that In what follows, we present evidence on the
the counterfactual is now simply no remit- poverty effects of remittances, based on three
tances. This approach measures the narrow sources: a poverty simulation model, a cross-
impact of remittances, which seems appropri- country regression analysis, and household
ate when migration is being treated as exoge- survey data from selected countries. The il-
nously given, and our interest is simply in the lustrative poverty simulation model asks a
remittance flows generated by the existing mi- straightforward question: how would poverty
gration stock. The second approach merits rates change in our sample of developing
close attention because the existing migrant countries if remittances were to disappear
stock is large, and also because not all remit- completely? Because this model is easy to im-
tances are received from migrant relatives plement for most countries, it can provide
abroad; third-party remittances are common.3 some sense of the effect of remittances across
Second, in evaluating the benefits of remit- countries. However, the model is relatively
tances, we also need to weigh the welfare of crude and cannot account for the fact that
the migrants themselves. To take a concrete while remittances affect poverty, the level of
example, imagine the migration decision fac- poverty also affects the volume of remittances.
ing a young husband and father in a country In comparison, cross-country regression
with a long-established migration history that analysis requires more data and is harder to
borders on a much richer country. Economic implement, but it is better able to control for
opportunity (and possibly social pressure) reverse causality between remittances and
may make remittance-motivated migration poverty. Household surveys are most likely to
irresistible. However, separated from family provide the data required for a rigorous analy-
and community support, this young man sis of the relationship between remittances
could end up living a quite miserable exis- and poverty. The surveys also allow one to an-
tence. Clearly, a simple tracking of cash pro- alyze the counterfactual loss of income due to
vides an inadequate guide to the welfare im- migration. It is difficult, however, to general-
plications of the move. ize across countries on the basis of household
data, particularly because most available
household surveys do not have usable
Remittances, poverty, data on remittances, especially international
and inequality remittances.

R emittances directly affect poverty by in-

creasing the income of the recipient. They
also indirectly affect poverty in the recipient
To understand how remittance flows might
affect measures of poverty, we start with an il-
lustrative poverty simulation model and ask a
country through their effects on growth, infla- straightforward (although unrealistic) ques-
tion, exchange rates, and access to capital. tion. The model relates the change in poverty
Measuring the impact of remittances is com- to income growth and inequality change. It
plex (in part because of the difficulties of is estimated using cross-country data for

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

Box 5.1 Estimating a cross-country poverty

change model
T he basic idea behind a poverty change model
is that a particular measure of poverty (say the
fraction of the population with incomes below
To estimate the relationship, we use the dataset
assembled by Adams and Page (2005), in which the
observations relate to the period between compara-
$1 per day) is a function of descriptive parameters ble nationally representative household surveys. For
of the income distribution, such as the mean and the each country, we have data from surveys for the ini-
Gini coefficient. Building on Ravallion (1997), we tial and final values of poverty, inequality, and per
posit a conditional constant elasticity specification, capita income. Below is the estimated poverty change
in which there is a constant growth elasticity of equation for the headcount measure of poverty.
poverty reduction that varies with the initial level Using survey income and consumption data as the
of inequality. After reformulation, the basic
income variable (last column), results from the
relationship (see annex 5A.1) relates the rate of
model are robust, with statistically significant im-
poverty change to a measure of inequality-adjusted
pacts from both income growth and inequality
income growth and an income-adjusted change in
inequality. change on the headcount rate.

Estimated poverty change model

Income variable Income variable Survey mean income or consumption

Dependent variable—proportionate GDP per capita
change in the headcount rate
Intercept 0.39 0.33
(2.21) (2.42)

Inequality-adjusted growth 4.93 5.60

(2.57) (2.27)

Income-adjusted inequality change 0.60 1.11

(1.57) (2.82)

R2 0.08 0.30
Number of observations 81 81

Source: World Bank staff estimates.

Note: Standard errors are robust to country-level clustering; t statistics are in parentheses. The model is estimated using first
difference of variables. The poverty line for the headcount poverty calculation is $1.08 in purchasing-power-parity-adjusted
1993 dollars (equivalent to $1 in 1985 dollars). The per capita income measure is mean survey income or consumption
(depending on availability from the survey).

81 countries. The methodology and results are assume that nothing else changes—that there
presented in box 5.1 and in annex 5A.1. The are no offsetting increases in domestic sources
premise behind the analysis is that the incre- of income or other adjustments to spending
mental income from remittances can be ana- behavior or labor supply.4
lyzed in the same way as incremental income Results from the simulation are summarized
from economic growth—so we can simulate in table 5.1 (see also annex 5A.1). We report
the impact of eliminating remittances by mod- averaged results for different groups of
eling an income decline equal to the original countries—first, by distinguishing between
remittance level. For the sake of simplicity, we higher-remittance recipients (greater than

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Table 5.1 Simulated impact of eliminating remittances on poverty rate

Change in headcount rate, no
Country group No. of countries Remittances/GDP (%) Poverty headcount rate Gini change

Low remittances 23 2.2 25.6 5.0

Low headcount rate 12 2.0 11.8 1.2
High headcount rate 11 2.5 40.6 9.1
High remittances 14 11.0 24.8 12.2
Low headcount rate 7 8.0 10.7 4.1
High headcount rate 7 14.1 38.9 20.3

Source: World Bank staff estimates.

Note: Low (high) remittances refer to the remittance share in GDP less (greater) than 4 percent. Low (high) headcount rate refers
to a rate less (greater) than 20 percent. Allowing inequality to change 2 Gini points has minimal effects on the change in the
headcount rate. See annex 5A.1 for detailed results.

4 percent of GDP) and lower (less than 4 per- across surveys in how remittances are ac-
cent but greater than 1 percent), and second, by counted for in the household surveys.5
the extent of poverty (headcount above 20 per- The results just described provide an indi-
cent or below 20 percent). cation of the role that remittances can play in
These results show that the impact of elim- reducing poverty, but because of the simplicity
inating remittances depends on how large they of the model and other limitations, the results
are to begin with (higher initial levels mean are not conclusive. More rigorous analytic
steeper income declines), the initial extent of work has been undertaken to investigate the
poverty, and the degree of inequality. For ex- link between remittances and poverty based
ample, the average increase in the headcount on careful analysis of cross-country data.
ratio for higher-remittance countries (12.2 per- In a model that relates national poverty
centage points) is more than twice that of levels to mean income and the Gini measure
the lower-remittance countries (5 percentage of inequality for 71 developing countries, a
points). Similarly, with each of these two 10 percent increase in per capita official inter-
groups, the impact is much greater for those national remittances leads to a 3.5 percent de-
countries with higher headcount ratios to start cline in the share of people living in poverty
with. The estimated impact of inequality— (Adams and Page 2005).6 Other recent studies
an assumed 2 point worsening in the Gini have broadly confirmed these findings, includ-
coefficient—has only a small marginal impact ing IMF (2005) (see chapter 2), which uses a
on the estimated change in the poverty rate. sample of 101 countries for the period
This simple analysis has significant limita- 1970–2003.
tions. First, the simulated effects depend on Although the available evidence is still rel-
accurate country-level measures of remit- atively limited, growing evidence from house-
tances, which, as emphasized in chapter 4, hold survey data complements the findings of
are of variable reliability. Second, many of the the model that international remittances have
country simulations are made outside the sam- reduced the incidence and severity of poverty
ple used for the regression analysis and are in several low-income countries. According
therefore subject to the standard out-of- to that evidence, remittances are believed to
sample prediction problems. Third, the analy- have reduced the poverty headcount ratio by
sis assumes that remittances are included in 11 percentage points in Uganda, 6 percentage
household income when calculating the mea- points in Bangladesh, and 5 percentage points
sures of poverty and inequality from house- in Ghana (Adams 2005b). Completely remov-
hold surveys. In reality, there is variation ing remittances for Lesotho would raise the

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

headcount poverty ratio (with a poverty line The effect of remittances on inequality
equal to 60 percent of mean household expen- is unclear
diture) from 52 to 63 percent (Gustafsson and In contrast to the link between remittances
Makonnen 1993). and poverty, no strong conclusion is found in
While remittances had only a limited role household studies of the relationship between
in reducing the number of poor people in remittances and inequality: remittances some-
Guatemala, they did significantly reduce the times go disproportionately to better-off
depth and severity of poverty (Adams households and so widen disparities, but in
2004a).7 International remittances accounted other cases they appear to target the less well
for 60 percent of income for households in the off, causing disparities to shrink. Some studies
lowest income decile, but were not very large suggest that the remittances from new migra-
for households located near the poverty line tion may raise inequality in the short term,
(roughly the fifth income decile). As a result, but the effect on inequality is small over the
international remittances had more impact on long term.8 Calculations that impute incomes
reducing the depth of poverty than on the for the migrant had he stayed and worked at
poverty headcount; in other words, they were home generally show an increase in inequality
really helpful for the poorest of the poor. from the combined effect of migration and re-
Wodon and others (2002) conclude that in mittances. For example, inequality was found
Guerrero and Oaxaca, two southern Mexican to have increased in Bluefields, Nicaragua,
states with significant international emigra- when an imputation was made for the lost do-
tion and remittance inflows, the share of the mestic income of migrants, but it fell when the
population living in poverty is lower by 2 per- domestic income of migrants was ignored
centage points due to remittance income. They (Barham and Boucher 1998).
argue that this poverty effect is similar in mag- Two recent studies, however, did not find
nitude to that of many government programs an increase in inequality even after control-
in poverty reduction, education, health, and ling for the counterfactual income loss from
nutrition. Taylor, Mora, and Adams (2005), migration: Adams (2005a) found that in
using data from a 2003 survey, find that inter- Ghana, the inclusion of international remit-
national remittances account for 15 percent of tances in household expenditures led to only a
per capita household income in rural Mexico. slight increase in income inequality, but that
They conclude that an increase in interna- the Gini coefficient remained relatively stable,
tional remittances would reduce both the between 0.38 and 0.40.9 De and Ratha (2005)
poverty headcount and the poverty gap. found that in Sri Lanka, the Gini coefficient
Yang and Martinez (2005) studied the im- drops from 0.46 to 0.40 as a result of remit-
pact of variations in the exchange rate on re- tance receipt.
mittances sent by Filipino workers and the ul- Differences in findings on the impact of
timate impact of remittances on poverty in the remittances on inequality also stem from vary-
recipient regions. Using a large dataset from ing geographic and historic circumstances,
the Overseas Filipino Survey, they found that such as the distance from high-income
an appreciation of the Philippine peso led to destination countries and the prevalence of
an increase in remittance flows, which con- networks of earlier migrants. Both proximity
tributed to the reduction in poverty. Interest- to high-income countries and established
ingly, increased remittances not only reduced networks will tend to reduce the cost of
poverty in the migrant families, they also migration, making migration an option for
had spillover effects on nonmigrant families. poorer (and often credit-constrained) house-
(We will have more to say on multiplier effects holds.10 For example, remittances to a Mexi-
later in this chapter.) can village with a well-established history of

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

international migration had an equalizing ef-

Figure 5.1 Sri Lankan migration
fect, whereas remittances to another Mexican
village for which international migration was a Percent of Sri Lankan households with
a relatively new phenomenon tended to make migrants abroad

the distribution of income more unequal Percent

(Stark, Taylor, and Yitzhaki 1986). For a large 12
number of Mexican communities, the overall 10
impact of migration and remittances is esti-
mated to reduce inequality for communities
with relatively high levels of past migration 6

(McKenzie and Rapoport 2004).11 4

One reason for the inconclusive empirical
evidence on inequality effects is that the Gini
coefficient does not adequately capture income
1 2 3 4 5 6 7 8 9 10
mobility; it remains unchanged, for example, Income decile
if one person moves up and another moves
Source: De and Ratha 2005.
down the income ladder. Using household sur-
vey data from Sri Lanka, De and Ratha (2005)
show that the poor income deciles also have
b Percent of Sri Lankan households that moved up
significant overseas migration and that remit- to a higher income decile after receiving remittances
tance recipients in the middle income deciles
move up the income ladder (figure 5.1).12
Beyond the contradictory or inconclusive
results, some scholars question whether the
link between remittances and inequality is all
that important. Inequality matters when it in- 10

terferes with the functioning of the economy 5

(for example, when credit constraints bind 0

more households13) or the political system
(for example, when growing inequality in-
creases support for governments that pursue 1 2 3 4 5 6 7 8 9 10

damaging populist policies14). Greater in- Income decile

equality may also be considered bad because Source: De and Ratha 2005.
of its impact on social welfare (see Sen 1973
for a discussion). But it should be kept in mind
that in the context of remittances, inequality
relates to income differences among groups that lack access to insurance and credit
that would all be viewed as relatively poor in markets are vulnerable to severe declines in
an industrial-country context. The rich in de- income from adverse shocks, and they may
veloping countries probably receive little in be forced to forgo income-generating—but
the way of remittances; the rich who migrate risky—strategies (Morduch 1994). Informal
tend to take their families with them. community institutions generally play a limited
role in mitigating risk (see, for example, Coate
and Ravallion 1993 and Fafchamps 2004), es-
Remittances and household pecially in the face of adverse events such as a
consumption smoothing community-wide crop failure. One strategy to

R emittances may play a significant role in

smoothing consumption. Poor households
reduce risk is for households to send family
members to other regions or countries, where

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

they are not likely to face the same income Consider, for example:
shocks as those found in the domestic
• Migrants responded to the cost of hur-
market.15 Migration patterns and policies that
ricane damage borne by Jamaican house-
encourage migrants to travel unaccompanied
holds, with each additional dollar of
by family members encourage this form of risk
hurricane damage leading to $0.25 in
additional remittances (Clarke and
There is some evidence that remittances
Wallsten 2004. The authors use panel
from internal migration provide insurance.
data to control for the household-level
Remittances to Botswana increased with the
risk aversion and vulnerability effects
extent of drought in the migrant’s home re-
that potentially bias the estimates.18)
gion, and the responsiveness of remittance lev-
• Remittances are estimated to have re-
els to drought was greater for households with
placed 60 percent of income loss due to
more drought-sensitive assets such as cattle
weather-related shocks in a sample of
(Lucas and Stark 1985).16 The anticipation of
Filipino households (Yang and Choi
insurance may allow the household to pursue
2005.19 Rainfall is used as an instrument
a more risky asset accumulation strategy—
for income to avoid reverse causality;
although it is also possible that households
panel data are used to control for the ten-
with more to lose from drought (whatever the
dency for risk-averse households to locate
reason) are simply more likely to receive re-
in places where incomes are more stable
mittances. The likelihood that Thai internal
and to send migrants to manage risk.)
migrants move to Bangkok is reduced the
• In cross-country data, a dollar’s worth of
more closely income in Bangkok aligns with
hurricane damage leads to roughly $0.13
income in the province of origin (Paulson
in additional remittances in the year of
2000).17 The effect is particularly strong for
the hurricane and $0.28 cents over five
remittances to rural households, which are
years (Yang 2005). (Yang uses meteoro-
likely to be poorer and have less access to for-
logical data to instrument for reported
mal insurance products to mitigate weather-
disaster damage, because damage reports
related risks. The more volatile a household’s
may be affected by the anticipation of
income (and the more restricted its ability to
financial flows.)
self-insure), the greater the distance that
households in rural India tend to send their
daughters to marry (Rosenzweig and Stark Remittances and indirect effects
1989). Greater distance means that the covari- on household income
ance of income shocks with the home region
will be smaller, facilitating consumption-
smoothing transfers between these related
R emittances may indirectly affect house-
hold income through changes to the labor
supply of those remaining behind; relaxation
households. of working capital constraints that expand in-
Studies of how remittances respond to ad- come from entrepreneurial or farming activi-
verse household shocks generally support the ties; and multiplier effects on household in-
view that remittances provide some insurance. come. Unfortunately, the evidence on each of
However, interpreting these correlations is these channels is quite limited, so we are con-
complicated by the likelihood of reverse strained here to identifying important areas for
causality (remittances can influence household additional research.
outcomes as well as be influenced by them)
and omitted variable bias (certain hard-to- Remittances may affect labor supply
measure household characteristics may affect Remittances may tend to reduce the supply of
a household’s susceptibility to risks as well labor provided by remaining household
as the likelihood of receiving remittances). members, who may take a portion of the

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

remittance gain as leisure. This income effect is household assets (which determines their value
generally not a concern, because it represents as collateral) and on the availability of inputs
part of the welfare gain from remittances. By that complement entrepreneurial activity (Tay-
contrast, remittances may change the return to lor and Wyatt 1996). The role of remittances
supplying labor, for example, if the migrant in relaxing household credit constraints in
conditions the remittance on low household rural cropping income in China dominated the
income.20 Such a substitution effect will re- direct loss of productive labor from migration,
duce the welfare gain from remittances by so that internal migration increased per capita
distorting household labor decisions. household income (excluding remittances) by
However, it is difficult to separate income 14 to 30 percent (de Brauw, Taylor, and
and substitution effects of remittances on the Rozelle 2001). Mishra (2005) found that a 1
labor supply of those remaining behind. Look- percentage point increase in remittance in-
ing at the overall effect, a rise in remittances flows in 13 Caribbean countries increased pri-
reduced labor force participation in Managua, vate investment by 0.6 percentage point (all
Nicaragua, but increased self-employment measured relative to GDP).
(Funkhouser 1992). Remittances were esti- Remittances may ease credit constraints be-
mated to reduce the participation rates of cause a stable stream of remittance income
remaining household heads in a number of may make households more creditworthy in
Caribbean countries, although the direction the eyes of formal sector financial institutions.
of causality was hard to establish (Itzigsohn Remittance receipts that increase when the
1995). Yang (2004) points to more encourag- household receives an adverse shock may be
ing labor-supply effects than the standard even more important in relaxing credit con-
model when he determined that remittances straints, since they increase the lender’s confi-
reduce the supply of child labor but increase dence that they will be repaid even if things
that of adult labor. turn out badly for the household. This credit-
worthiness effect deserves careful empirical
Remittances provide working capital investigation, given the increasing interest in
There is some evidence that remittances pro- channeling remittances through formal finan-
vide working capital to households that lack cial channels.
access to credit markets. For example, migra-
tion to South Africa’s mines initially reduced Remittances may have multiplier effects
agricultural production in countries of origin, Some studies have found that remittances
because labor was removed from the farm have a multiplier effect, whereby the increase
(Lucas 1987). However, over time production in domestic income is some multiple of the re-
rose with migration, perhaps due to remit- mittance income. For example, each dollar
tance-funded capital investment and a greater sent by Mexican migrants to the United States
willingness to take risks with agricultural pro- was estimated to boost Mexican GDP by
duction, owing to the more diversified sources $2.90 (Adelman and Taylor 1992). Such mul-
of family income. Remittances had a small tipliers will occur if output is constrained by
negative effect on household income for Mex- insufficient demand. However, in many devel-
ico in 1982, but a large positive effect for 1988 oping countries sustained underemployment is
(Taylor 1992). One possible explanation is likely to have supply-side causes, for example,
that over time the development of migrant net- government policies that increase the cost of
works allowed migration from poorer house- hiring and firing workers, so that increased
holds that are more likely to be credit con- demand will ultimately result in higher infla-
strained (see the discussion of inequality, tion rather than increased output.21
above). The effect of remittances on household Nevertheless, there may be greater scope
income depends on both the liquidity of for sustained multiplier effects at the regional

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

level. The local spending of remittance income possible existence of positive externalities from
will generate further income for other local investment expenditure.24 Thus the way that
households, which in turn is likely to cause remittances are allocated by households may
local inflation for nontraded goods and possi- affect the social value of a given remittance
bly a small increase in national inflation. A na- flow.
tional government with a formal or informal The rate of investment of remittance in-
inflation target is likely to respond to any in- come will be high when:
crease in the national inflation rate by tighten-
ing monetary policy, thereby leading to an off- • Remittance flows are viewed by the
setting effect on national aggregate demand. household as transitory rather than per-
The net effect would be multiplier effects at manent and thus should be saved (and
the local level but not at the national level. In- invested) rather than spent.
deed, the local gains come partly at the ex- • The sender conditions the remittance on
pense of the regions that do not receive the re- it being spent for particular purposes,
mittances but are forced to suffer the tighter which are more likely to involve invest-
monetary policy. ment than current consumption. Exam-
Remittances also may have multiplier ef- ples include education or the purchase of
fects in the context of increasing returns, typi- new farm machinery.
cally as the expansion of one sector increases • The remittance is targeted (or “tagged”)
the optimal size of other sectors.22 Although to household members more likely to
such income-expanding feedback loops could use the funds for investment purposes
be present at the national level, they are again (women rather than men).25
more likely to be relevant at the regional level, • Households practice a form of mental
because expanding regions attract labor and accounting with their overall budget,
capital from elsewhere in the economy. The with remittances being disproportion-
bottom line is that remittance-induced multi- ately put in accounts set aside for invest-
plier effects cannot be ruled out—especially at ment purposes.26
the regional level—but our current empirical
understanding of their importance is quite On the other hand, some of the literature al-
limited. ready reviewed suggests reasons to expect that
the marginal propensity to invest remittance
income will be low when (a) remittances are
Remittances, savings, targeted to poor households that are struggling
and investment to meet subsistence needs and (b) they are tar-

D oes it matter how households allocate re-

mittance income between consumption
and saving? Allocations to the latter may boost
geted to credit-constrained households that are
experiencing adverse consumption shocks.
The empirical challenge in identifying the
household investment or national investment causal effect of remittances on investment is
through allocation to financial assets. But from that remittances are likely to be correlated
a welfare perspective, an extra dollar of invest- with the extent of opportunities for invest-
ment is only better than an extra dollar of pre- ment, thereby biasing the estimated remittance
sent consumption if the marginal social value effect. That correlation could be positive or
of investment is greater than its marginal pri- negative. When more enterprising households
vate value.23 Although a number of factors can are the ones sending migrants and the ones
drive a wedge between social and private val- with substantial investment opportunities,
ues (such as capital income taxes, monopoly high remittances will be wrongly associated
powers, and credit constraints), one prominent with high investment. On the other hand, to
reason raised in the development context is the the extent that households send migrants

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

when investment opportunities are absent at largest impact (an additional 0.89 years of
home, then high remittances will be wrongly schooling) for girls in households where the
associated with low investment. The empirical mother has a low level of education (Hanson
solution is to find a source of variation in re- and Woodruff 2003).
mittances that is plausibly unrelated to house- Health status is both an important compo-
hold investment opportunities. nent of human capital and a central element of
Measuring the impact of remittances on in- well-being in its own right. Unfortunately,
vestment—either in physical or in human cap- the effect of migration on the health of
ital—is not easy. Household budget surveys family members remaining behind—notably
are best suited for this purpose, but most of children—is poorly understood. Migration
the existing surveys either do not record data from Mexico is associated with lower (by
on international remittances or are poorly 3 percent) infant mortality and higher birth
designed. Since remittances are fungible, it is weights of children left behind (Hildebrandt
difficult to isolate their effects from those of and McKenzie 2005). The positive health ef-
other sources of income. Simply asking how fects come through increased access to health-
remittances are spent is unlikely to reveal the related knowledge as well as through in-
true marginal effect on spending, because re- creased household wealth. Notwithstanding
mittances, even when used for investment pur- these encouraging outcomes, the authors cau-
poses, may free up the marginal dollars for tion that the impact of migration on child
consumption spending.27 health is quite nuanced, with migration asso-
ciated with lower measures of preventive
health care such as breast-feeding and vacci-
Remittances can lead to investments nations.30 De and Ratha (2005) find that in
in education and health Sri Lanka, remittance income has a positive
Some of the clearest evidence for remittance- and significant impact on the weight of chil-
induced investments comes from work on dren under five; this result is especially strong
human capital. The dramatic depreciation of for female-headed households. However, the
the Philippine exchange rate during the Asian health impact of absenteeism of one of the
financial crisis increased remittances from parents is negative.
Filipino migrants (because from the migrants’
perspective, exchange-rate depreciation raised Remittances can encourage
the relative price of their own consumption entrepreneurship
in the destination country compared with There has been a marked shift from the belief
consumption by household members back that migrants are unlikely to establish new
home), leading to greater child schooling, re- business enterprises in their countries of origin
duced child labor, and increased educational (either upon return or through remittance
expenditure in origin households (Yang financing) to the view that migration encour-
2004).28 In El Salvador, remittances are esti- ages entrepreneurship. Large receipts of
mated to reduce the probability of children remittances from the United States are associ-
leaving school by 10 times the effect of other ated with a greater likelihood of productive
sources of income in urban areas and by 2.6 investment in Mexico (Massey and Parrado
times in rural areas (Cox Edwards and Ureta 1998).31 A survey of 6,000 small firms in 44
2003).29 They speculate that remittances have urban areas in Mexico shows that remittances
a disproportionate influence on schooling ex- are responsible for almost 20 percent of the
penditures because the migrant has made it a total capital in urban micro-enterprises
condition for the financial support. Mexican (Woodruff and Zenteno 2001). The share rises
children in households with migrants com- to one-third for the 10 states with the highest
pleted significantly more schooling, with the rates of United States–bound migration.

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

Remittances also appear to ease credit con- perfect substitutes by the household. In con-
straints on new business formation in the trast, studies such as Yang (2004) econometri-
Philippines (Yang 2004). The effect of exoge- cally estimate expenditure propensities given
nous increases in remittance income on the exogenous changes in remittance income, so
probability of entering into entrepreneurship that the estimates should be less susceptible to
is larger for low- to middle-income house- the fungibility problem. A second explanation
holds, which are the ones most likely to face for the different results is that the econometric
credit constraints. Policies that facilitate easy studies measure marginal propensities,
exit and reentry for migrants may encourage whereas the direct surveys measure average
increased involvement in remittance-funded propensities. It is the marginal propensity that
investments or enterprises. is of interest when we consider the expendi-
ture effects of policies that increase remittance
Remittances are often invested flows.33
by recipient households The role of remittances in funding invest-
Contrary to the conventional wisdom that ment has recently been questioned in a macro-
remittances tend to be “frittered away” by re- economic paper by Chami, Fullenkamp, and
cipient households, recent work has estimated Jahjah (2005), who find that remittances tend
that a large proportion of remittance income to be negatively associated with economic
is saved. Only 12 percent of net increments growth. This countercyclical behavior of re-
to expenditure by rural Egyptian households mittances is consistent with the evidence dis-
were allocated to consumption, with large cussed above that remittances respond to ad-
propensities to invest in the construction and verse household shocks. But the observation
repair of houses, and in agricultural or build- that remittances tend to move countercycli-
ing land (Adams 1991). This relatively high cally does not necessarily obviate their role in
propensity to invest is assumed to result from funding investment. The micro studies we re-
households treating remittance receipts as viewed point to remittances as both smooth-
temporary income flows, which forward- ing consumption and providing funds for in-
looking households save (and invest) rather vestment. Moreover, the increased flow of
than consume. These findings are largely con- remittances in the face of adverse shocks may
firmed in a later study of Pakistani households allow households to sustain funding for key
(Adams 1998).32 In Guatemala, remittance- investments in areas such as business working
receiving households are found to have lower capital, education, and health care.
marginal propensities to consume and a The evidence reviewed in this chapter sug-
higher propensity to invest in education, gests that remittances play multifaceted roles
health, and housing than other households in poverty reduction, consumption smoothing,
(Adams 2005c). and investment, with the balance of roles
It should be noted that some survey results varying by time and place.
for a number of Latin American countries
point to much higher propensities to consume
remittance income (see, for example, IADB-
MIF 2004). The percentages of remittances
Annex 5A.1 Poverty simulation
spent on household expenditures are 78 per- model: description and results
cent in Mexico, 77 percent in Central Amer-
ica, and 61 percent in Ecuador, while spending
on real estate and education is low. However,
T he poverty change model assumes that a
particular measure of poverty (say the
fraction of the population with incomes below
surveys of how income from a particular $1 per day) is a function of descriptive parame-
source is spent tend to be unreliable, because ters of the income distribution, such as the
monies from different sources are considered mean and the Gini coefficient. Building on

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

Ravallion (1997), we assume that there is a measure of inequality (which we take to be the
constant growth elasticity of poverty reduction, Gini coefficient), and A and ␤ are parameters.
but we allow that elasticity to vary with the Differentiating the poverty equation and writ-
initial level of inequality.34 We call this a ing it in rate-of-change form yields our basic
conditional constant elasticity specification. poverty change model,
Specifically, the poverty measure, P, is given by,
dP dY
 ␤(1  I)  ␤lnY dI
P  AY␤(1  I), P Y

where Y is per capita income (measured as This equation can be interpreted as saying
mean survey income or consumption), I is the that the rate of poverty change depends on a

Table 5A.1 Effect of removing remittances on the poverty headcount rate

Poverty headcount rate Remittances as share
Country Survey year ($1 a day, PPP, $1993) of GDP (%) Change in headcount rate (Gini ⴝ 0)

Using per Using survey

capita GDP mean income
Armenia 1998 13 5 2 4
Azerbaijan 2001 4 2 0 0
Bangladesh 2000 36 4 5 13
Bolivia 1999 14 1 0 1
Botswana 1993 31 2 1 3
Burkina Faso 1998 45 3 3 4
Colombia 1999 8 2 0 0
Côte d’Ivoire 1998 16 1 0 1
Ecuador 1998 18 3 2 3
Egypt, Arab Rep. 2000 3 3 0 1
El Salvador 2000 31 13 10 41
Gambia, The 1998 54 7 9 12
Georgia 2001 3 6 0 0
Guatemala 2000 16 3 1 2
Honduras 1999 21 6 3 4
India 2000 35 3 3 23
Kenya 1997 23 3 2 2
Lao PDR 1997 26 2 2 4
Lesotho 1995 36 44 29 34
Mali 1994 72 6 10 17
Mexico 2000 10 1 0 1
Moldova 2002 22 19 13 21
Mozambique 1997 38 2 2 2
Nepal 1996 39 1 1 2
Nicaragua 2001 45 2 3 13
Pakistan 1999 13 2 1 2
Panama 2000 7 2 0 0
Paraguay 1999 15 4 1 2
Peru 2000 18 1 1 2
Philippines 2000 16 8 3 10
Senegal 1995 22 3 2 3
Sierra Leone 1989 57 3 3 4
Sri Lanka 1996 7 6 1 3
Swaziland 1996 8 6 1 1
Tajikistan 1999 14 6 3 3
Uganda 1999 85 4 9 40
Yemen, Rep. 1998 16 19 10 7

Source: World Bank staff estimates.

Note: Remittance share is 1995 for Sierra Leone. Estimates are based on poverty reduction model.

R E M I T T A N C E S , H O U S E H O L D S , A N D P O V E R T Y

measure of inequality-adjusted growth and an states that remittances “did nothing to convert stagna-
income-adjusted change in inequality.35 tion to development” in abandoned regions.
2. Since data on what the migrant was earning be-
To estimate the relationship, we utilize the
fore leaving are typically unavailable, the lost domestic
dataset assembled by Adams and Page (2005).
income is estimated or imputed based on observed
The observations relate to spells between com- characteristics of the migrant and on knowledge of
parable nationally representative household how those characteristics are rewarded in the domestic
surveys. For a given spell, we have data for the economy.
initial and final values of poverty, inequality, 3. In the Sri Lanka Integrated Survey 1999–2000,
and per capita income.36 The estimated equa- nearly a third of households receiving remittances did
not report having a migrant member overseas. It is pos-
tion for the headcount measure of poverty is
sible that those households received remittances from
reported in the table in box 5.1 in the main
their extended family; it is also possible that they re-
text. Similar results were found for the ceived remittances from friends (De and Ratha 2005).
poverty gap measure. The literature also notes third-party remittances in
The next step is to simulate the effect of other countries (see, for example, Yang 2004 for the
removing remittances on the poverty measure Philippines). A survey of African diasporas in Belgium
under various assumptions about how remit- found that more than one member in a typical migrant
household sent regular remittances and that each re-
tances affect inequality. The proportionate in-
mitter might send remittances to different recipients.
crease in per capita income due to remittances
4. This situation is akin to the second counterfactual
is given simply by the share of remittances in (a decline in remittances but no change in migration
GDP multiplied by the ratio of per capita GDP and hence household income) discussed at the outset.
to mean survey income/consumption.37 It is To the extent that a decline in remittance income may
important to emphasize that the simulated encourage households to devote more labor hours to
poverty-increasing effect of removing remit- domestic income-generating activities, the total decline
in household income and the consequent poverty effect
tances applies to the latest year for which a
may be smaller than assumed in the simulation, a point
survey is available for that country, and thus
made in Adams (2004a).
different years are being used for different 5. If we adopt the other extreme assumption—that
countries. Care should thus be taken in mak- remittances are not included in household income—the
ing comparisons about the importance of re- results can be interpreted with a simple reversal of the
mittances in reducing poverty across different sign, which gives the reduction in poverty that would
countries. Where the headcount rate is below result if remittances were included.
6. As higher poverty increases the incentive to mi-
2 percent we do not attempt to estimate the
grate, the ordinary least squares (OLS) estimates of
poverty change effect of remittances.
the impact of remittances and stock of migrants are
Table 5A.1 shows results on the poverty biased downward. On the other hand, when credit-
headcount when remittances are removed constrained poor families do not have the resources to
(assuming there is no impact of remittances on send migrants, the OLS coefficients are biased upward
inequality) for 37 countries where remittances as poorer countries send fewer migrants. To deal with
are above 1 percent of GDP and where the the bias, Adams and Page (2005) allow for remittances
and emigrant stock variables in their regressions, using
poverty headcount rate is greater than 2 per-
measures of international distance, government stabil-
cent at the outset.
ity, and levels of education.
7. The poverty depth is the average shortfall below
the poverty line expressed as a fraction of the poverty
Notes line (or simply the poverty gap ratio); and poverty
1. This represents a significant shift from the tradi- severity is the squared poverty gap ratio. A key feature
tional earlier pessimism about the role of remittances of this severity measure is that it is sensitive to the dis-
in development. For example, Papademetriou and tribution of income among the poor (Foster, Greer, and
Martin (1991) emphasize how migration increases the Thorbecke 1984).
dependence of emigration countries that are unable “to 8. Stark, Taylor, and Yitzhaki (1986) found that a
regulate or channel remittances,” while Jacobs (1984) 1 percent increase in international remittances leads to

G L O B A L E C O N O M I C P R O S P E C T S 2 0 0 6

a 0.14 percent increase in the Gini coefficient in the 17. Rainfall is used as an instrument for provincial
case of Mexican villages with a short migration his- income in establishing the covariance pattern.
tory; but in other villages with a long migration his- 18. Simple cross-sectional estimates of how remit-
tory, the Gini coefficient actually declines by 0.01 per- tances respond to hurricane damage will be biased down-
cent. Taylor (1992), McKenzie and Rapoport (2004), ward if more risk-averse households are more likely to
and Taylor, Mora, and Adams (2005) also find nega- send migrants as a general insurance strategy and are
tive effects on the Gini in the case of Mexico. more likely to take actions to reduce the risk of costly
9. Nearly 40 percent of households in a representa- damage to the home. They will be biased upward if
tive sample in Ghana receive remittances, of those re- households with more vulnerable dwellings are more
mittances, 4 times as many households receive internal likely to send migrants and more likely to suffer hurricane
than international remittances (Adams 2005b). House- damage. An additional complication is moral hazard,
holds receiving internal remittances are disproportion- where insured (remittance receiving) households have
ately poor, indicating the importance of internal remit- less incentive to avoid risky behavior. Clarke and
tances in reducing poverty. Households surveys in Wallsten (2004) deal with the potential endogeneity
Europe and Central Asia also show that in a number of problem by using the average damage done in the neigh-
countries (Albania, Kosovo, Moldova, and Tajikistan, borhood as an instrument for household-specific damage.
for example) a large number of households receive re- 19. They cannot reject the null hypothesis that all
mittances, many in rural areas (World Bank 2005). of an exogenous decline in income is matched by an
10. Research has also shown that the inclusion of increase in remittances.
remittance-induced, indirect effects on income—such 20. There is a disincentive to work if remittances
as income-induced reductions in nonmigrant labor are conditioned on low income. Conversely, if remit-
supply or increases in entrepreneurial income due to tances are conditioned on domestic labor supply—“I
the relaxation of credit constraints—can change the will help you if you help yourself”—there is an added
direction of the inequality effect. incentive to work. From the migrant’s perspective,
11. The authors experiment with various instru- there is some similarity with the challenges faced by
ments for location-specific migration levels in their re- governments in providing social assistance without cre-
gressions. The instruments include the historic (1924) ating poverty traps and dependency. A traditional
state-level migration rate and unemployment rates for “welfare” model conditions remittances on household
the U.S. state that includes the city that is the likely income, whereas the modern “workfare” model at-
destination for migrants from a particular Mexican tempts to condition remittances on household effort in
location. an attempt to avoid putting the household in a trap