Managing International Migration For Development in East Asia

Public Disclosure Authorized
93255
Managing International Migration

for Development in East Asia
Citation
Adams, Richard H. Jr; Ahsan, Ahmad; Adams, Richard H. Jr; Ahsan, Ahmad. 2014. Managing International
Migration for development in East Asia (English). Washington, DC: World Bank Group.
http://documents.worldbank.org/curated/en/825411468023944864/Managing-International-Migration-for-
development-in-East-Asia
Editors
Richard H. Adams, Jr.

Ahmad Ahsan
Office of the Chief Economist

East Asia and Pacific Region
World Bank
Managing International Migration
For Development in East Asia
Editors
Richard H. Adams, Jr.

Ahmad Ahsan
Office of the Chief Economist

East Asia and Pacific Region
World Bank
Washington, DC 20433
June 2014
© 2014 International Bank for Reconstruction and Development / International Development
Association or
The World Bank
1818 H Street NW
Washington DC 20433
USA
Telephone: 1-202-473-1000
Internet: www.worldbank.org
This report was printed by The World Bank Printing & Multimedia Department using only
FSC certified paper and environmentally sustainable water based inks.
This work is a product of the staff of the World Bank with external contributions. The
findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views
of the World Bank, its Board of Executive Directors, 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 ASEAN or 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 subject to copyright. Because the World Bank encourage
dissemination of its knowledge, this work may be freely quoted or reprinted, in whole or in part, for
noncommercial purposes as long as full attribution to this work is given.
Any queries on rights and licenses, including subsidiary rights, should be addressed to the
Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-
522-2422; e-mail: pubrights@worldbank.org.
ACKNOWLEDGEMENTS
This collection of papers was produced by the Office of the Chief Economist, East Asia and
Pacific Region and used as background papers for the World Bank report, International
Migration and Development in the East Asia and Pacific Region, 2014, forthcoming. Bert
Hofman (Chief Economist of East Asia and Pacific) provided guidance to the work. The
Institute for Policy Studies, National University of Singapore, kindly acted as the host for the
2010 Workshop where paper authors presented the original drafts of the papers. Participants
at that workshop, including Fabio Baggio, Yukon Huang, Vikram Nehru, Chia Siow Ye,
provided very helpful comments on the papers. Doris Chung and Mildred Gonsalvez
processed this report. The World Bank also gratefully acknowledges support from the
Government of Japan and Government of Korea that funded several of the papers in this
volume
TABLE OF CONTENTS
TABLE OF CONTENTS .........................................................................................................................i
OVERVIEW ......................................................................................................................................... iii
PART 1. MIGRATION AND ITS IMPACTS
Chapter 1: Labor Migration and Economic Growth in East and South-East Asia
by Terrie Walmsley, Angel Aguiar, S. Amer Ahmed ............................................................................. 1
Chapter 2: Remittances, Household Investment and Poverty in Indonesia

by Richard H. Adams, Jr . and Alfredo Cuecuecha .......................................................................... 29
Chapter 3: More or Less Consumption?: The Effect of Remittances on Filipino

Household Spending Behavior
by Emily C. A. Cabegin and Michael Alba, ........................................................................................ 53
Chapter 4: Impacts of International Migration and Remittances on Child Outcomes

and Labor Supply in Indonesia: How Does Gender Matter?
by Trang Nguyen and Ririn Purnamasari ....................................................................................... 84
Chapter 5: The Effects of Immigration on the Thai Wage Structure

by Dilaka Lathapipat ........................................................................................................................ 111
Chapter 6: The Impact of Foreign Labor on Labor Productivity and Wages in Malaysian
Manufacturing, 2000-2006
by Tham Siew Yean and Liew Chei Siang ................................................................................... 136
PART 2: MANAGING MIGRATION
Chapter 7: Indonesia’s Regulatory, Institutional and Governance Structure for

International Labor Migration
by Ari Kuncoro, Arie Damayanti and Ifa Isfandiarni ......................................................................... 159
Chapter 8: The Philippine Labor Migration Industry for Health and Educational
Services: Regulatory and Governance Structures
by, Tereso S. Tullao, Jr., Mitzie I. P. Conchada and John P. R. Rivera........................................... 177
i
Chapter 9: Vietnam’s Regulatory, Institutional and Governance Structure for
International Labor Migration
by Nguyen Huyen Le and Daniel Mont .................................................................................199
Chapter 10: Singapore’s System for Managing Foreign Manpower

By Yap Mui Teng , Senior Research Fellow, Institute of Policy Studies, Lee Kuan Yew School of
Public Policy, Singapore ..................................................................................................................... 220
Chapter 11: The Management of Foreign Workers in Malaysia: Institutions and Governance
Regime
by Azizah Kassim, Terence Too and Mahani Zainal Abidin .................................................241
Chapter 12: A Cost-Benefit Analysis of the Legal Status of Migrant Workers in Thailand
by Charamporn Holumyong and Sureeporn Punpuing ................................................................... 263
Contributors ……………………………………………………………………………………….283
ii
OVERVIEW
INTRODUCTION
International migration, defined as the movement of people across international borders, is

emerging as an important development issue in the East Asia and Pacific region (EAP). East Asia
currently has an international migrant population of about 21 million people. More than a third of
these migrants in the East Asia region are regional migrants working in other countries within the
region. The wide disparities in income across the countries in the region make East Asia home to
both “labor receiving” (or destination) countries and “labor-sending” (or source) countries. Migrant
workers in the region remitted more than US $112 billion in 2013 to their home countries. These
remittance flows are huge not just for the small Pacific Island countries but also for the large
economies of Philippines and Vietnam.
Five stylized facts characterize international migration in the East Asia region. First, East
Asia represents its own migration hub. Five of the top 10 destination countries for migrants from the
East Asia region are also in the region – Hong Kong, Malaysia, Singapore, Thailand, and Republic of
Korea. East Asia serves as its own migration hub because of the large income differences between
labor receiving and labor sending countries in the region. For example, per capita incomes in the
richer labor receiving countries of Japan, Korea and Thailand are often ten times those of incomes in
the poorer, labor sending countries of Indonesia, Vietnam and the Philippines. Second, remittance
flows in the region are huge and expanding faster than those to the developing world as a whole.
Between 1989 and 2009, remittances to East Asia grew at an average annual rate of 15.8 percent vs.
7.8 percent for the developing world as a whole. Third, most migrants in East Asia are unskilled,
defined as those having a secondary school education or less. For example, 67 percent of the
migrants from Indonesia have less than a secondary school education, and about 70 percent of the
migrants from Vietnam fall into this category. Fourth, there is a high level of irregular and
undocumented migration in the region. In the early 2000s it was estimated that there were more than
1.6 million undocumented migrants from the Philippines, and that more than half of all Indonesians
migrating abroad were undocumented. The high level of irregular migration in the region reflects the
strong desire of migrants to escape the twin problems of poverty and unemployment in their home
countries. Fifth, females dominate migration flows in East Asia. In 2000, the number of female
migrants surpassed that of male migrants for the first time: 5 million females versus 4.9 million
males. These female migrants tend to work as domestic helpers in such countries as Saudi Arabia,
Malaysia, Hong Kong and Singapore.
The objective of this book is to analyze the economic and social impact of international
migration on labor sending and labor receiving countries in the East Asia region. More specifically,
the book seeks: (a) to examine the impact of international migration on key development indicators,
including poverty, investment, labor force participation, labor productivity and wages; (b) to
evaluate current government structures and institutions for managing migration, with a view to
identifying future policies for maximizing the benefits of international labor migration. The book
includes new work on these key policy issues from six East Asian countries: three labor sending
countries (Indonesia, Philippines and Vietnam) and three labor receiving countries (Singapore,
Malaysia and Thailand).
The twelve chapters in this book are divided into two parts. Those in Part 1, Migration and
Its Impacts, examine the empirical impact of migration and remittances in East Asia on various
development indicators and measures. These indicators include poverty; investment in education,
health and housing; labor force participation; child labor and education; and labor productivity and
iii
wages. These chapters focus on different East Asian countries, use data collected from large,
nationally representative studies, and employ different econometric tools. Their results, however, are
surprisingly consistent.
The chapters in Part 2, Managing Migration, analyze government and institutional efforts to
manage international migrant flows between countries in East Asia. These chapters focus on the key
question: how to devise effective migration policies that match migrant workers from labor sending
countries with the labor needs of labor receiving countries in a way that protects workers’ rights and
maximizes the contributions of migration to both sets of countries? The chapters in this part analyze
different East Asian countries, and use different methodological techniques. On the whole, these
chapters show just how difficult it is to implement effective and efficient migration policies, given the
inherent complexity of the migration industry, with its numerous stakeholders, governments,
employers and recruiting agencies.
In the next section of this overview, we review each of the twelve chapters in this book by
emphasizing their contribution to the migration literature, and highlighting the answers they provide
to ongoing debates in the East Asia region. However, before turning to these descriptions, it should
be emphasized that this volume focuses on economically-motivated migration in East Asia. While
there are certainly other types of international migration flows in the region, including those by
refugees and human traffickers, they are not covered here. Because relatively little has been written
to date on economically-motivated worker migration in East Asia, the choice was made to concentrate
the studies in this book narrowly rather than more broadly.
OVERVIEW OF THE CHAPTERS
In this section we provide an overview of the main findings presented in the chapters. Rather
than presenting the findings of each chapter individually, we examine how the findings relate to some
of the most important policy issues relating to migration and development in the East Asia region.
PART 1: MIGRATION, DEMOGRAPHY, POVERTY AND INVESTMENT
Over the next few decades East Asia faces major demographic changes as many countries’
labor forces will start to decline, while other countries will experience higher labor force growth as
populations and labor force participation rates increase. For example, the Japanese labor force has
already started to decline, and the labor forces of China, South Korea, Singapore, Taiwan, and Hong
Kong will begin shrinking by 2025. By contrast, the labor forces of Indonesia, Vietnam and
Cambodia will either increase or remain stable until 2025.
On the basis of these demographic trends, Walmsley et al in Chapter 1 show how a well-
managed migration strategy would represent a useful mechanism for reducing impending labor
shortages in some East Asian countries, while providing an opportunity for other countries in the
region to provide migrant workers that will contribute to their development through greater
remittance flows. Specifically, Walmsley et al show how a more flexible migration policy in East
Asia would be beneficial to most economies in the region in terms of real GDP or incomes over the
period 2007 to 2050.
As suggested by Walmsley et al, one of the most important development effects of

international migration is its direct impact on income and poverty in labor sending countries.
International migration usually results in remittances, and since remittances typically account for a
large share of household income in labor sending countries, most empirical studies find that
remittances tend to reduce poverty in the developing world.
In Chapter 2 Adams and Cuecuecha use a panel household data set from Indonesia to
examine the impact of international remittances on poverty. They find that the receipt of international
remittances reduces the probability of a household being poor by 28 percent.
iv
While the findings of Adams and Cuecuecha on remittances and poverty in Indonesia are in
line with other international evidence, their findings on remittances and investment are perhaps more
unique. In the more general literature, the question of how migrants spend and invest their remittance
earnings is a topic of much debate. Some studies find that international migrants spend most of their
remittances on consumption goods (food and consumer goods). However, other studies find that
households receiving remittances tend to spend them on investment goods (education and housing)
and that these patterns of expenditure can help build human and physical capital in developing
countries.
In Chapter 2 Adams and Cuecuecha find that households receiving international remittances
in Indonesia spend 332 percent more at the margin on education, and that these marginal expenditures
on education can help build human capital in the country at large. According to the authors,
households receiving remittances in Indonesia spend more at the margin on education because they
treat their remittances as transitory (rather than permanent) income, and the marginal propensity to
invest out of transitory income is higher than that for other sources of income.
These findings are supported by Cabegin and Alba in Chapter 3. Using nationally
representative data from the Philippines, Cabegin and Alba find that households receiving
international remittances spend 57 percent more at the margin on education. The authors also find
that households with remittances in the Philippines spend less at the margin on food and more at the
margin on housing. In other words, households receiving remittances in the Philippines are helping to
build both human capital (education) and physical capital (housing) in their country.
However, these findings on remittances and education are at least partially challenged by
Nguyen and Purnamasari in Chapter 4. Using panel household data from Indonesia these authors find
that international migration and remittances have no statistical impact on child school enrollment. In
other words, while Chapters 2 and 3 find that households receiving remittances spend more at the
margin on education, Nguyen and Purnamasari in Chapter 5 find that remittances have no impact on
school enrollment. One possible reason for these contradictory findings might be that primary school
enrollment rates are already over 90 percent in Indonesia.
In Chapter 4 Nguyen and Purnamasari also examine the impact of international migration on
labor force participation. They find that migration reduces the working hours of household members
left in Indonesia, but this is only true for households with male migrants. Households with female
migrants do not reduce their working hours. This result suggests that the gender of the migrant may
affect development outcomes in labor sending countries.
PART 1: MIGRATION, LABOR PRODUCTIVITY AND WAGES
In the literature the impact of international migration on labor productivity and wages is a
much debated issue. Most studies find that while international migration may reduce the wages
received by unskilled native workers in labor receiving countries, the decline is usually small, around
1 percent on average, and statistically insignificant.
In Chapter 5 Lathapipat examines the impact of low-skilled worker migration on the wages of
native workers in Thailand. Like other authors, Lathapipat finds that such migration has a small,
negative impact on low-skilled native workers. He also finds that low-skilled worker migration has a
much more negative impact on the wages of existing foreign workers in Thailand, and that migration
raises the productivity of high-skilled native workers with high-school and college education. In
Thailand, a doubling in the number of worker migrants causes the wages of low-skilled native
workers to decline by 0.03 to 0.79 percent, while it causes the wages of high-skilled native workers to
rise by 0.5 percent.
In Chapter 6 Yean and Siang come to a similar conclusion in Malaysia. Analyzing the impact
of migrant workers on the manufacturing sector in Malaysia, these authors find that the use of foreign
v
workers has a small, negative effect on both labor productivity and total wages for all workers (native
and foreign). However, the negative impact of foreign workers on labor productivity is smaller than
the negative impact on total wages for all workers. This suggests that the use of migrant workers has
helped improve the competitiveness and profitability of manufacturing firms in Malaysia by lowering
unit labor costs.
PART 2: MANAGING MIGRATION IN LABOR SENDING COUNTRIES
In the East Asia region strong supply and demand forces underlie the whole process of
international worker migration. On the one hand, workers in labor sending countries are driven by
supply pressures to migrate abroad: poverty, high unemployment and low wages. At the same time,
labor receiving countries have a high demand for foreign workers because of rapid economic growth,
tight labor markets and relatively high wages. These supply and demand forces create the need for
governments to try to manage or control migration flows in the region.
In the three labor sending countries in the region (Indonesia, Philippines and Vietnam)
governments typically try to manage migration by maximizing the outflow of migrants while making
(some) efforts to protect the rights of migrant workers abroad.
In Chapter 7 Kuncoro et al. analyze Indonesia’s institutional structure for managing the
migration of about 750,000 legal and 1.8 million irregular migrants. Most of these migrants find
overseas work through private recruiting firms, and in recent years the number of such firms has
increased from 100 to around 500. According to Kuncoro et al., growth in the number of recruiting
firms has affected the distribution of the rents generated by the migration industry. Now migrants
from Indonesia have to pay large sums to go work abroad, with perhaps half or two-thirds of their
payments going to government agencies or officials (legally or illegally). The size of these payments
causes many migrants to go to work abroad on an undocumented basis. While it may be cheaper to
work abroad illegally, undocumented migrants enjoy few, if any, rights abroad.
The Philippines is probably the East Asian country with the most experience in managing
migration outflows, and in Chapter 8 Tullao et al examine the institutional structure for managing the
1.4 million legal migrants from that country. In the Philippines a large government agency, the
Philippine Overseas Employment Administration (POEA), oversees the recruitment of migrants
through a large network of private recruiting firms. Most of these recruiting firms are small, and
specialized in particular occupations, like nursing. According to Tullao et al, since the expected
private returns to migration (and education) are very high in certain fields (like nursing), there is a
strong incentive for students to get educated in these fields. This in turn leads to a misallocation of
educational and other resources in the Philippines and – because not everyone can become a migrant -
- the creation of a large pool of underemployed people in the country.
In Chapter 9 Le and Mont analyze the governance structure for managing migration in
Vietnam. Vietnam is probably the smallest labor exporter in East Asia, with only about 0.5 million
legal migrants. Most of these migrants find overseas work through private service agencies. These
private agencies typically use rural brokers to find prospective migrants. According to Le and Mont,
these brokers represent a breeding ground for fraud, because they extract large rents from migrants.
The services offered by brokers are neither regulated nor supervised by the government, which is still
in the process of implementing procedures to protect the rights of migrant workers at home and
abroad.
PART 2: MANAGING MIGRATION IN LABOR RECEIVING COUNTRIES
In the three labor receiving countries in the region (Singapore, Malaysia and Thailand)
governments typically try to manage migration by using inflows of migrants to alleviate domestic
labor shortages and to maintain export competiveness while keeping an eye on preserving social
peace.
vi
In East Asia Singapore is probably the receiving country with the strictest controls for
managing migration flows. In Chapter 10 Teng analyzes Singapore’s two-pronged work pass system
for admitting foreign workers: one pass for skilled and professional workers, and the other pass for
unskilled workers. In Singapore the Ministry of Manpower uses this work pass system, along with
levies, dependency ceilings and source country restrictions, to encourage skilled worker migration
and to discourage unskilled migration. Despite these policy efforts, Teng finds that Singapore’s
reliance on unskilled foreign labor continues to rise: fully 80 percent of foreign workers with work
passes in Singapore are unskilled. Public concern about these rising numbers has led companies to
invest more in efforts to upgrade productivity and the skills of native workers.
Like Singapore, Malaysia has a system of levies and source country restrictions to try to
manage the inflow of foreign workers. However, as Kassim et al. describe in Chapter 11, these
controls have not been too effective at managing the estimated 1.9 million legal and 0.6 to 1.0 million
undocumented workers in Malaysia. According to Kassim et al., part of the problem is the large
number of government agencies and institutions involved in managing migration in Malaysia.
Another part of the problem is the long and porous border with Indonesia, which allows large
numbers of irregular migrants from that country to come and work in Malaysia. Efforts to regularize
these workers in Malaysia, or to impose strict penalties on employers who hire these workers, have
not been too successful.
As examined in Chapter 12 by Holumyong and Punpuing, Thailand also has a major problem
managing its 2.8 million migrant workers, over 50 percent of whom are irregular. As in Indonesia,
most undocumented workers in Thailand come from neighboring East Asian countries: Lao PDR,
Cambodia and Myanmar. According to Holumyong and Punpuing, the Thai government has tried to
manage these irregular workers through a system of worker registration, work permits and border
controls. Since 2007 the government has also signed Memorandums of Understanding (MOUs) with
neighboring countries to try to reduce illegal migration and to encourage legal migration. However,
time consuming processes and high costs have limited the successful implementation of these MOUs.
vii
Chapter 1: Labor Migration and Economic Growth in East and South-
East Asia
TERRIE WALMSLEY
Purdue Univerity
ANGEL AGUIAR, Purdue University, and S. AMER AHMED, The World Bank 1
ABSTRACT: East and South-East Asia face major demographic changes over the next few decades
as many countries’ labor forces start to decline, while other countries experience higher labor force
growth as populations and/or participation rates increase. A well-managed labor migration strategy
therefore represents a useful mechanism for ameliorating the impending labor shortages in some
East-Asia Pacific countries, while providing an opportunity for other countries with excess labor to
provide migrant workers that will contribute to their development through greater remittance flows.
On the one hand, migration will be unable to offset the economic impact of declining labor
forces in East Asian countries with shrinking populations. However, a more flexible migration policy,
allowing migrants to respond to the major demographic changes occurring in Asia over the next 50
years, would be beneficial to most economies in the region in terms of real incomes and real GDP
over the 2007-2050 period. Such a migration policy could deeply affect the net migration position of
countries in the region. Countries that were net recipients under current migration policies might
become net senders under the more liberal policy regime.
1. INTRODUCTION
The East Asia and Pacific (EAP) region experienced sharp demographic changes between
1965 and 1990, with declining mortality and fertility rates contributing significantly to rapid
economic growth (Bloom and Williamson, 1998). While these rapid demographic changes are
projected to persist in the future, their impact will not be as positive as they have been in the past. The
changing demographics of Asian countries will create labor shortages in some countries, while other
Asian countries will experience large expansions in their labor force.
A well-managed labor migration strategy thus has the potential for ameliorating impending
labor shortages in some EAP countries, while providing an opportunity for other countries with excess
labor to provide migrant workers that will contribute to their development through greater remittance
flows.
This study examines the potential impact of such a labor migration strategy on the economies
of East and South-East Asian countries. The potential effect of increased labor migration between
EAP countries is analysed using a global dynamic simulation model, with migrant labor flows and
remittances used to examine the impact of migration.
1 Terrie Walmsley is Associate Professor and Director of the Center for Global Trade Analysis, Purdue University and
Principal Fellow at the University of Melbourne, Australia (email: twalmsle@purdue.edu). Angel Aguiar is a Research
Economist of the Center for Global Trade Analysis, Purdue University (email: aaguiar@purdue.edu). S. Amer Ahmed is an
Economist in The World Bank’s Development Prospects Group, World Bank (email: sahmed20@worldbank.org). This
paper was prepared for the “Cross Border Mobility and Development in the East Asia and Pacific Region” project of the
World Bank’s Office of the Chief Economist, East Asia and Pacific Region, under the guidance of Ahmad Ahsan (Lead
Economist, EASPR). The authors are also grateful to the participants of the June 2010 World Bank and Institute for Policy
Studies conference on “Cross Border Mobility and Development in the East Asia and Pacific Region”, Ahmad Ahsan, and
an anonymous referee for many helpful comments and feedback.
1
2. LITERATURE REVIEW
There is a large and growing economic literature on migration and its impacts, going back to
Samuelson’s (1964) discussion of the implications of immigration into the USA on labor supply and
wages. This section will review a few broad segments of this literature.
Drivers of Migration
Contemporary theories of international migration (see Massey et al., 1998) suggest that
people move because of expected improvements elsewhere, where improvements might include
higher wages, employment, health, and education. From an economic perspective, migration occurs
because of expected differences in wages or income. This is the basis of the Harris – Todaro model,
which explains the migration decision as being based on expected income differentials between rural
and urban areas (Harris and Todaro, 1970). In this model, migration from rural areas to urban areas
occurs if the rural wage rate (i.e. the rural marginal productivity of labor, MPL) is less than the urban
wage rate (i.e. urban MPL) times the urban employed-labor force ratio (i.e. ratio of total employed to
total employed plus seeking jobs in the urban area).
The applicability of this theoretical framework is supported by broader empirical analyses.

For example, Pissarides and McMaster (1990) found that inter-regional migration within the UK
responds to changes in regional relative wages and to differences in employment opportunities. In the
context of international migration, Hanson and Spilimbergo (1999) explain that high wage
differentials between the USA and Mexico have traditionally been the cause of migration to the USA.
Demographic factors can also represent another economic force. If there is an excess supply
of labor (relative to domestic demand and employment opportunities) in a labor sending country, then
there is an incentive for those workers to move to markets where there is an excess demand.
Authukorala’s (2006) review of East Asian migration provides several examples of this in the context
of undocumented migrant workers from high-population countries moving to Thailand.
Economic Impacts of Demographic Change
Most analyses of the coming changes in Asian labor forces due to demographic changes focus
on the implications of an aging population. For example, a seminal study by Bloom and Williamson
(1998) found that rising working-age shares in the population – and rising labor forces –increased per
capita income while holding output per worker constant.
Updating the findings of Bloom et al. (2000) to the period 1960-2005, Bloom and Finlay
(2009) find that the working age share and labor force growth rates remain important contributors to
economic growth in Asia. Columns I and II of Table 1 demonstrate the importance of these two
population variables for economic growth. In the case of Singapore, for example, the labor force
growth rate contributed to 2.21 percent of average economic growth between 1965 and 2005, with the
growth of the working-age share of the population contributing to more than half of average growth in
the period. However, when the study applies projected growth of the working age shares and
population to determine their contribution to future economic growth, it is found that with the
exception of Malaysia and the Philippines, the growth rates of all East and South-East Asian countries
become negative as a result of the demographic changes (column III, Table 1).
2
Table 1: Contribution of Demographic Change to Average Economic Growth in Asia
(Percentage Points)
Contribution to Economic Growth from Growth of
Labor Force and Working-Age Share Working-Age Share
Population (1965-2005) and Population (1965- and Population (2005-
2005) 2050)
I II III
China 0.91 16.46 –0.36
Japan 0.35 9.53 –0.91
South Korea 2.01 36.4 –0.87
Singapore 2.21 51.13 –0.78
South-East Asia
Indonesia 1.19 41.01 –0.05
Malaysia 1.12 26.71 0.13
Philippines 0.53 36.18 0.46
Thailand 0.88 20.81 –0.45
Bangladesh 0.15 9.53 0.42
India 0.01 0.27 0.36
Nepal –0.63 –50.32 0.66
Pakistan –0.10 –3.90 0.73
Sri Lanka 1.15 32.3 –0.26
Source: Reproduced from Bloom and Finlay (2009)
Given the potentially large differences in labor force growth rates between countries in East
and South-East Asia, the migration of workers into countries that will soon experience declining labor
forces presents itself as a potential policy response to the coming demographic transitions. In recent
years there have been a number of studies on the role of migration in meeting such demographic
transitions in Asia and other regions of the world.
Tyers and Shi (2007) explore the impact of migration on growth by simulating a policy
scenario where Western Europe, North America, and Australia respond to declining labor forces by
allowing for sufficient migration from the rest of the world to hold non-working age dependency
ratios constant from 2000 onwards. They find that even though labor receiving countries are able to
maintain high growth rates, the rest of the world experiences substantial welfare losses.
Back-of-the-envelope estimates from Rodrik (2004) and Winters (2001) indicate that even
modest liberalization of temporary migration from the developing to developed economies can lead to
substantial global welfare gains. Similarly, Hanson’s (2008) review of the empirical literature on
migration finds that liberalizing international migration is generally beneficial to expanding global
output. Hanson’s review also finds that while international migration undoubtedly has wage impacts
on the receiving countries, these wage effects are not well-captured by many statistical analyses.
Hanson’s review suggests that this shortcoming is best addressed through global general equilibrium
(GE) analyses.
Computable GE simulations by the World Bank (2006), and Walmsley et al. (2009), predict
that greater liberalization of labor movement from the South to the North would lead to global welfare
increases. In these studies, the migrant-sending, less developed countries receive large shares of the
welfare gains.
3
Migration and Migration Policy in Asia
Within the East Asia region, Manning and Sidorenko (2007) point out that intra-regional
liberalization of skilled worker migration would address the growing phenomena of skill shortages
and surpluses in the same occupations across Asian countries. The study cites Singapore as an
example, where there has been growing excess demand for healthcare professionals, managers,
accountants, and engineers. At the same time, there is evidence that neighbouring Indonesia and the
Philippines have surpluses in several of these professions. A policy framework that would encourage
intra-regional migration between Singapore, Indonesia and the Philippines could therefore have
tremendous potential welfare gains for all countries concerned.
Over time net migration – i.e. the difference between migrants entering a country and
migrants leaving a country – within East Asia has generally increased (Table 2). Indonesia is the
most notable exception, with a general fall in migration rates, and Malaysia’s migration has been
variable with a large increase in the 1960s, followed by a small decrease in the 70s. Over the entire
period Vietnam has recorded the largest increase in migration rates.
Table 2: Percentage Change in Net Migration* per Annum

1960-1970 1970-1980 1980-1990 1990-2000 1960-2000 1980-2000
China 1.3 -15.1 5.2 9.0 -0.3 7.1
Hong Kong 0.5 1.9 0.7 1.9 1.2 1.3
Indonesia -4.5 -4.5 -4.5 -10.7 -6.1 -7.7
Japan 0.5 1.0 2.9 4.6 2.2 3.8
Malaysia 29.2 -0.9 3.5 4.7 8.5 4.1
Philippines -0.1 -5.6 1.1 9.0 1.0 5.0
Singapore 0.2 -0.1 3.3 6.4 2.4 4.8
South 2.9 11.4 0.8 -0.1 3.6 0.4
Korea
Thailand -3.3 -2.4 0.5 9.1 0.9 4.7
Vietnam 1.0 1.0 4.1 18.7 6.0 11.2
Source: Özden et al. (2011)
Note: *Net migration refers to migration into a country minus migration out of a country.
3. ANALYTICAL FRAMEWORK
Our analytical framework involves applying a dynamic global general equilibrium model
(GMig2Dyn) to simulate a projected growth path of the world based on current best-estimates of
population, real GDP and labor growth from international institutions over the period 2007-2050.2 As
part of this projected growth path, we simulate more liberal international migration policies that allow
the labor-force shrinking Asian countries to import labor from those Asian countries where the labor-
force is expanding.
Simulation Modelling
The dynamic migration model (GMig2Dyn) is based on the Dynamic GTAP (GDyn) model
developed by Ianchovichina and McDougall (2001) and the bilateral migration model (GMig2),
developed by Walmsley et al. (2009). Both the GDyn and GMig2 models are based on the GTAP
standard general equilibrium model. The standard GTAP model is a comparative-static general
equilibrium model of the world economy (Hertel 1997).
In the standard GTAP model, capital can move between industries within a region, but not
across regions. The GDyn model extends the standard model by incorporating international capital
2
We show results from 2007, but the GTAP 7 Data Base has a base year of 2004 so we had to update it first to 2007.
4
mobility and capital accumulation. Furthermore, GDyn takes account of foreign income flows and
wealth, by keeping track of both the ownership and location of capital assets. In the GDyn model,
international capital mobility is modelled using a disequilibrium approach. GDyn assumes an adaptive
expectations mechanism that permits errors in expectations. These errors in expectations are
gradually eliminated, and rates of return on investment gradually equalize across regions, resulting in
a gradual movement of economies towards steady state growth.
The GMig2 model extends the GTAP model to consider skilled and unskilled bilateral labor
movement across countries, and their impact on growth, remittances and the real incomes of migrants
and permanent residents. The bilateral nature of the GMig2 model allows us to analyse the effect of
changes in the receiving country’s immigration policy, targeting particular migrant sending countries.
The movement of labor of type i from region c to region r (i.e. changes in labor force which are
changes in LFi,c,r) can be determined exogenously, for example through changes in quotas, or
endogenously in response to changes in relative real wages. Where migration occurs endogenously,
workers (or labor supply) are assumed to respond to changes in the expected real wages between the
sending (RWi,c,c) and potential receiving (RWi,c,r) region according to equation (1).
ESUBMIGi,c, r
 RWi,c,r 
LFi,c,r  A i,c,r    (1)
 RWi,c,c 
Ai,c,r is a coefficient which takes into account other factors in the migration decision (e.g.,
language, distance etc.) and is calibrated from the underlying GMig2 Data Base. ESUBMIGi,c,r is a
parameter reflecting the extent to which migration responds to changes in the relative expected real
wages and is set to 1 in this paper. The extent to which migration is endogenous is dependent on this
parameter. Increasing this parameter increases the number of migrants moving, but does not change
the directions of their movements.3 A low parameter value means that the ability of migrants to
respond to changes in real wage differentials is limited, due to excessively high costs associated with
such movements. Such costs could include difficulties finding a job in the host economy or distance
from ones family.
Note that Equation (1) is calibrated on actual data and incorporates the current state of
restrictions on migration in the receiving country.4 Figure 1 is used to further explain this. It is
assumed that demand and supply of migrants are equal and hence the labor market is in equilibrium
(see Figure 1). The labor supply curve is upward sloping – as wages available to migrants in the
receiving region rise relative to those in the sending region, migration to the receiving country
increases. When there are no restrictions on migration imposed by the receiving country, the
equilibrium is represented by point A in Figure 1 where demand equals supply (Ls=Ld). Equation (1)
is the labor supply curve.
3
The authors also tested the model with ESUBMIG=0.4.
4
Given ESUBMIGi,c,r=1.
5
Figure 1: Demand and Supply of Migrants (Equation 1)
Turning now to the more likely case of restrictions on migration, the dashed line in Figure 1 is
used to depict the situation where a quota (or cost to migrants, either implicit or explicit) has been
applied to restrict the number of migrants. In this case migration is lower than it would be without the
quota, and the wage paid by firms (Point B, determined by Ld) is much higher than the wage required
by the migrant worker (Point C, determined by Ls). The difference between B and C is the rent or cost
of the quota induced by the migration restrictions. There are a number of alternative agents who are
likely to share this rent, including 1) migration agents in the sending or receiving country who charge
fees for obtaining visas and/or finding the migrant a job; 2) the receiving country employer who could
pay migrant workers lower wages, keeping the rent for themselves; 3) the receiving government
through charges for visas or additional taxes on migrants; 4) the sending country government, if
agreements have been made between governments for transfers (for example, to pay for education
expenses); or 5) the migrant worker themselves. In this case we assume that all of the rent is earned
by the migrant worker.5
We argue that this quota reflects the status quo in the receiving economy and the preferences
of its people and firms for migrant workers. This means that the current equilibrium is the point at
which the preferences of the incumbent population for migrants are exactly balanced against the
firms’ desire for more workers. Rather than eliminating this quota altogether, we assume that as
populations, and in turn labor forces, change and economies grow, the receiving economies will adjust
their quotas in response to firms’ demands for more labor. The resulting labor supply curve is given
by Ls1, and is higher than the original supply curve (Ls). As mentioned above, this is due to the fact
that the initial rents, which reflected the preferences of the country towards migration, remain in
place. Equation (1) therefore represents this derived labor supply curve (Ls1 in the case of
restrictions). The implication of this is that only changes in the expected relative wages will drive
new migration. If you want migrants to respond to the initial differences in wages, i.e., the initial
restriction depicted by the value of the rent, then this rent would have to be removed so as to move the
equilibrium back to point A.
5
Due to data limitations it is difficult to ascertain the value of this rent, and by making this assumption we avoid the need to
calculate its value. Given we do not reduce this rent during the simulations this has assumption has minimal effect on our
results. If this rent were to be reduced, it would be important to clearly allocate this rent appropriately.
6
In summary, the dynamic migration model (GMig2Dyn) therefore features:
a) the accumulation of capital over time;

b) the ownership of capital and the income flows to those capital owners;
c) the movement of migrants and other changes in the labor force over time;
d) the flow of remittances back to the families of the migrants; and
e) the real incomes of migrants and permanent residents.
The model also separately identifies domestic and foreign workers by sector of employment.
Foreign and domestic workers of the same skill type are treated as imperfect substitutes, but there is
no distinction between foreign countries. That is, firms demand foreign workers without regard to
their country of origin (Aguiar, 2009).
In addition, this version of the model also includes unemployment of endowments (capital,
skilled and unskilled domestic and foreign labor) through the inclusion of an elastic segment in the
previously inelastic labor supply curve.6 This is achieved through a complementarity which sets
employment equal to the natural rate of employment, unless a fall in demand is sufficient to drive the
expected real wages down by more than a threshold rate of change. In the next period, the
employment rate will attempt to move back to the natural rate, but this will only be achieved if
demand is sufficient to return the economy to the natural rate without further lowering wages more
than the threshold rate of change. Provided the economy does not continue to be hit by negative
shocks, employment is expected to gradually move back to the natural rate of employment. This
allows us to capture unemployment resulting from the global financial crisis.7
In Equation (1) the real wage is adjusted to take account of the probability of employment
once the migrant arrives in the host region. Real wages (RWi,c,r) are calculated as the average
expected real wage of the entire labor force, regardless of employment status; hence the expected real
wage that a migrant faces once the migrant arrives in the receiving region includes unemployment. If
unemployment in the receiving economy rises then the expected real wage falls, due to a higher
probability of unemployment, and hence migration falls. This reflects a main assumption of the
Harris-Todaro model that the migration decision between rural and urban areas is based on expected
income differentials, rather than just wage differentials.
Underlying the GMig2Dyn model is a database that captures both bilateral labor (GMig2
Database) and foreign ownership of capital (GDyn Database); as well as the core GTAP 7 Data Base
(Narayanan and Walmsley, 2008). For this study, we have updated the GMig2 database to 2004, the
base year of the GTAP 7 Data Base. The bilateral migration data is based on a new migration data
base by Özden et al. (2011) and the remittances data were also updated, using the IMF's balance of
payments statistics on remittances and workers compensation.
Further adjustments are also made to the database in order to improve the distribution of
unskilled foreign workers across sectors. Estimates of the number of unskilled foreign workers by
industry were obtained for Malaysia, Korea, Singapore, and Thailand from Yean and Siang (2010);
Hur (2010); Teng (2010); and Holumyong and Punpuing (2010) respectively, see Table 3. For other
ASEAN countries – Indonesia, Philippines, and Vietnam – we use the average distribution of
unskilled foreign workers by industry of the aforementioned countries as a proxy. For all other
countries foreign and domestic workers are assumed to be allocated across sectors in the same
proportions according to the underlying total data available in the GTAP Data Base. The choice to
redistribute unskilled workers by industry for Indonesia, the Philippines and Vietnam, but not for the
6
For simplicity of the diagrams, we assume that labour supply is fixed. In the model there will be some interaction between
total labour supply in an economy and the demand and supply of migrants, which responds endogenously to real wage
changes and unemployment, hence the total labour supply curve is slightly upward-slopping.
7
See Strutt and Walmsley (2010) for more details on how unemployment was incorporated.
7
other economies, reflects the fact that Asia is the focus of this paper, and there is reason to believe that
the sectoral relationships in the other Asian economies exist across all Asian economies. ASEAN thus
tends to use unskilled foreign workers more intensely in agriculture and food processing; while
Singapore and South Korea use unskilled foreign workers in food processing, light and heavy
manufactures, construction and services, in some cases.
Table 3: Distribution of unskilled foreign workers across sectors in ASEAN countries

Manufacturing Construction Services Agriculture
Malaysia 35.12 14.79 24.74 25.35
South Korea 41.61 22.30 34.16 1.93
Singapore 25.00 26.00 49.00 0.00
Thailand 15.39 16.75 31.31 36.55
Average (Indonesia,
27.88 19.01 33.15 20.26
Philippines and Vietnam)
Sources: Yean and Siang (2010); Hur (2010); Teng (2010); and Holumyong and Punpuing (2010)
Scenarios
The model simulates the world economy from 2007 to 2050, under migration policies that
allow labor to move freely within the EAP region in response to changes in real wages. These labor
movements are in addition to, and the result of, the demographic and labor force changes that are also
projected to occur. The results are then decomposed to examine the impact of the regional migration
liberalization on the world economy over time, as well as the other changes in the world economy.
The analysis takes into account the impact of the demographic changes expected to occur in
Asia over the next 43 years, from 2007 to 2050. It includes population and labor force forecasts by
skill that are based on the World Bank’s and the UN’s World Population Prospects 2008 Revision.
The scenario also tracks actual and expected future changes in Real GDP from 2007 to 2012. Beyond
2012 calibrated technological change is assumed to persist (declining gradually) to 2050. Between
2007 and 2012 additional assumptions are made to take account of the global financial crisis,
including additional adjustments to investment through a rise in errors in expectations, unemployment
of labor and capital, government spending and a negative productivity shock aimed at capital,
following Strutt and Walmsley (2010). After 2011 the financial crisis gradually comes to an end and
unemployment falls back to pre-crisis levels.
Ignoring endogenous migration for the time being, both the domestic and foreign labor forces
are assumed to grow at the projected growth rate in total labor. Increases/decreases in the domestic
labor supply are implemented as changes in the natural rate of population growth, while foreign labor
increases are implemented through changes in migration.8 The change in foreign labor keeps the
share of migrants in the total population constant.9 Since these labor forecasts are meant to include
migration we assume that foreign labor grows at the same rate as the domestic labor force. This
means that a country is willing to exogenously increase migration (increase the quota) to keep the
migrant shares constant. The impact of including forecasts is discussed below – these are referred to
as the “forecast” results. Next we liberalize migration by allowing it to respond to wage changes –
these results are labelled “liberal” below and are also referred to as ‘endogenous migration’ in the
text.
Thus migration in this scenario results from two sources: first, we assume that a country is
willing to exogenously increase migration to keep the migrant shares constant (forecasts); and second,
8 We assume that children born to migrant workers are included in the natural population and that all new migrant workers
come from abroad.
9
With the exception of Japan and South Korea where we assume that migrant growth is zero. Hence when combined with
negative growth of incumbent labour, migrant shares increase slightly.
8
we assume that migration is liberalized so that migrants can respond endogenously to changes in the
real wages in the sending and receiving economies (liberal). These two sources of migration may
work in opposite directions. For instance, an economy with a growing population will increase the
number of migrants to keep the share of migration in its total population constant; on the other hand,
the same country may experience declining wages due to its rising population which will cause
outward migration of its own population and return migration of its foreign population, thereby
negating the exogenous increase in migration.10
4. SIMULATION RESULTS
In this section we examine the results of the scenario in terms of the projected demographic
changes, changes in bilateral migration in response to the changes in wages resulting from the
demographics and the impact of these on the EAP economy.
Demographic Changes and Real Wages
Figures 2 and 3 show the yearly forecasted growth rates of skilled and unskilled labor forces
obtained from the World Bank and United Nations. According to these data there is an unmistakable
downward trend in skilled labor growth rate in all countries, with forecasted growth rates becoming
negative or falling to almost zero in all of developed economies of Asia –Hong Kong, Japan,
Singapore, South Korea, and Taiwan – as well as in China and Thailand. The forecasted unskilled
labor growth, on the other hand, does not decline as significantly as skilled labor, although it is
negative or close to zero for the same seven economies –China, Hong Kong, Japan, Singapore,
Taiwan, Thailand, and South Korea.
Figure 2: Forecasted Annual Growth Rates in Skilled Labor by Region

9
Philippines
Malaysia
7
Indonesia
Rest S. E. Asia
Yearly Growth Rates
5
Rest E. Asia
Vietnam
3
China
South Korea
1
Thailand
Taiwan
-1
Singapore
Hong Kong
-3
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Japan
Source: Authors’ results
10
The reader should therefore not interpret that these large decreases in migration or return migration are large flows of
people going back to their home economies, instead they are migrants choosing not to migrate in the first place or choosing
to migrate elsewhere instead.
9
Figure 3: Forecasted Annual Growth Rates of Unskilled Labor by Region
2.5
Philippines
2 Malaysia
Rest E. Asia
1.5
Indonesia
Yearly Growth Rates
1 Rest S. E. Asia
Vietnam
0.5
China
0 South Korea
Thailand
-0.5
Taiwan
-1 Singapore
Hong Kong
-1.5
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Japan
Figures 4 and 5 show the cumulative percentage changes over time of the real factor prices of
skilled and unskilled labor resulting from these forecasted demographic changes in the EAP region.
Hong Kong stands out as an economy with the largest changes in real wages as a result of the
demographic changes. This is not surprising given the low growth in skilled and unskilled workers,
combined with the high accumulation of capital and forecasted changes in real GDP. The figures
show that real wages in Hong Kong are 350% higher in 2050 than they were in 2007 for skilled
workers and 600% higher for unskilled workers. The next highest increases are in the Rest of East
Asia and Thailand where real wages rise by 200% relative to 2007. As expected, six of the seven
economies – China, Hong Kong, Japan, Taiwan, Thailand, and South Korea – highlighted above for
having the lowest growth in skilled and unskilled labor, are also those the economies where factor
prices rise consistently over time; while the other regions experience much slower growth in real
wages. One such a case is Singapore, where despite low/negative growth in skilled and unskilled
labor, the growth rate of wages is slower than in Hong Kong or even Thailand. This is explained by
the slower GDP growth forecasted for Singapore when compared to the rapid GDP growth forecasted
for Hong Kong and Thailand.
10
Figure 4: Cumulative Percentage Changes in Real Wage of Skilled Workers due to Forecasted
Demographic changes
350
Hong Kong
300 Japan
Rest E. Asia
250
Cum % changes in 2050
Thailand
200 Taiwan
Vietnam
150
Rest S. E. Asia
100 China
Philippines
50
South Korea
0 Singapore
Malaysia
-50
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Indonesia
Figure 5: Cumulative Percentage Changes in Real Wages of Unskilled Workers due to

Forecasted Demographic changes
600
Hong Kong
500 Thailand
South Korea
400
Cum % changes in 2050
Japan
Philippines
300
China
200 Vietnam
Taiwan
100 Rest S. E. Asia
Singapore
0
Malaysia
Rest E. Asia
-100
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Indonesia
11
The Impact of the Liberalization of Migration Policies on the Labor Force and Migration
Table 4 (next page) presents an overview of the changes in the labor force, by skill type and
by country. The total change in the labor forces are decomposed into two main components: forecasts
and changes due to liberal migration policies. The forecasts can be further divided into changes due
to the natural rates of changes in permanent residents and our assumption that migration increases
with population forecasts. The liberal migration policies component can be further divided into
changes in changes in the number of migrants located in the country/region and changes in permanent
residents through outward or return migration. The labor forces of all countries, except for Japan,
increase over time. The decrease in Japan is due to the fact that the demographic changes occurring in
this country are not offset by migration. China’s and Indonesia’ labor forces increase the most, due to
positive natural population growth over the period and high initial populations. These changes are the
result of the forecasts discussed above.
12
Table 4: Decomposition of the Changes in Labor Force by Country between 2007 and 2050
(Millions of People)
Total Forecasts Liberal Migration

Change in
Change in
Permanent
Permanent
Change in Labor Residents Change in
Change in Residents
Force (Natural growth Migrants
Migrants (Return or
due to
Outward
births/deaths)
Migration)
Unskille Skille Unskille Skille Unskille Skille Unskille Skille Unskille Skille
d d d d d d d d d d
I II III IV V VI VII VIII IX X
China 157.0 31.9 158.2 31.6 0.02 0.04 -0.01 -0.02 -1.20 0.29
Hong Kong 1.0 0.8 -0.5 0.1 0.01 0.41 1.46 0.25 0.00 0.02
Indonesia 85.0 29.1 85.0 28.7 0.03 0.11 -0.04 -0.10 0.02 0.43
Japan -13.1 -10.1 -13.1 -10.2 0.00 0.00 -0.01 0.07 0.00 0.02
Malaysia 9.7 6.3 9.6 6.0 0.35 1.62 -0.21 -1.30 -0.02 -0.05
Philippines 46.6 10.9 46.4 10.5 0.12 0.26 -0.04 -0.14 0.05 0.33
Rest E. Asia 9.4 0.7 9.5 0.7 0.08 0.03 0.00 0.00 -0.14 0.06
Rest S.E.
24.9 4.7 24.9 4.5 0.10 0.16 0.00 0.00 -0.05 0.06
Asia
Singapore 0.2 0.1 0.2 0.0 0.06 0.07 -0.06 -0.01 -0.04 0.08
South
-0.6 3.7 -0.7 3.7 0.02 0.08 0.09 -0.07 0.04 -0.04
Korea
Taiwana 1.3 -0.1 1.3 -0.1 NA NA NA NA 0.00 0.02
Thailand 2.8 5.2 2.6 5.0 0.02 0.16 0.11 -0.03 0.05 0.11
Vietnam 38.6 2.9 38.6 2.9 0.01 0.01 -0.01 0.00 0.00 0.03
Note: a. no data on migrants are available for Taiwan.
In this model, it is the changes in the real wages occurring as a result of the financial crisis,
economic growth and demographic changes that drive the endogenous migration. Migration between
two countries depends on the changes in the real wages in both the receiving and the sending country
and any changes in unemployment (Equation 1). Moreover, this model depends on a database that
only captures migrants recorded in the census – illegal migration is therefore likely to be under-
reported.
Hong Kong’s labor force increases the most as a result of more liberal migration policies,
with 1.72 million more workers, of which 1.71 million are new migrant workers and 0.02 returning
migrants. China and Malaysia have the largest decreases in their labor forces as a result of liberal
migration, with 0.95 and 1.57 million fewer workers respectively. The implication is that migrants
that would have otherwise moved to China or Malaysia (as in the baseline), would rather migrate to
Hong Kong when given the opportunity (as under endogenous migration as a function of changes in
expected real wages).
Under the more liberal migration framework, China, Malaysia, Indonesia, the Philippines, and
Singapore receive fewer migrant workers by 2050, although the overall impact on the labor force (in
terms of people) in Indonesia and the Philippines is positive due to return migration. Skilled workers
in China and Singapore also increase for the same reason, return migration of skilled workers
(Column X in Table 4). The liberal migration policies also cause the return migration of people from
Hong Kong and an increase in outward migration of Chinese and Malaysian residents.
13
Figures 6 and 7 compare the forecasted growth in skilled and unskilled labor with the growth
in skilled and unskilled labor under more liberal migration for selected countries, respectively, over
time. Looking at Figure 6, this comparison highlights several interesting points.
 First, endogenous migration does not alter the growth of the labor force significantly in most
economies. Only in Hong Kong, where migration is highest, does the growth rate of labor rise
by just over 1%. This is because migrants are generally a small share of the labor force in
most Asian economies.
 Second, another interesting feature of endogenous migration is the extent to which migrants
react to the global financial crisis. During the global financial crisis, real GDP and real wages
fall, while unemployment rises. The combination of falling real wages and higher
unemployment in the host economy generally reduce migration flows under the endogenous
migration scenario, albeit the direction can be sometimes unclear if wages and employment
also fall in the home country. Prior to 2020, Figures 6 and 7 show clear evidence of some
temporary declines in migration (or increased return migration) due to the financial crisis –
only migration to East Asia increases. Labor growth with endogenous migration is therefore
generally below forecasted labor growth during the financial crisis (2007-2012) – in 2012, the
number of migrants globally was 0.82 million lower than under the forecast scenario.
 Third, once the effects of the global financial crisis have dissipated (usually between 2012
and 2020), the demographic effects come into play and migration towards those Asian
economies experiencing lacklustre population growth becomes more evident. Overall,
migration due to the financial crisis is temporary and does not affect the demographic story.
Figure 6 indicates that if migrants are free to move in response to wages, Hong Kong and
Singapore would have mostly increasing growth rates of skilled labor force after 2020,
relative to the forecasts. For Malaysia and the Rest of East Asia the model predicts only a
marginally smaller growth rate of skilled workers than originally forecasted.
Figure 7 compares forecasted growth with endogenous growth for unskilled migrant labor,
with similar conclusions to those found with skilled labor in Figure 6. The growth of unskilled labor
with endogenous migration would be higher than forecasted unskilled labor for Hong Kong after the
crisis; prior to the crisis growth rates are lower, but recovery is quick with growth exceeding forecasts
by 2011. In Singapore, endogenous migration causes a larger decline in the growth rate of unskilled
labor relative to the forecasts until 2015. After 2015, the unskilled labor force growth rate recovers
quickly, but it is not until 2035 that the growth rate under endogenous migration surpasses the
forecasted growth. As with the skilled labor force, Malaysia's unskilled labor force growth would be
slower than originally forecasted when considering endogenous migration, albeit the differences are
minimal (Figure 7).
14
Figure 6: Forecasted versus Endogenous Migration Annual Growth Rates of Skilled Labor for Selected Countries
Hong Kong Singapore

5 2.5
4 2
1.5
3
Endogenous 1 Endogenous
2 Migration Migration
0.5
1 Forecasted Skilled Forecasted Skilled
0
Labor Labor
Yearly Growth Rates

Yearly Growth Rates
0 -0.5
-1 -1
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
Malaysia Rest E. Asia

9 2.5
8
7 2
6
Forecasted Skilled 1.5 Forecasted Skilled
5
4 Labor Labor
1
3 Endogenous Endogenous
2 Migration 0.5 Migration
Yearly Growth Rates

Yearly Growth Rates
1
0 0
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
15
Figure 7: Forecasted versus Endogenous Migration Annual Growth Rates of Unskilled Labor for Selected Countries
Hong Kong Thailand

1 0.6
0.5 0.4
0 Endogenous 0.2 Endogenous

Migration Migration
-0.5 0
Forecasted Unskilled Forecasted Unskilled
-1 Labor -0.2 Labor
Yearly Growth Rates

Yearly Growth Rates
-1.5 -0.4
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
Singapore Malaysia
2 2.5
1.5 2
1 Endogenous 1.5 Forecasted Unskilled

Migration Labor
0.5 1
Forecasted Unskilled Endogenous
0 Labor 0.5 Migration
Yearly Growth Rates

Yearly Growth Rates
-0.5 0
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
2007
2011
2015
2019
2023
2027
2031
2035
2039
2043
2047
16
Changes in the labor force depend on the changes occurring within the bilateral migration
corridors, and the relative productivities of those migrants. Figures 8A, 8B, 9A, and 9B describe the
unskilled and skilled labor migration corridors that expanded or contracted the most over the period.
These changes in the migration corridors are the result of both the exogenous changes in forecasts and
the endogenous migration due to more liberal migration policies.
The largest changes in bilateral migration are the outflow of Chinese migrants from Singapore
and Indonesia into Hong Kong in response to the increase in real wages in Hong Kong relative to
those in Singapore and Indonesia. This is the result of the very large increases in real wages in Hong
Kong and the large number of Chinese already living in Hong Kong. Migrants from Rest of East Asia
are also entering Hong Kong and South Korea in response to the higher wages, while Chinese
migrants also enter the Rest of East Asia11 to fill the gap left by migrating East Asians. Unskilled
South Koreans also return home from Japan as relative wages at home rise, as do skilled Filipinos.12
Thailand, which was also experiencing large increases in real wages in response to
demographic changes, receives large increases in migrants from South East Asia. Even though real
wages in Malaysia decline relative to many of the other countries and migrants from the Philippines
and Thailand return home, they are still able to attract migrants from Indonesia, where growth in the
labor force continues to be robust over the entire period. Moreover many unskilled Malaysian
migrants return home from Singapore, and skilled Singaporean return home from Malaysia.
Figure 8A: Unskilled Worker Migration Corridors in EAP that Expanded the Most between
2007 and 2050 (Excluding Hong Kong / China).a
0.35
0.3
0.25
Millions of People
0.2
0.15
0.1
0.05
0
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Malaysia / Indonesia South Korea / Rest of East Asia

Thailand / Rest of SE Asia Rest of East Asia / China
Hong Kong / Rest of East Asia
Note a. Number of unskilled migrant workers by location/home region
11
The inflow of Chinese into East Asia is primarily due to the forecasted increase in the labour supply in East Asia, rather
than as a result of more liberal immigration policy.
12
Skilled real wages of Filipinos increase faster than those in Japan up until 2040 (see Figure 4).
17
Figure 8B: Skilled Worker Migration Corridors in East Asia and the Pacific that Expanded the
Most between 2007 and 2050. a
0.6
0.5
0.4
Millions of People
0.3
0.2
0.1
0
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Malaysia / Indonesia South Korea / Rest of East Asia
Thailand / Rest of SE Asia Rest of East Asia / China
Hong Kong / Rest of East Asia
Note: a. Number of unskilled migrant workers by location/home region
Figure 9A: Unskilled Worker Migration Corridors in East Asia and the Pacific that Declined
the Most between 2007 and 2050. a
0.25
0.20
Millions of People
0.15
0.10
0.05
0.00
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Malaysia / Philippines Malaysia / Thailand Singapore / Malaysia

Singapore / China Japan / South Korea
18
Figure 9B: Skilled Worker Migration Corridors in East Asia and the Pacific that Declined the
Most between 2007 and 2050 (excluding Singapore / China). a
0.14
0.12
0.1
Millions of people
0.08
0.06
0.04
0.02
0
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
Japan / Philippines Indonesia / China Malaysia / Philippines
Malaysia / Thailand Singapore / Malaysia
These trends generally reflect the relative changes in real wages, although there are other
factors that also feature:
a) trends that also include the exogenous changes in migration13 and hence population
due to the forecasted changes in labor;
b) the movement of the migrants themselves also affect real wages and hence the
decision to migrate;
c) migrants and domestic workers are not perfect substitutes in this model and hence the
wages of migrants may differ from the average wages depicted in Figures 4 and 5;
and
d) the real wages in Figures 4 and 5 do not take account of changes in unemployment
due to the financial crisis.
When we examine the disposition of the nationals from a specific country – specifically, how
many of them are expatriates – we see that all countries have seen increases in the number of people
that are living outside the home region (column I, Table 5). However, as a share of their total
population regardless of their current location, there are fewer Indonesians, Malaysians, Filipinos, and
Vietnamese living overseas (column II, Table 5). For example, more than 5.2 percent of all
Malaysians were living outside Malaysia, representing 0.65 million people in 2007. By 2050, the total
number of Malaysians living overseas increased by 0.19 million, although the share of all Malaysians
living overseas has fallen by 2.13 percent.
13The extent to which migration occurs also depends on the choice of the parameter ESUBMIG in Equation (1).
We chose a value of one and undertook sensitivity analysis. In this sensitivity analysis we found that lowering
ESUBMIG to 0.4 lowered absolute changes in migration, however the direction of the flows of the migrants was
the same.
19
Table 5: Changes in National Populations Living Outside the Home Region, 2007-2050
Additional Effect of
Liberal Forecasts
Endogenous Migration
Change in Change in Change in
Change in share of Change in share of Change in Share of
Number of Expatriates Number of Expatriates Number of Expatriates
Expatriates in Total Expatriates in Total Expatriates in Total
(millions) Nationals (millions) Nationals (millions) Nationals
(%) (%) (%)
I II III IV V VI
China 2.93 0.21 1.17 0.06 1.77 0.15
Hong Kong 0.11 6.51 0.09 7.28 0.02 -0.77
Indonesia 0.59 -0.09 0.64 0.09 -0.05 -0.18
Japan 0.12 0.60 0.09 0.64 0.02 -0.05
Malaysia 0.19 -2.13 0.10 -2.36 0.09 0.23
Philippines 0.49 -1.57 0.51 -1.19 -0.02 -0.37
Rest E. Asia 0.19 -0.21 0.09 -0.56 0.11 0.35
Rest S.E.
Asia 4.36 1.53 0.20 -0.27 4.16 1.80
Singapore 0.06 2.13 0.03 0.53 0.04 1.60
South
Korea 0.14 0.05 0.12 0.06 0.02 -0.01
Taiwan 0.06 0.23 0.05 0.35 0.01 -0.12
Thailand 0.07 0.02 0.15 0.35 -0.08 -0.33
Vietnam 0.27 -0.64 0.26 -0.61 0.01 -0.03
The Macroeconomic Impact of the Liberalization of Migration
Table 6 shows the impact on GDP decomposed into changes in capital, skilled and unskilled
permanent residents and migrants, and technological change. The changes in real GDP depend on
these changes in the workforce caused by the migration and on these changes and the importance of
each of these in real GDP (shown by the initial shares). The results are also divided into forecasted
changes in natural population growth and migration; and endogenous migration due to more liberal
migration policies. Overall real GDP rises over the period due to forecasted changes in the labor
force, increased capital accumulation and technological changes. Only Hong Kong, Japan, Singapore,
and Taiwan experience declining skilled and/or unskilled forecasted labor growth, which would have
adversely affected the overall positive growth in real GDP. The decline in technological change in
Singapore stems from the fact that growth in capital is strong, while forecasted growth in real GDP
over the period is relatively low, particularly when the financial crisis is taken into account.14
The impact of new migrants on real GDP is the result of both changes in forecasted migrants
and the more liberal migration policy. Overall only Japan and Singapore experience a decline in
unskilled migrant workers which impact real GDP negatively (VIII outweighs VII, Table 6). Skilled
migration is positive overall for all regions (XII outweighs any declines in XIII, Table 6).
14
Technological change is calibrated as the residual between real GDP growth and growth in endowments (Solow growth
residual).
20
Although there is an increase in migration into Japan and eventually also into Singapore, the
inflow of new migrants seems surprising low when compared to Hong Kong, especially given that
Japan and Singapore experience similar demographic changes to Hong Kong. This lack of migration
into Singapore and Japan can be attributed to the lacklustre growth in forecasted real GDP over the
period. The new liberal policies allow migrants to respond to real wages. However, if real GDP does
not increase substantially (as it does in Hong Kong) then there are no incentives for migrants to move
to Singapore and Japan. In the case of Singapore, only skilled Singaporeans living in Malaysia are
incentivized to move home. It is not until the long run effects of the financial crisis have dissipated
(2035) that migration into Singapore from other regions increases enough for labor growth to exceed
forecasted growth (Figures 6 and 7) and for the beneficial effects of migration to at last be seen. In
Japan, the low initial share of migrants in the labor force, lacklustre growth in real GDP, and
increased competition from Hong Kong for Japan’s traditional Asian migrants, make Hong Kong a
much more attractive destination to migrants than Japan. Despite the relatively small size of the
increase in migrants into Japan, even this small increase in migration has a positive effect on Japan’s
real GDP. Singapore’s growth is also much higher in the later years when migration finally becomes
positive.15 For this reason Singapore and Japan may want to consider more aggressive liberalization of
their migration policies, by reducing the implicit costs to migration (or rents in Figure 1), thereby
allowing demand by firms and the supply of migrants to increase.16 Japan and Singapore might then
be able to attract migrants from alternative sources, such as the Philippines, Thailand or from outside
of the region.
China also gains in terms of real GDP as a result of the liberalization of migration policies,
although the increase in skilled labor is due to return migration, not new migrants. In general the
economies in East Asia (China, Hong Kong, Japan, and South Korea) all gain from the liberalization
of migration and from the increased capital accumulation that accompanies it (column III, Table 6),
while those in South-East Asia (except Vietnam) experience small losses. The small losses in South
East Asia occur because of the declines in the labor force and/or capital. Changes in the labor force
are driven by changes in the number of workers weighted by their productivities. Decreases in the
labor force therefore occur if decreases in the number of new migrants and/or outward migration
offset return migration, or if migrants leaving are more productive than those entering. In Indonesia
and the Philippines the skilled labor forces decline with migration, despite large return migration17,
causing slight falls in real GDP. In Singapore, despite the high return migration of skilled
Singaporeans, reduced capital accumulation and unskilled labor cause real GDP to fall. This
highlights the importance of access to unskilled workers by developed economies like Singapore. In
Thailand the loss in capital and skilled migrants as a result of the liberalization of migration policy
causes a slight decline in real GDP. Malaysia’s real GDP falls with increasing outward migration of
both skilled and unskilled labor, and decreased inward migration. Finally, Vietnam, gains due to an
increase in new and returning skilled migrants.
Forecasted capital accumulation is the result of the dynamic mechanisms in the model that
cause investment to add to available capital stocks, and the forecasted increases in skilled and
unskilled labor (column II, Table 6). Capital also responds to the liberalisation of migration policies.
Countries that receive more migrants also experience increased production and hence increased
returns to capital, thereby causing more investment and the expansion of capital over time. The
reverse is true for countries experiencing outward migration, albeit the increase in remittances does
offset this to some extent. In Malaysia and the Philippines the outflow of skilled workers abroad
results in substitution towards capital which in turn has led to an unexpected increase in the return to
capital; and hence in the long run, an increase in capital stocks. This increase in capital stocks
however only begins after 2045 when migration flows start to reverse (Figures 9A and 9B).
15
Although by 2050, the gains had not yet outweighed the losses from earlier years where there was considerable outward
migration of unskilled workers. It is expected that if migration flows into Singapore continued, real GDP would have risen
above forecasted soon after 2050.
16
This would then allow migrants to respond to absolute wage differentials, not just changes in wages.
17
Returning skilled migrants are less productive than the skilled migrants that are returning to Singapore, Malaysia and East
Asia.
21
Table 6: ecomposition of the cumulative change in Real GDP between 2007 and 2050 into Capital, Unskilled and Skilled Permanent residents and
Migrants and Technological Change due to Forecasts and More Liberal Migration
Tech
Capital Unskilled Skilled Actual GDP
changec
Initial Initial Permanent Migrants Initial Permanent Migrants
share Forecasts Liberal share Forecasts Liberal Forecasts Liberal share Forecasts Liberal Forecasts Liberal Forecasts Forecasts Liberal
In VAa % Δb %Δ In VA %Δ %Δ %Δ %Δ In VA %Δ %Δ %Δ %Δ %Δ %Δ %Δ
I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI
China 43% 391.3 0.08 40% 18.7 -0.1 18.7 -13.4 11% 81.3 0.4 81.3 -24.5 157.7 434.4 0.05
Hong Kong 54% 941.2 5.1 25% -23.5 0.2 0.4 129.1 20% 34.2 5.7 57.1 22.4 108.2 664.2 14.1
Indonesia 51% 519.4 -0.2 32% 80.2 0.01 68.7 -20.8 8% 199.3 1.1 199.3 -57.5 54.1 328.5 -0.1
Japan 44% 99.4 0.1 35% -27.5 0.01 0.0 -4.0 21% -49.3 0.2 0.0 21.3 61.9 55.8 0.2
Malaysia 46% 509.7 0.2 37% 96.2 -0.08 82.8 -26.9 12% 307.5 -0.6 307.5 -61.0 84.7 564.8 -2.9
Philippines 58% 1112.4 0.1 26% 130.4 0.07 110.3 -4.9 11% 232.5 2.3 232.5 -17.8 141.8 1179.1 -0.02
Rest E.
49% 257.9 -0.2 34% 72.7 -0.7 72.7 0.0 12% 40.1 2.8 40.1 0.0 48.6 223.4 -0.02
Asia
Rest S.E.
43% 658.2 -0.2 29% 67.2 -0.09 0.0 0.0 8% 103.4 0.5 0.0 0.0 112.4 385.2 -0.04
Asia
Singapore 52% 602.9 -0.8 30% 15.8 -2.6 12.6 -14.2 18% -6.1 21.2 11.9 -4.3 -31.8 124.7 -0.5
South
48% 333.6 0.02 34% -1.8 0.2 8.5 33.5 15% 63.0 -0.4 63.0 -35.2 109.1 244.1 0.00
Korea
Taiwan 40% 467.0 0.2 35% 11.9 0.00 11.9 0.0 24% -9.2 1.5 7.7 0.0 18.2 148.0 0.4
Thailand 64% 258.3 -0.4 21% 7.1 0.1 7.3 18.5 9% 61.7 1.0 61.7 -8.2 310.9 471.2 -0.04
Vietnam 38% 349.5 -0.01 38% 86.5 0.00 72.8 -14.4 12% 51.3 0.4 51.3 10.3 203.1 405.3 0.06
Note: a. In VA – this is the initial share in value added.
b. % Δ – Percent change
c. Technological change is calibrated as the residual between Real GDP growth and growth in endowments (Solow growth residual).Technological change for a given country has been has been
weighted by considering whether it was on capital, labor or other endowments.
22
Table 7: Macro Results: Cumulative change between 2007 and 2050 due to Forecasts and More Liberal Migration
Trade Real
Real wage-Skilled Real wages-Unskilled Exports Imports Remittances ina Remittances outa
Balancea Incomeb
Forecast Libera Forecasts Liberal Forecast Liber Forecast Liber Liberal Liberal Forecast Liberal Forecast Liberal
s l s al s al s s
%Δ %Δ %Δ %Δ %Δ %Δ %Δ %Δ $US Mill $US Mill $US Mill $US $US Mill $US
Mill Mill
China 84.3 -0.2 70.8 0.1 654.8 0.6 229.5 0.9 2,317 12890 12245 447 147 -10
Hong Kong 341.8 -6.6 579.3 -17.9 664.5 17.5 706.2 10.9 3,407 25219 1 0 12684 603
Indonesia -21.8 0.08 -6.0 -0.3 371.5 -0.1 373.4 0 275 215 2366 26 94 -35
Japan 189.8 -0.3 119.3 0.2 47.9 -0.2 160.1 0.6 -10,296 13903 1771 -26 5430 70
Malaysia -5.7 15.4 36.5 1.6 715.7 -3.1 771.4 -2.0 -3,087 1846 1145 42 2714 -316
Philippines 75.0 0.9 116.8 0.2 3360.0 0.2 1762.4 0.3 -188 -506 12425 -120 741 -19
Rest E. Asia 94.9 -1.8 34.0 0.3 238.9 -0.3 305.8 0.2 -207 415 1614 58 51 0
Rest S.E. Asia 120.7 -0.5 41.5 -0.1 194.0 -0.1 545.0 0.02 51 62 483 -3 340 0
Singapore 54.5 -5.6 39.5 4.0 150.5 -1.0 165.9 0.05 -4,167 1156 1439 8 1420 -89
South Korea 67.8 0.8 128.6 -0.6 357.9 0.6 262.4 0.4 2,935 441 1375 14 901 -12
Taiwan 91.9 -0.9 55.9 0.2 178.9 0.7 177.1 0.6 356 614 2625 -37 0 0
Thailand 194.7 -0.4 186.3 -0.8 508.8 0.09 402.4 -0.03 564 -1301 2554 -117 971 -3
Vietnam 206.8 -0.4 59.4 0.08 323.4 -0.08 306.6 0.07 76 471 3823 -9 52 0

Note: a. Absolute change
b. Real income of permanent residents
23
Table 7 provides results for some other macroeconomic variables. The real wages of
unskilled and skilled labor generally respond as expected – with real wages falling (rising) with
increased (decreased) migration and labor forces in the long run. With the exception of Hong Kong,
Malaysia and Singapore, most of the changes in real wages and labor forces are relatively small and
therefore this amount of migration only partially offsets the demographic uncertainties experienced by
these economies.
In those countries where inward migration increases, remittances out also rose; the reverse
occurs in those countries that experience outward migration. With remittances out increasing the
current account balance, and hence the trade balance, moves into surplus. Exports rise relative to
imports. Consumption and hence imports also rise due to the increase in the migrant population and
incomes. The exception is Japan, where remittances sent home are very small18 and the increase in
investment and capital inflows caused by the migration generally outweighs the impact of remittances
on the trade balance.
Real incomes of the incumbent populations generally rise as a result of more liberal migration
policies. The reason for this is that people are choosing to return home from countries with relatively
lower wages (lowering remittances, but also raising incomes earned at home) or migrating to those
countries with relatively higher wages (raising remittances). The reverse occurs in Thailand and the
Philippines as migrants return home, lowering remittances and real incomes.
The Sectoral Impact of the Liberalization of Migration
The sectoral results are shown in Table 8. In general sectoral production grows over time
with increased labor, capital and technological change (forecasts). Only Japan and Singapore
experience some declines in production over the period 2007-2050 as decreases in unskilled labor
cause production in certain labor intensive sectors to decline.
Under more liberal migration policies, Hong Kong gains considerably as new migrant
workers enter the labor force, allowing sectors to expand. Since there are increases in both skilled and
unskilled migrant workers, all sectors gain as a result of the liberalization of migration policy. In
Thailand, on the other hand, unskilled labor rises significantly more than skilled labor, and capital
falls. This causes expansion in unskilled labor intensive industries and declines in skilled labor and
capital intensive industries. The reverse happens in China and Japan where we see larger increases in
skilled-intensive services and manufactures following the increase in skilled labor. Many of the gains
in Thailand are in the agricultural sector which is very intensive in unskilled migrant workers.
Malaysia and Singapore experience the largest declines in sectoral output. Malaysia is similar
to Thailand in that the largest declines are in skilled labor and hence skill-intensive industries suffer
the most; while Singapore losses mostly unskilled workers like Japan. Since construction is an
important part of investment goods, the gains or losses to the construction sector are closely aligned
with the changes in the capital stock, although the sector will also be affected by the availability of
unskilled migrant workers since most countries use migrant workers intensely in construction (Table
3, earlier in page 8).
18
This is due to the fact that migrants in Japan are based on nationality rather than birth. Hence there are many Japanese
born people listed as “migrants” living in Japan who not have families abroad to send remittances home to. Hence the
remittance rate per person is very low.
24
Table 8: Cumulative Changes in Sectoral Production 2007-2050 due to Forecasts and the Liberalization of Migration Policy ($US Millions)
China Hong Kong Japan Malaysia Singapore Thailand
Forecastsa Liberal Forecasts Liberal Forecasts Liberal Forecasts Liberal Forecasts Liberal Forecasts Liberal
Rice 74477 10 5 0 1468 12 784 -8 -1 0 60393 2
Wheat 55777 7 19 0 1138 -2 9 0 -7 0 52 0
Grains Crops 651722 -223 1059 7 11103 38 4076 -5 -203 -4 76256 -15
Cattle and Wool 112796 -40 940 4 4252 12 369 -1 -1 0 1990 -1
Other Animals 489377 282 5324 49 3372 24 5247 10 -12 0 10243 13
Meat Products 150836 78 711 61 3261 5 -264 14 127 -3 18422 13
Processed Rice 75832 10 2 1 1267 14 197 -1 92 -2 32349 16
Other Food 854926 98 24964 1074 136505 394 82565 -578 9144 -52 53530 14
Forestry 207409 -25 17 0 6609 -18 12447 -9 -9 0 8281 -1
Extraction 613449 7 71041 67 27507 2 46065 -5 -70 0 84474 1
Textiles 1306142 -1977 73981 4225 -11676 -788 8274 -168 -502 -52 39421 -143
Apparel 701527 -379 29541 3426 2928 -99 2324 -39 -395 -25 25869 -15
Leather Products 354667 -8 3032 361 -1060 -18 1470 -8 -211 -20 10443 -6
Wood and Paper Products 943963 -306 68036 1682 96975 90 35254 -180 3069 -107 40764 -45
Motor Vehicles 435941 -181 35741 300 307680 -483 61687 -185 1312 -15 131973 -34
Electronics 2068471 3604 133943 1008 14011 1806 944064 -2440 54847 -572 185559 251
Other Machinery and equip 1085880 -647 157785 1221 161595 -151 148953 -983 -11232 -370 76582 -27
Petroleum and Coal Products 595971 -136 3172 2 45003 121 52118 -206 41041 121 93406 -49
Chemicals, Rubbers and 1814282 -185 164222 1773 123697 199 266086 -699 52096 -473 172358 20
Plastics
Metals 539549 -159 -5461 -639 130277 831 36629 -277 13881 -34 10843 1
Metal Products 289791 11 25143 348 71600 391 25032 -133 6313 -65 13118 1
Other Manufactures 1032318 1210 5392 62 71440 612 156469 -666 11916 -128 41634 2
Utilities 530825 -65 47415 961 87727 264 51837 -213 4340 -24 49813 -10
Construction 75593 -403 250696 1164 632354 3214 77886 -139 53229 219 77574 -111
Transport and 3087662 -2775 1937158 38613 1491577 1243 206060 -1078 88027 -538 245549 -282
Communications
Business Services 1414860 104 267193 9785 772761 1771 131407 -855 87865 -458 150526 -38
Govt, Health and Educ 2836662 978 105254 3920 838907 2874 127815 -895 22710 69 179395 17
Note: a. These are results determined by the model corresponding to forecasts used in the baseline simulation.
25
4. CONCLUSION
This study analyses the impact of a more liberal migration policy within the East Asia and Pacific
region. Migration flows in this scenario come from two sources. First, we assume that the migration status
quo of a country is one where the country’s migrant shares in the labor force remain constant over time.
Second, we assume that migration is liberalized so that migrants can respond endogenously to changes in
the real wages in the sending and receiving economies (liberal), in addition to the migrant flows that
would occur to maintain the migrant share status quo. Overall, we find that increased migration results in
gains to both the sending and receiving countries in terms of real GDP or incomes.
When migrants are able to respond endogenously to changes in relative wages arising from
changes in demography and other economic factors, there is increased migration to East Asia, as well as
return migration by East Asians previously living as migrants elsewhere. With the exception of Thailand
and the Philippines, all the East and South-East Asian economies gain in terms of real income. Thailand
and the Philippines experience substantial return migration, leading to lower remittances, which cause
incomes to fall. The large inflow of migrants and return migrants into those East Asian economies
experiencing the strongest demographic changes, also causes an increase in real GDP for those
economies.
While the increases in migration are insufficient to completely offset the declining labor forces in
East Asian countries with shrinking populations, when migration is able to endogenously respond to
international differences, labor and wages adjust to reduce the economic effects of the demographic
changes over the period. Countries that receive more migrants also experience increased production and
greater returns to capital, subsequently attracting more investment and capital growth over time. The
combination of increased labor and capital leads to the increase in real GDP found in East Asia.
Even in Japan and Singapore, where the response of migration to the demographic changes is
considered low, positive gains in real GDP from migration are evident. For this reason Singapore and
Japan might want to consider more aggressive liberalization of their migration policies to attract migrants.
Significant changes in migration patterns are also expected to occur over the period examined,
2007-2050. Countries that are currently net senders of migrants may become net recipients under a more
liberal migration policy that allows endogenous movements in response to wage changes. For example,
China, Indonesia, and the Philippines are currently net senders of skilled migrant workers. However, they
became net recipients of skilled migrants as skilled workers return home in response to changes in relative
wages. Conversely, net recipients of skilled migrants, like Malaysia, and unskilled migrants, like
Singapore, under current migration policies, may become net senders of those migrants under a more
liberal migration policy.
26
REFERENCES
Aguiar, A.H. (2009) "An Analysis of U.S. Immigration and Policy Reforms". Ph.D. dissertation,
Purdue University.
Athukorala, P. (2006) “International Labor Migration in East Asia: Trends, Patterns and Policy
Issues,” Asian-Pacific Economic Literature, Vol. 20, No. 1, pp. 18-39.
Bloom, D.E. and J.E. Finlay (2009) “Demographic Change and Economic Growth in Asia,”
Asian Economic Policy Review, 4, 45–64.
Bloom, D.E. and J.G. Williamson (1998) “Demographic Transitions and Economic miracles in
Emerging Asia,” World Bank Economic Review, 12, 419–455.
Bloom, D.E., D. Canning, and P. Malaney (2000) “Population Dynamics and Growth in Asia,”
Population and Development Review, Vol. 26, pp. 257-290.
Hanson, G. H. (2008) “International Migration and Development," Working Paper, 42, Commission on
Growth and Development
Hanson, G.H. and A. Spilimbergo (1999) "Illegal Immigration, Border Enforcement, and Relative Wages:
Evidence from Apprehensions at the U.S.-Mexico Border," American Economic Review, Vol. 89, No. 5,
pp. 1337-1357.
Harris, J. R. and M.P. Todaro, (1970) "Migration, Unemployment and Development: A Two-Sector
Analysis", American Economic Review, Vol. 60, No. 1, pp. 26–142.
Hertel, T. (1997) Global Trade Analysis: Modeling and Applications. Cambridge: Cambridge University
Press.
Holumyong, C. and S. Punpuing, (2010) "Thailand’s Regulatory, Institutional and Governance System for
Managing Foreign Workers: A perspective on a cost-benefit of the migration status of migrants"
Presented at the IPS and World Bank Labor Mobility and Development Conference, Singapore, 1-2 June
2010.
Hur, J.J. (2010) "The Economics and Governance of Korea’s Migrant Worker Policy" Presented at the
IPS and World Bank Labor Mobility and Development Conference, Singapore, 1-2 June 2010.
Ianchovichina, E., and R. McDougall (2001) “Theoretical Structure of Dynamic GTAP.” Working paper,
Center for Global Trade Analysis.
IMF (2010) "Balance of Payments Statistics Online" http://www.imfstatistics.org/, Retrieved on July 19

2010.
Manning, C., and A. Sidorenko (2007) “The Regulation of Professional Migration: Insights from the
Health and IT Sectors in ASEAN,” The World Economy, 30(7):1084–113.
Massey, D., J. Arango, G. Hugo, A. Kouaouci, A. Pellegrino, and J. Taylor (1998) Worlds in Motion:
Understanding International Migration at the end of the millennium. Oxford University Press Inc., New
York.
27
Narayanan G., B., and T.L. Walmsley Editors (2008) Global Trade, Assistance, and Production: The
GTAP 7 Data Base, Center for Global Trade Analysis, Purdue University.
Özden, Ç., C.R. Parsons, M. Schiff, and T.L. Walmsley (2011) “Where on Earth is Everybody?" The
Evolution of Global Bilateral Migration 1960-2000,”. World Bank Economic Review, June, pp.1-45.
Pissarides, C.P. and I. McMaster (1990) "Regional Migration, Wages and Unemployment: Empirical
Evidence and Implications for Policy," Oxford Economic Papers, Vol. 42, No. 4 (Oct., 1990), pp. 812-
831
Rodrik, D. (2004) “How to Make the Trade Regime Work for Development,” mimeo. Available:
ksghome.harvard.edu/~drodrik/How%20to%20Make%20Trade%20Work.pdf.
Samuelson, P. (1964) "Theoretical Notes on Trade Problems," Review of Economics and Statistics 46:
145-54.
Strutt, A., and T.L. Walmsley (2010) “Trade and Sectoral Impacts of the Financial Crisis: A Dynamic
CGE Analysis.” 13th Annual Conference on Global Economic Analysis, Bangkok, Thailand.
(https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=3229)
Teng, Y.M. (2010) " Singapore’s System for Managing Foreign Workers" Presented at the IPS and World
Bank Labor Mobility and Development Conference, Singapore, 1-2 June 2010.
Tyers, R. and S. Qun (2007) “Demographic Change and Policy Responses: Implications for the Global
Economy,” The World Economy, Vol. 30, No. 4, pp. 537-566. DOI: 10.1111/j.1467-9701.2007.01004.x
United Nations (2009) World Population Prospects: The 2008 Revision, United Nations, New York.
Walmsley, T.L. and S.A. Ahmed (2008) “Asian Migration Prospects: 2007–2012,” ADB Economics
Working Paper No. 133, Asian Development Bank, Manila.
Walmsley, T.L., S. A. Ahmed and C. Parsons (2007) "The GMig2 Data Base: A Data Base of Bilateral
Labor Migration, Wages and Remittances", GTAP Research Memorandum No. 6.
Walmsley, T.L., L.A. Winter, and S.A. Ahmed (2009) “Measuring the Impact of the Movement of Labor
Using a Model of Bilateral Labor Migration”, Second Conference on International Migration and
Development, World Bank, Washington DC, September 10- 11, 2009.
Winters, L. A. (2001) “Assessing the Efficiency Gain from Further Liberalization: A Comment,” in P.
Sauve and A. Subramanian, eds., Efficiency, Equity and Legitimacy: The Multilateral Trading System
and the Millennium, Chicago University Press, Chicago.
World Bank (2006) Global Economic Prospects 2006: Economic Implications of Remittances and
Migration, World Bank, Washington, DC.
Yean, T.S. and L.C. Siang (2010)"Impact of Foreign Labor on Average Labor Productivity in Malaysian
Manufacturing, 2000-2006" Presented at the IPS and World Bank Labor Mobility and Development
Conference, Singapore, 1-2 June 2010.
28
Chapter 2: Remittances, Household Investment
and Poverty in Indonesia
RICHARD H. ADAMS, Jr.*

Consultant, The World Bank
and
ALFREDO CUECUECHA
Universidad Iberoamericana Puebla, Economia y Negocios
Blvd. Del Niño Poblano 2901, Unidad Territorial Atlixcayotl, CP 72430, Puebla, Puebla,
México, E-Mail: alfredo.cuecuecha@iberopuebla.edu.mx
ABSTRACT: This paper analyzes the impact of international remittances on household investment and
poverty using panel data (2000 and 2007) from the Indonesian Family Life Survey (IFLS). Using a three-
stage conditional logit model with instrumental variables to control for selection and endogeneity, it finds
that households receiving remittances in 2007 spend more at the margin on one key consumption good
(food) and more at the margin on one important investment good (education) compared to what they
would have spent on these goods without the receipt of remittances. Using a bivariate probit model with
random effects to control for selection and simultaneity, the paper also finds that households receiving
remittances are less likely to be poor compared to a situation in which they did not receive remittances.
These findings are important because they show that households can use remittances to help build human
capital and to reduce poverty in remittance-receiving countries.
1. INTRODUCTION
International remittances refer to the money and goods that are transmitted to households by
migrant workers working outside of their origin communities. At the start of the 21st Century these
resource transfers represent one of the key issues in economic development. In 2009 international
remittances to the developing world amounted to US $307 billion (World Bank, 2011) and were about 50
percent larger than the level of official development aid to the developing world.
From the standpoint of economic development, two key questions surround these large remittance
flows: (a) How are international remittances spent or used by households in origin countries?; and (b)
What is the impact of these remittances on poverty in the developing world? Answers to these two key
questions seem central to any attempt to evaluate the overall effect of migration and remittances on the
developing countries of Latin America, Asia and Sub-Saharan Africa.
29
In the literature there are at least three competing views on how international remittances are
spent or used by households, and their effect on economic development. The first, and probably most
widespread, view is that remittances are fungible and are spent at the margin like income from any other
source. In other words, a dollar of remittance income is treated by the household just like a dollar of
wage income, and remittance income is spent just like any other source of income. The second view
argues that the receipt of remittances can cause behavioral changes at the household level and that
remittances tend to get spent on consumption rather than investment goods. For example, a review of the
literature by Chami, Fullenkamp and Jahjah (2003:10-11) reports that a “significant proportion, and often
the majority” of remittances are spent on “status-oriented” consumption goods. A third, and more recent,
view arising out of the permanent income hypothesis is that since remittances are a transitory type of
income households tend to spend them more at the margin on investment goods -- human and physical
capital investments – than on consumption goods, and that this can contribute positively to economic
development (Adams, 1998). For instance, in a recent study in Guatemala, Adams and Cuecuecha (2010)
find that households receiving remittances spend less at the margin on food and more at the margin on
investment goods (education and housing). In a similar study in the Philippines, Yang (2008) reports that
positive exchange rate shocks lead to a significant increase in remittance expenditures on education.
On the issue of remittances and poverty, the literature is a bit clearer: most studies find that
international remittances reduce poverty in developing countries. For example, using data from
household surveys in 71 developing countries, Adams and Page (2005) report that, on average, a 10
percent increase in international remittances in a developing country will lead to a 3.5 percent decline in
the share of people living in poverty. In a similar study using household survey data from 10 Latin
American countries, Acosta et al (2006) find that international remittances reduce poverty by 0.4 percent
for each percentage point increase in the remittances to GDP ratio. Finally, at the country level, Lopez-
Cordova (2005) in Mexico, Yang and Martinez (2006) in the Philippines and Lokshin et al (2010) in
Nepal all find that international remittances reduce poverty.
The purpose of this paper is to extend the debate concerning the impact of international
remittances on household investment and poverty by analyzing the results of a large, panel household
budget survey in Indonesia. Indonesia represents a good case study for examining these issues because
the country produces a large number of international migrants to Malaysia, Saudi Arabia and other
countries.1 Also, the presence of panel household data from Indonesia makes it possible to overcome
several of the methodological problems – simultaneity, reverse causality, and omitted variable bias – that
bedevil any economic work on international remittances.
The paper is based on two methodologies, one for the analysis of the effect of international
remittances on household investment, and another for the analysis of the impact of remittances on
poverty. Evaluating the impact of remittances on household investment faces the obvious challenge of
selection, that is, households receiving remittances might have unobserved characteristics (e.g. more
skilled, able or motivated members) which are different from households not receiving remittances. If
these unobserved characteristics are constant through time, the use of panel data methodologies can
eliminate the bias that they produce. However, if the unobserved characteristics change over time, it is
still important to address the problem of selection in unmeasured characteristics.
To address this issue we use a three stage conditional logit model that tests for the existence of
selection bias in the household receipt of remittances. The identification of this model is based on the use
of instrumental variables. Since past research has found that historical distance to railroad lines and
1
According to the World Bank (2008: 8), 85 percent of the Indonesians that were approved to work abroad in 2006 went to
Malaysia and Saudi Arabia.
30
changes in rainfall patterns are important in the receipt of international remittances (e.g. Adams and
Cuecuecha, 2010; Woodruff and Zenteno, 2007; Munshi, 2003), our identification strategy focuses on
these variables.
This instrumental approach enables us to control for selection and to compare the predicted
marginal budget shares for households conditional on their household characteristics and their receipt of
remittances with the counterfactual marginal budget shares of households conditional on their household
characteristics and on the hypothetical condition where they do not receive remittances. By comparing
the predicted and counterfactual marginal budget shares of households we are able to pinpoint how
households receiving remittances spend at the margin on a broad range of consumption and investment
goods, including food, housing and education.
Our methodology for analyzing the impact of remittances on poverty not only faces the challenge
of selection but also the problem of potential simultaneity. These problems exist because the levels of
income and choices made by households that lead them to be poor are very likely correlated to their
choice of whether or not to receive remittances. Moreover, these household decisions are all made in the
presence of unobserved (to the econometrician) heterogeneity. To address these concerns, we use the type
of bivariate probit model with random effects proposed by Carrasco (2001). This bivariate probit model
allows us to calculate household probabilities of being poor and not being poor conditioning on the
receipt of remittances. This in turn enables us to obtain the average treatment effects of remittances on
the probability of a household being poor or not being poor.
The paper proceeds in seven further parts. Section 1 presents the data. Since the problems of
selection and identification are so important, Section 2 presents the three-stage conditional logit model
and discusses the various identification issues involved in estimating this model. Section 3 estimates the
three-stage model with selection controls. Section 4 presents the predicted and counterfactual marginal
budget shares for households receiving remittances and uses average treatment effects to compare
outcomes. Section 5 presents the bivariate probit model for estimating the impact of remittances on the
probability of a household being poor. Section 6 estimates this probit model and presents average
treatment effects on outcomes. The final section, Section 7, concludes.
2. DATA SET
Data come from Indonesia Family Life Survey (IFLS), an on-going panel household survey in
Indonesia. The IFLS Survey includes four waves of surveying, IFLS 1 (1993), IFLS 2 (1997), IFLS 3
(2000) and IFLS 4 (2007). However, since this paper is on international remittances, and consistent
definitions of remittance variables could not be developed for all four waves of the IFLS survey, the focus
here is on the last two waves of the survey, IFLS 3 (2000) and IFLS 4 (2007). These two waves include a
total of 5301 urban and rural households. While the IFLS Survey was never designed to be nationally
representative, the last two waves of the survey do include households from 19 of Indonesia’s 33
provinces. In terms of data collected, the IFLS Survey was comprehensive, collecting detailed
information on a wide range of topics, including expenditure, education, health, nutrition, financial assets,
household enterprises and remittances.
It should, however, be emphasized that the IFLS Survey was not designed as a migration or
remittances survey. In fact, it collected very limited information on these topics. With respect to
international migration, the survey collected only limited information on migrants who have been gone
31
from the household for more than one year: their age, education or income earned away from home.2
This means that limited data are available on the characteristics of most international migrants who are
currently living outside of the household. With respect to international remittances, the IFLS Survey only
contains information from three types of questions: (1) Does your household receive remittances from
spouse, parents or children?; (2) Where do these people sending remittances live?; and (3) How much
(remittance) money did your household receive in the past 12 months? The lack of data on individual
migrant characteristics in the IFLS survey is unfortunate, but the presence of detailed information on
international remittances and household expenditures makes it possible to use responses in the survey to
examine the impact of remittances on household expenditure behavior.
Since the focus here is on remittances, it is important to clarify how these income transfers are
measured and defined. Each household that is recorded as receiving international remittances is assumed
to be receiving exactly the amount of remittances measured by the survey. This means that households
which have migrants who do not remit are not recorded in this study as receiving remittances; rather these
households are classified as non-remittance receiving households. This assumption seems sensible
because migration surveys in other countries generally find that about half of all migrants do not remit.3
Because of data limitations, the focus throughout this study is on the receipt of international remittances
by the household rather than on migration or the type of person sending remittances. Finally,
international remittances include both cash and in-kind remittances. The inclusion of in-kind remittances
(food and non-food goods) is important because it leads to a more accurate measure of the actual flow of
remittances to households in Indonesia.
Table 1 presents summary data from IFLS 3 (2000) and IFLS 4 (2007). It shows that the number
of households receiving international remittances in Indonesia is fairly small: in 2000, 169 households
(3.2 percent of all households) receive remittances, and in 2007, 179 households (3.3 percent of all
households) receive remittances.4 According to the table, households receiving international remittances
in Indonesia have older household heads, have fewer household members with high school and university
education, and are more likely to be located in rural areas. Households receiving international
remittances also tend to have lower mean per capita expenditures than households without remittances.
For households receiving remittances, remittances represent 26.0 percent of total household expenditures
in 2000 and 29.0 percent of expenditures in 2007. However, since households receiving international
remittances in Indonesia also have low levels of expenditure, the absolute amount of remittances received
in annual per capita terms by households is quite low: not exceeding US $30 in either year.5
Since the focus of this paper is on household expenditure behavior, it is important to present the
type of expenditure data contained in the IFLS Survey (2000 and 2007). Table 2 shows that the survey
collected detailed information on five major categories of expenditure, and on several subdivisions within
each category. While the time base over which these expenditure outlays were measured varied (from
last 7 days for most food items, to last year for most durable goods), all expenditures were aggregated to
2
The IFLS Survey contains detailed information on international migrants who are listed on the household roster (that is,
migrants who have been gone less than one year), but it does not contain any information on migrants who are not listed on the
household roster (that is, those who have been gone for more than one year)/
3
For example, in their study in the Dominican Republic, de la Briere, Sadoulet, de Janvry and Lambert (2002) find that fully half
of all international migrants do not remit.
4
By contrast, recent nationally-representative household surveys in Guatemala (2000) and Ghana (2005/06) show that the share
of households receiving international remittances was 7.1 percent and 5.4 percent, respectively. For details on these surveys, see
Adams and Cuecuecha (2010) on Guatemala and Adams, Cuecuecha and Page (2008) on Ghana.
5
By contrast, the household surveys cited in note (4) show that the absolute amount of international remittances received in
annual per capita terms by remittance-receiving households was US $365 in Guatemala and US $417 in Ghana (nominal terms).
32
Table 1. Summary of Data on Non-Remittance and Remittance-Receiving Households, Indonesia,
2000 and 2007
2000 2000 2000 2007 2007 2007

t-test t-test
Receive no Receive (Receive Receive no Receive (Receive
Variable remittances remittances remittances remittances remittances remittances
vs. no vs. no
remittances) remittances)
Mean age of 50.17 55.03 2.19*** 52.80 56.53 4.26***
household (29.13) (14.22) (12.82) (15.26)
head (years)
Number of 0.38 0.40 0.37 0.28 0.35 1.93*
children (0.59) (0.65) (0.53) (0.60)
below 5
years in
household
Number of 1.38 1.22 -1.63 1.03 1.05 0.33
children (1.22) (1.17) (1.08) (1.06)
between 6
and 18 years
old in
household
Number of 1.46 1.20 -0.07 1.32 1.33 0.12
household (1.15) (.82) (1.10) (1.06)
members
with
primary
education
Number of 0.74 0.59 -2.16** .93 0.61 -3.78***
household (1.17) (.93) (1.25) (0.93)
members
with high
school and
university
education
Area 0.33 0.21 -3.15*** 0.38 0.28 -3.26***
(0=rural, (0.47) (0.41) (0.48) (0.45)
1=urban)
Mean 702.5 614.2 -1.34 1007.8 931.3 -.032
annual per (861) (471) (3573) (1240)
capita
household
expenditures
(000
Indonesian
rupiah) at
2000 prices
33
2000 2000 2000 2007 2007 2007
t-test t-test
Receive no Receive (Receive Receive no Receive (Receive
Variable remittances remittances remittances remittances remittances remittances
vs. no vs. no
remittances) remittances)
Remittances NA 26.0 NA NA 29.0 NA
as percent of (42) (63)
total per
capita
household
expenditure
N 5132 169 5122 179
Notes: N=5301 households. Standard deviations are in parentheses. In 2000, 8422 Indonesian rupiah= US$1.00; in 2007, 9141
Indonesia rupiah=US$1.00.
Source: Indonesia Family Life Survey (IFLS), 2000 and 2007.
*Significant at the 0.10 level. **Significant at the 0.05 level. ***Significant at the 0.01 level.
Table 2. Expenditure Categories and Average Budget Shares, Indonesia, 2000 and 2007
Expenditure Description No Remittances No Remittances

category remittances in 2000, no remittances in 2000,
in 2000, no remittances in in 2000, remittances
remittances 2007 remittances in 2007
in 2007 in 2007
2000 2000 2000 2000
2007 2007 2007 2007
Food Purchased food 0.600 0.609 0.615 0.627
Non-purchased 0.551 0.535 0.550 0.563
food
Education Educational 0.049 0.049 0.045 0.029
expenses 0.051 0.051 0.056 0.049
Housing Housing value 0.100 0.091 0.092 0.093
0.112 0.157 0.114 0.110
Health Health expenses 0.018 0.024 0.010 0.017
0.020 0.030 0.020 0.052
Other Household 0.232 0.227 0.238 0.234
durables, 0.266 0.226 0.260 0.225
Transport,
Communications,
Legal
1.000 1.000 1.000 1.000
1.000 1.000 1.000 1.000
Notes: N=5301 households. All values are weighted. International remittances include remittances received from spouse,
parents and children.
Source: Indonesia Family Life Survey (IFLS), 2000 and 2007.
34
obtain yearly values. For household durables (stove, refrigerator, automobile, etc), annual use values
were calculated to obtain an estimate of the cost of one year’s use of that good. Annual use values were
also calculated to obtain an estimate of the one year use value of housing (rented or owned).
Table 2 also shows the average budget shares devoted to the five categories of goods for each of
the four groups of households. On average, each of the four groups of households spends over 53 percent
of their budgets on one key consumption item – food – and less than 6 percent of their budgets on
education.
An Econometric Model of Household Incomes with Selection Controls and the Estimation of the
Marginal Expenditure Behavior of Households
The purpose of this section is to analyze the marginal expenditure patterns of households
receiving remittances. To do this, it is necessary to choose a proper functional form for the econometric
model. The selected functional form must do several things. First, it must provide a good statistical fit to
a wide range of goods, including food, housing and education. Second, the selected form must
mathematically allow for rising, falling or constant marginal propensities to spend over a broad range of
goods and expenditure levels. A model specification that imposes the same slope (or marginal budget
share) at all levels of expenditure would not be adequate. Third, the chosen form should conform to the
criterion of additivity (i.e. the sum of the marginal propensities for all goods should equal unity).
One useful functional form which meets all of these criteria is the Working-Leser model, which
relates budget shares linearly to the logarithm of total expenditure. This model can be written as:6
Cs /EXP = βs + as /EXP + γs (log EXP) (1)

where Cs /EXP is the share of expenditure on good s in total expenditure EXP. Adding up requires that Σ
Cs/ EXP = 1.
Equation (1) is equivalent to the Engel function:
Cs = as + βs EXP + γs (EXP) (log EXP) (2)
To estimate the marginal expenditure shares of households we use a three-stage model to estimate
predicted and counterfactual expenditures for households taking into account selection bias. In the first
stage of the model, we estimate the probability of a household receiving remittance using a conditional
logit specification. 7 In the second stage, we use a Gaussian kernel to estimate weights that assign larger
weights to observations with lower selection bias. In the third stage of the model, we use a first difference
regression for households that always receive remittances or households that never receive remittances in
order to obtain the parameters for our expenditure equations. This regression is weighted according to the
weights obtained in the second stage and controls for selection using the probabilities of receiving
remittances that are estimated in the first stage.
6 The functional form used in this analysis differs from the Working-Leser model because it includes an intercept in equation (1).
In theory, Ci should always equal zero whenever total expenditure EXP is zero, and this restriction should be built into the
function. But zero observations on EXP invariably lie well outside the sample range. Also, observing this restriction with the
Working-Leser model can lead to poorer statistical fits. Including the intercept term in the model has little effect on the
estimation of marginal budget shares for the average person, but it can make a significant difference for income redistribution
results. For more on the Working-Leser model, see Prais and Houthakker (1971).
7
The original method proposed by Kiriazidou (1997) uses a conditional maximum score estimator. Charlier, Melenberg and van
Soest (2001) propose the use of a conditional logit.
35
The panel data from Indonesia is for two years (2000 and 2007) and this gives us certain
advantages over simple cross-sectional data. For example, we know whether households have chosen to
receive remittances in each of four states: (1) receive no remittances in either year; (2) receive
remittances in 2000 but not in 2007; (3) receive remittances in both years, 2000 and 2007; and (4) receive
remittances in 2007 but not in 2000. Moreover, some of the characteristics of our households are fixed,
and thus do not change according to their remittance status, while other unobservable characteristics
change depending on how the households choose between the four states.
In the first stage of our model, we assume that the decision process of the households can be
represented as follows: in time period 1, households can select between two states (r): (1) receive no
remittances; (2) receive remittances. Once households have chosen their state, they decide their level of
expenditure Ctr, where Ctr is the optimal expenditure for households that chose r=r. At time period 2, they
can again select between two states (r): (1) receive no remittances; (2) receive remittances. Once
households have chosen their state, they decide their level of expenditure Ct+1,r. We assume that this
decision tree is represented by a conditional logit process and that for any good s we have a system of
equations (we abstract from the subscript s for simplicity):
= ( + − ≥ 0) (3)
= + ( )( )+∑ + ( ) + + =0
(4)
= + ( )( )+∑ + ( ) + + =1
(5)
Where represents the jth characteristic of the household that enters the remittances decision
equation, Zj represents the jth characteristic of the household that enters the consumption equations ,
represents expenditure by household i in time t, when the household does not receive remittances;
is the total expenditure of the household, represents the jth characteristic of the household in time t
that enters the consumption equation, represents the fixed effect that enters the equation for
households that do not receive remittances, and represents the error term in the consumption equation
for households that do not receive remittances. represents the expenditure for each household i in
time t that receives remittances, represents the fixed effect that enters the equation for households that
receives remittances, and represents the error term in the consumption equation for households that
receive remittances.
To identify the model, we need instrumental variables that enter the conditional logit estimation,
but do not enter the other stages of the model. In our case, these instrumental variables are two: (1) the
distance from kabupaten (district) to railroad station in 1930; and (2) the level of rainfall in 1995-1999.8
Our rationale for using these two instrumental variables is as follows.
The first railroad line in Indonesia opened in 1867. A continuous railroad line between Jakarta
and Surabaya, the two largest cities in Java, opened in 1894. Between 1900 and 1930 smaller railroad
lines were constructed in Madura, Sumatra and South Celebes. In Indonesia distance to railroad lines in
1930 represents a good instrumental variable because it is related to migration costs in the past and to the
need for sending migrants in the past, and therefore to the development of present day migrant social
8
In a simpler version of the model, one in which the effect of receiving remittances is modeled as a change in intercept in the
expenditure equation, the two instruments are tested for under-identification, weakness and over-identification. The three
instruments are significant at the 1% level in the first stage, the instruments reject the null hypothesis of under-identification, the
instruments present a Cragg-Donald F statistic that demonstrates that they are not weak, and the tests do not reject the null of
valid instruments.
36
networks, but it is not correlated with the expenditure patterns of households at the time of the 2000 and
2007 IFLS Surveys. We calculated distance to railroad lines for each household using the distance from
the kabupaten (district) to the nearest railroad station in 1930, using historical maps from the Indonesian
Railway Authority, and then cross-checking this information with the IFLS Surveys. This type of
instrument has been used before in the literature by Woodruff and Zenteno (2007) for the case of Mexico,
and Adams and Cuecuecha (2010) for the case of Guatemala.
Changes in rainfall have also been used before in the literature as an instrumental variable in the
cases of Mexico, the Philippines, and Guatemala (Munshi, 2003; Yang and Choi, 2007; Adams and
Cuecuecha, 2010). The argument here is that rain is closely correlated with agricultural production and
income, and so too little rain in one or several years may cause people to migrate out of rural areas. For
this reason, changes in historical rain are correlated with the formation of migrant networks and with the
receipt of remittances, but changes in historical rain are not correlated with unobserved changes in
consumption patterns. We obtained historical rainfall information at the meteorological station level in
Indonesia from the IFLS data. We then calculated the average level of rainfall for the period 1995 to
1999, by district. Our argument here is that changes in migration patterns and the receipt of remittances
are influenced by the actual level of rainfall for 1995 to 1999, since the level of rainfall is exogenous at
the beginning of the decision process estimated with our data.
For the two instrumental variables, our claim is that conditional on the set of human capital,
household and district characteristics included in our specification, the unobserved components in the
expenditure equation of the households are uncorrelated with the two instruments.
Equations (4) and (5) can be first differenced for households that did not change their remittance
status, either because they have never received remittances or because they have always received
remittances. We need to use first differences to eliminate the fixed effect in the expenditure equation. So,
for households whose remittance status has always been r, the change in consumption is defined as:
∆ = ∆ + ( )( )−( )( ) +∑ ∆ + ( ) −
( ) + (6)
Where we have that = − . Notice that equation (6) is only defined for either households that
always receive remittances or for households that never receive remittances. Moreover, if unobservable
components that participate in the expenditure function actually change over time, the estimation of
equation (6) may still suffer from selection bias. In our case, since we have a time lag of seven years
(2000 to 2007) between surveys, it is quite possible that unobservable components might have changed
over time. Therefore, to control for selection, we apply two further corrections to equation (6). The first
correction follows Kyriazidou (1997) and Charlier, Melenberg and van Soest (2001). The second
correction follows Dubin and McFadden (1984) and Bourguignon, Fournier and Gurgand (2004).
The first correction represents the second stage in our method. Kyriazdiou (1997) and Charlier,
Melenberg and van Soest (2001) propose to estimate equation (6) by weighting observations according to
a function of the differences between the linear predictions of the selection equations for times t and t-1.
In our case, this corresponds to obtaining the difference between the linear predictions at time t and time
t-1, which becomes:
( )= ( − − + ) (7)
Note that households for which Dit is near zero have a very similar probability of being in their
respective branches and consequently similar selection bias, while households for which Dit is different
from zero differ more in their selection bias over time. Function F(.) is obtained using a Gaussian kernel
37
function. The Gaussian kernel function is selected because it generates weights that assign higher
probabilities to events near zero and lower probability to events farther away from zero, in either
direction. Therefore, households with smaller selection bias are given a higher weight in the estimation.
The second correction represents the third stage in our method. It is based on a method proposed
by Dubin and McFadden (1984) and Bourguignon, Fournier and Gurgand (2004). This is a method which
is used in cross-section data and that corrects selection bias when multiple sources of bias are present.
Here, we realize that our conditional logit structure generates k=4 four potential household types: (1)
households that never receive remittances; (2) households that did not receive remittances in 2000, but
changed their status to receive remittances in 2007; (3) households that receive remittances in 2000, but
change their status to not receiving in 2007; and (4) household that always receive remittances. Since we
only need to estimate the equations for either households that never receive remittances or households
that always receive remittances we need to express the expected value of taking into account the type
of household k that we are studying. It can be shown that the expected value of conditional on being
in the equation of a household of type k’ can be represented as a linear combination of the probabilities of
being a household of each of the other three types. We use this to express the change in consumption in
good s as follows:
∆ = ∆ + ( )( )−( )( ) +∑ ∆ + ( ) −
( ) +∑ ´ ´ + (8)
Where the polynomial on represents the probability of being a household of type k. Therefore we
estimate equation (8) weighting observations using function (7). Equation (8) is obtained using
constrained least squares.9 The marginal budget share (MBS) for good s (we omit subscript for good s, for
simplicity) can be shown to be equal to:
= + (1 + )+∑ (9)
Notice that our estimation of the MBS only depends on values of total expenditure at time t and
observables Zjt, consequently it depends only on information at time t and therefore we obtain the value
for the MBS at time t with remittance status r. For our case, we will obtain the MBS for households that
receive remittances in 2007 and households that do not receive remittances in 2007. To obtain the effect
of remittances on MBS it would be tempting to compare the MBS for households that receive remittances
with the MBS for households that do not receive remittances. However, this comparison would confuse
the effect of remittances on MBS with differences in observable characteristics between households that
receive remittances and households that do not receive remittances.
The average treatment effect on the treated (ATT) that we estimate is based on the difference
between predicted MBS and a counterfactual MBS that let us condition on the characteristics of the
households that receive remittances, as follows:
= − = − +( − )(1 + )+ −
(10)
9 To normalize the changes in expenditure shares over time we follow the following reasoning: all changes in the five
expenditure goods should add up to the aggregate change in expenditure observed for each household. Therefore, all changes in
expenditure are expressed as a fraction of the total change in expenditure per household. Moreover, we constrained the estimation
to guarantee that the sum of the different MBS adds to one.
38
The above ATT can be calculated for two types of households: those that received no remittances
in 2000, but receive remittances in 2007; and those that received remittances in both 2000 and 2007.
Because both of these groups of households have been received remittances, we would expect a different
magnitude in the effects of remittances. We will obtain the ATT for these two types of households as well
as the overall difference. Notice that the ATT can be understood as comparing the marginal behavior of a
household that receives remittances in 2007 with the hypothetical marginal behavior of that household if
it did not receive remittances in 2007.
Estimating the Econometric Model with Selection Controls
Table 3 shows the results for the first-stage conditional logit on the household probability of
receiving international remittances in 2007. It shows that households with more household members
below five years old are significantly more likely to receive remittances.
Table 3. Conditional Logit Model on the Household Probability of Receiving Remittances,

Indonesia, 2007
Dependent variable: Does household receive international remittances in 2007

Coefficient Sd
Variable
Human Capital
Number of household members over age 15 with primary education -0.051 0.147
Number of household members over age 15 with junior secondary 0.251 0.190
education
Number of household members over age 15 with high school to -0.001 0.003
university education
Household Characteristics
Age of household head -0.268 0.318
Sex of household head (1=male) 0.339 0.196
Number of children below 5 years 0.056* 0.121
Number of children between 6 and 18 years old -0.001 0.003
Instrumental variables
Distance from kabupaten (district) to railroad in 1930, adjusted -0.009** 0.004
Rainfall, 1995-1999 3.43E- 1.78E-
06** 06
Log likelihood -146.48 N/A
Likelihood ratio test for model 28.65** N/A
Chi squared test for rainfall and distance to railroad 5.97** N/A
N 464 N/A
Notes: Table reports the coefficients of a variable on the probability of household receiving international remittances in
2007. The model also includes a dummy for urban/rural areas, a dummy for whether there is a financial institution in the
village, and dummies for four Indonesia regions. The distance to railroad variable is adjusted in the following manner:
for households that never receive remittances the variable is the simple distance to railroad; for households that receive
remittances in 2007 but not in 2000, it adds 3 to the distance to railroad variable; for households that receive remittances
in 2000 but not in 2007, it adds 2 to the distance to railroad variable; and for households that receive remittances both in
2000 and 2007, it adds 4 to the distance to railroad variable.
**Significant at the 0.05 level.

* Significant at the 0.10 level.
39
In Table 3 both of the instrumental variables are also significant. The instrumental variable,
distance from kabupaten (district) to railroad, is negatively related to the receipt of remittances. This
suggests that households living further away from a railroad in 1930 are less likely to receive international
remittances in 2007. The other instrumental variable, rainfall from 1995 to 1999, is positively related to
the receipt of remittances. This means that households living in areas with more rainfall before 2000 have
a higher probability of receiving remittances in 2007.
Figure 1 shows the distribution of weights obtained in the second stage equation. The figure
shows that the distribution is centered at zero, but with a long left tail. Consequently, households with a
lower difference in linear predictions between time t and time t-1 receive a larger weight in the
estimation.
Tables 4 and 5 show the results of estimating the marginal expenditure behavior of two types of
households: (1) households that never receive remittances (Table 4); and (2) households that receive
remittances in both 2000 and 2007 (Table 5).
The most important variable in these two tables is the selection term, which is the II variable. For
households that never receive remittances (Table 4) the selection term is significant for all goods, except
“other” goods. For households that always receive remittances (Table 5) the selection term is significant
only for education and health. These results suggest that selectivity in unobservable components matters
40
Table 4. Household Expenditure Estimates (selection corrected, fixed effects) for Households That Never
Receive Remittances
Variable Food Education Housing Health Other
Expenditure 1.867*** 0.063 1.451*** 1.004*** -4.547***
(0.281) (0.040) (0.120) (0.053) (0.142)
Expenditure*log(Expenditure) -0.164*** -0.005 -0.167*** -0.092*** 0.498***
(0.013) (0.005) (0.014) (0.006) (0.017)
Human Capital
Number of household members over age 15 74.400 64.792*** 21.922 -80.391*** -231.826***
with primary education (71.963) (8.592) (26.088) (11.489) (30.846)
Number of household members over age 15 171.426** 9.208 -169.595*** -63.468*** 168.313***
with junior secondary education (69.452) (9.188) (27.896) (12.285) (32.985)
Number of household members over age 15 131.663*** 134.828*** -60.939*** 38.260*** -270.871***
with senior secondary and above education (19.090) (6.915) (20.995) (9.246) (24.824)
1.381 1.823*** -15.061*** -3.635*** 10.789***
Age of household head (1.111) (0.602) (1.827) (0.805) (2.160)
Sex of household head (1=male) 58.938 -88.122*** 261.422*** 294.038*** -305.366***
(59.028) (21.086) (64.022) (28.194) (75.700)
Number of children below age 5 201.893* -102.322*** 149.135*** 129.105*** -530.135***
(102.310) (12.093) (36.717) (16.170) (43.415)
Number of children between 6 and 18 years -18.753 42.642*** 20.220 22.164** -172.904***
old (66.469) (6.745) (20.478) (9.018) (24.214)
Bank in the village (1=yes) 234.656*** -23.092* 407.224*** -84.165*** -410.125***
(14.286) (13.513) (41.031) (18.069) (48.515)
Π2 10.093 -2.980** -14.485*** -4.007** -0.214
(7.228) (1.470) (4.464) (1.966) (5.278)
Π3 -0.004*** 3.15E-04 0.001 -0.002 0.003
(0.001) (1.34E-03) (0.004) (0.002) (0.005)
Π4 -1243.741 -275.024*** 562.058*** -181.255* 392.722
(748.532) (81.930) (248.761) (109.549) (294.137)
Adjusted R2 .40 .29 .39 .97
41
.29
Test of joint significance (F)
17.8*** 4.9*** 5.5*** 2.5* .67
N
5023 5023 5023 5023 5023
Notes: Dependent variable is the change in expenditure in good i. All variables shown are introduced as changes, except for selection controls. The equation includes interactions between
expenditure and each characteristic. It also includes a dummy for rural areas and four regional dummies. Figures in parentheses are standard errors. Results are weighted estimations using
Kyriazidou (1997) weights.
*** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level
Table 5. Household Expenditure Estimates (selection corrected, fixed effects) for Households Receiving Remittances
in both 2000 and 2007

Expenditure 14.512*** -3.330*** 0.307 -5.821*** -9.522***
(4.275) (0.797) (1.188) (1.872) (2.953)
Expenditure*logExpenditure -2.762*** 0.462*** -0.063 0.374 1.700***
(0.596) (0.111) (0.166) (0.261) (0.412)
Human Capital
Number of household members over age 105.660 -29.901 -39.890 39.952 47.482
15 with primary education (139.251) (25.957) (38.692) (60.970) (96.190)
Number of household members over age -163.576 30.241 -171.048 204.288 -62.299
15 with junior secondary education (341.692) (63.692) (94.940) (149.607) (236.029)
Number of household members over age
15 with senior secondary and above -149.483 6.634 129.933*** -51.023 -221.086*
education (159.269) (29.688) (44.253) (69.735) (110.018)
-77.196*** 2.585 3.214 -4.583 14.941
Age of household head (12.694) (2.366) (3.527) (5.558) (8.769)
Sex of household head (1=male) 49.762 68.490 44.001 -311.166** 261.648
(288.855) (53.843) (80.259) (126.473) (199.531)
Number of children below age 5 25.838 -33.445 -0.436 -128.189 -266.412**
(183.944) (34.288) (51.110) (80.539) (127.063)
42
Number of children between 6 and 18 -37.295 -25.491* -10.725 -70.840** -66.556
years old (66.697) (12.433) (18.532) (29.203) (46.072)
Bank in the village (1=yes) -1100.919*** 65.376 216.300** 319.420** 909.155***
(357.532) (66.645) (99.342) (156.543) (246.971)
Π1 740.723 -282.737** -159.661 -577.261** 644.574
(619.293) (115.438) (172.073) (271.152) (427.787)
Π3 -550.407 29.194 99.572 -57.496 -154.419
(476.676) (88.854) (132.446) (208.709) (329.272)
Π4 36.069 2.251 -9.931 10.216 13.899
(37.622) (7.013) (10.453) (16.473) (25.988)
Adjusted R2
.98 .98 .92 .95 .96
Test of joint significance (F)
.62 4.5** .46 32.41*** .93
N
47 47 47 47 47
Notes: Dependent variable is the change in expenditure in good i. All variables shown are introduced as changes, except for selection controls. . The equation includes interactions between expenditure and each
characteristic. It also includes a dummy for rural areas and four regional dummies. Figures in parentheses are standard errors. Results are weighted estimations using Kyriazidou (1997) weights.
*** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level.
43
for both households with no remittances and households that always receive remittances in Indonesia. In
other words, estimations ignoring the selectivity part of the model would be biased.
Remittances and Marginal Budget Shares
Table 6 takes the coefficients from Tables 4 and 5 and calculates the predicted and counterfactual
marginal budget shares for the five categories of expenditure for each type of household. This table also
shows the overall Average Treatment Effects on the Treated (ATT), which averages the ATT for all
households receiving remittances in 2007 and compares it to the counterfactual of what would have
happened if these households did not receive remittances in 2007.1
Two of the ATT results in Table 6 (column 7) are noteworthy. First, compared to a
counterfactual situation in which they did not receive international remittances in 2007, households
receiving remittances in 2007 spend more at the margin on one key consumption good: food.
Households receiving remittances in 2007 spend 5.9 percent more at the margin on food than what they
would have spent on this good without the receipt of remittances. Second, compared to a counterfactual
situation in which they did not receive international remittances in 2007, households receiving
remittances in 2007 spend more at the margin on education. Households receiving remittances in 2007
spend 332 percent more at the margin on education than what they would have spent on this good without
the receipt of remittances. This result is important because it shows that households can use remittances
to invest in human capital.
A Model for Estimating the Impact of Remittances on Poverty
To analyze the effect of remittances on poverty, it would be tempting to use a probability model
to calculate the marginal effect of receiving remittances on the probability of receiving remittances.
However, since the discrete variable -- receiving remittances or not -- is correlated with unobservable
characteristics, this kind of approach would be biased. To overcome this problem, we follow the approach
of Carrasco (2001), who uses a bivariate probit model with random effects to analyze the effect of an
endogenous discrete variable on another discrete variable.
10 The ATT’s reported in this section average out the effect of remittances for the two types of counterfactual experiments that
we perform. We obtain a weighted average in which each type of household involved in the
comparison is weighted according to their importance in the population studied. Standard errors reported in this section also
adjust for these weights.
44
Table 6. Marginal Budget Shares on Expenditure and Average Treatment Effects (ATT) for Non-Remittance
and Remittance-Receiving Households, Indonesia, 2000 and 2007.
No Remittances No remittances in 2000, Percent Remittances in 2000, Percent
remittances in 2000, no remittances in 2007 remittances in 2007
in 2000, no remittances Difference Difference
remittances in 2007
in 2007
Predicted Predicted Predicted Counterfactual (3) vs. (4) Predicted Counterfactual (5) vs. (6)
(1) (2) (3) (4) (5) (6)
Food 6.72% 3.37%
0.328 0.323 0.385 0.361 (3.7)*** 0.355 0.343 (3.8)***
Housing -2.99% -4.02%
0.261 0.258 0.218 0.225 (-26.7)*** 0.208 0.217 (-16.3)***
Education 382.27% 182.52%
0.010 0.014 0.050 0.010 (4.4)*** 0.028 0.010 (2.0)**
Health -54.46% -9.82%
0.107 0.105 0.045 0.098 (-2.4)** 0.098 0.108 (-.58)
Others -1.31% -3.11%
0.293 0.301 0.302 0.306 (-1.4) 0.312 0.322 (-.55)
1.000 1.000 1.000 1.000 1.000 1.000
Notes: Column (7) shows the Average Treatment Effects (ATT) of remittances on indicator i. It is calculated as the weighted
average of two ATT that are calculated subtracting column (4) from (3) and column (6) from (5). T-statistics shown in
parenthesis. T-tests conducted using clustered standard errors and weighting observations. A conditional logit is used to calculate
probabilities and Kiriazidou weights, regression with selection correction. ** Significant at the 0.01 level. ** Significant at the 0.05
level. * Significant at the 0.10 level.
In our case, we assume that such a bivariate probit model is based on the following decision tree:
households choose whether or not to receive remittances and then conditional on that decision,
households make other decisions that combined with the play of nature lead to the result that they are
poor or not. The econometrician, however, cannot observe the whole decision tree and can only see
whether or not a household is poor. Let y be a random variable that will be one if the household is poor
and zero in every other case. Because the choice of receiving remittances influences the set of choices that
the household makes, the best way to represent this random variable follows the approach proposed by
Carrasco (2001):
y = I(γ + B x + ω ≥ 0) iff d = 1
y = (11)
y = I(γ + B x + ω ≥ 0)iff d = 0
45
ω =φ +v j = 0,1 (12)
Where x represents the characteristics of the household that influence the probability of being
poor and ω represents the errors in the choice equation that depend on a random effect φ and the
random variable v . Let the variable d be a random variable that takes the value of 1 if the household
receives remittances and the value of 0 in any other case:
= ( + + ≥ 0) (13)
Unlike Carrasco (2001), we lack instruments that enter into the remittance equation, but do not
enter in the poverty equation.1 For this reason, the identification of our model is based entirely on the
assumption of the bivariate probit model.
Once we have estimated the probit model, we obtain the effect of remittances on poverty for
households that receive remittances in 2007 as the difference in probability of being poor and receiving
remittances in 2007 P and the probability of not being poor and receive remittances in 2007P :
= − (14)
Estimating the Probit Model for Poverty
Table 7 presents the results for the estimation of the bivariate probit model with random effects.
The table shows results for both the household probability of being poor (poverty equation) and the
household probability of receiving remittances (remittances equation). In the table poverty is defined
using poverty lines calculated by the World Bank as follows: for 2000, 1,099,584 rupiah/person/year at
2000 prices for urban households and 883,776 rupiah/person/year at 2000 prices for rural households; for
2007, 308,000 rupiah/person/year at 2000 prices for urban households and 235,500 rupiah/person/year at
2000 prices for rural households. For 2000, 14.3% of the households are poor and in 2007 7.3% of the
households are poor.
As expected, the poverty equation in Table 7 shows that households with more educated members
are significantly less likely to be poor. Also, households living in villages with banks and households
living farther away from railroad stations in 1930 are less likely to be poor. Finally, as might be expected,
households living in villages with more rain between 1995 and 1999 are less likely to be poor.
In Table 7 the remittance equation shows that households with more educated members are
significantly less likely to receive remittances. The table also shows that distance to railroad station in
1930 is positively related to the probability of receiving remittances, and that rainfall between 1995 and
1999 is not statistically related to the receipt of remittances. These two latter results are different from
those of our conditional logit model, and may suggest that the results of the conditional logit are affected
by the unobserved wealth of households. In the conditional logit we do not control for the effect of
wealth while in the bivariate probit we are able to condition on the poverty status of households and so
control for the effect that wealth has on the receipt of remittances.
Table 8 takes the coefficients from Table 7 and calculates the average treatment effects (ATT) of
the receipt of remittances on the probability of being poor. Specifically, this table shows the probability of
1
In the bivariate probit we attempted to use as instruments our two variables – distance to railroad stations in 1930 and rainfall in
1995 to 1999 – but found that these variables are part of the entire bivariate probit model.
46
Table 7. Bivariate Probit Results (random effects) for the Household Probability of Being
Poor and Receiving Remittances in Indonesia, 2000-2007
Equation: Poverty Equation: Remittances
Coefficient sd Coefficient sd
Variable
Human Capital
Number of household members over age 15 with primary
education 0.135*** 0.029 0.042 0.062
Number of household members over age 15 with junior
secondary education -0.077* 0.043 -0.082 0.156
Number of household members over age 15 with high school
to university education -0.265*** 0.042 -0.189** 0.076
Age of household head 1.31E-04 6.73E-04 -2.23E-04 1.02E-03

Sex of household head (1=male) 0.113 0.087 -0.365** 0.149
Number of children below 5 years 0.415*** 0.049 0.376** 0.188
Number of children between 6 and 18 years old 0.208*** 0.028 0.074 0.083
Bank in the village (1=yes) -0.176* 0.093 -0.170 0.168
Distance from kabupaten (district) to railroad in 1930,
adjusted
-4.59E-03*** 3.77E-04 0.011*** 0.002
Rainfall, 1995-1999 -2.89E-04*** 9.50E-05 6.00E-05 1.75E-04
Log likelihood test for Rho=0 258.31*** N/A N/A N/A
Log likelihood for model -927 N/A N/A N/A
Wald test for model 1022*** N/A N/A N/A
N 5460 N/A N/A N/A
Notes: Table reports the coefficients of a variable on the probability of household receiving international remittances and being poor. The years included in the estimation are 2000 and
2007. The model also includes a dummy for urban/rural areas and dummies for four Indonesia regions. The distance to railroad variable is adjusted in the following manner: for
households that never receive remittances the variable is the simple distance to railroad; for households that receive remittances in 2007 but not in 2000, it adds 3 to the distance to
railroad variable; for households that receive remittances in 2000 but not in 2007, it adds 2 to the distance to railroad variable; and for households that receive remittances both in 2000
and 2007, it adds 4 to the distance to railroad variable. Poverty is defined using poverty lines calculated by the World Bank as follows: for 2000, 1,099,584 rupiah/person/year at
2000 prices for urban households and 883,776 rupiah/person/year at 2000 prices for rural households; for 2007, 308,000 rupiah/person/year at 2000 prices for urban households and
235,500 rupiah/person/year at 2000 prices for rural households.
*** Significant at the 0.01 level.** Significant at the 0.05 level.* Significant at the 0.10 level.
47
Table 8. Average Treatment Effects (ATT) of the Receipt of Remittances on the Household Probability
of Being Poor, Indonesia, 2007.
Households with remittances in Households with remittances in All Households receiving

2007 and in 2000 2007 and not in 2000 remittances in 2007
Probability of being poor and

receiving remittances
0.17 0.26 0.24
Probability of being non-poor and

receiving remittances
0.24 0.36 0.33
ATT (Percent difference) -29.17% -27.78%*** -27.27%***
(-1.4) (-3.1) (-3.4)
Notes: T statistics in parenthesis. T-tests conducted using clustered standard errors and weighting observations. Probabilities estimated using a random effects bivariate probit model based on Carrasco
(2001). See Table 7 for definition of poverty line in 2000 and 2007.
*** Significant at the 0.01 level.
48
being poor and receiving remittances in 2007, , and the probability of not being poor and receiving
remittances in 2007, . The table includes all households receiving remittances in 2007, that is,
households receiving remittances in 2007 but not in 2000 and households receiving remittances in both
2000 and 2007.
Table 8 shows that the receipt of remittances clearly reduces the probability of a household being
poor. For households receiving remittances in both 2000 and 2007 the probability of being poor falls by
29.2 percent. For households receiving remittances in 2007 but not in 2000 the probability of being poor
falls by a statistically significant 27.8 percent.
3. CONCLUSION
This paper has used data from a large, panel household survey in Indonesia to analyze the impact
of international remittances on household investment and poverty. The paper has three key findings, and
one of these findings merits comment.
First, when we compare households receiving remittances in 2007 with a counterfactual situation
in which they did not receive remittances in 2007, we find that households receiving remittances increase
their marginal expenditures on one key consumption good – food – by 5.9 percent.
Second, when we compare households receiving international remittances in 2007 with a

counterfactual situation in which they did not receive remittances in 2007, we find that households
receiving remittances increase their marginal expenditures on one important investment good – education
– by a very large, 332.5 percent.
Third, we find that the receipt of remittances reduces the probability of a household being poor.
For households receiving remittances in 2007 but not in 2000 the probability of being poor falls by a
statistically significant 27.8 percent.
The second and third findings of this paper are important because they are identical to the results
of similar work on remittances, investment and poverty in other developing countries. For example, both
Adams and Cuecuecha in Guatemala (2010) and Yang (2008) in the Philippines find that households
receiving remittances tend to spend them on investment goods, like education. Since remittances are a
transitory (and possibly uncertain) type of income it is possible that households receiving remittances tend
to spend them more on investment goods (like education) than on other types of goods. These results are
important because they suggest that households receiving remittances can use them to build human capital
in remittance-receiving countries.
Similarly, with respect to remittances and poverty, the finding of this paper that remittances
reduce the likelihood of a household being poor is consistent with results in other
studies. For instance, Adams and Page (2005), Acosta et al (2006), Yang and Martinez (2006) and
Lokshin et al (2010) all find that remittances reduce poverty in developing countries.
49
However, the first finding of this paper is unexpected and is at odds with earlier work in this area.
Specifically, the finding that households receiving remittances in Indonesia spend more at the margin on
one key consumption good – food – is at odds with both the results of Adams and Cuecuecha (2010) in
Guatemala and the permanent income hypothesis that suggests that the marginal propensity to invest in
consumption goods (like food) should be less – not more – with the receipt of transitory income (like
remittances).
One possible explanation for this unexpected result is as follows. In Guatemala households
receiving international remittances receive much more in annual per capita terms from remittances than
those in Indonesia (US $365 vs. US $30 per year). As a result, mean annual per capita expenditure levels
for remittance-receiving households in Guatemala are much higher than those in Indonesia.42
Remittance-receiving households in Guatemala thus have more income and are able to devote more of
their marginal expenditures to investment in both human and physical capital: education and housing.
By contrast, households receiving international remittances in Indonesia are much poorer and thus are
anxious to improve their consumption situation first (by spending more on food) and are able to devote
only a portion of remittances to investment (by spending more on education). In the future, as
remittance-receiving households in Indonesia continue to raise their average per capita expenditures
through the receipt of international remittances, it is likely that they will reduce their marginal
expenditures on consumption (food) and devote more of their marginal expenditures to investment in
other types of investment goods (like housing).
42
Mean annual per capita expenditures for households receiving international remittances in Guatemala (in 2000) were 47.3
percent higher than those for households receiving international remittances in Indonesia (in 2007): US $1127 in Guatemala vs.
US $765 in Indonesia (nominal terms)
50
REFERENCES
Acosta, P., Calderon, C., Fajnzylber, P. and Lopez, H. 2006. Remittances and Development in
Latin America. World Economy 29, pp. 957-987.
Adams, Jr., R. 1998. Remittances, Investment and Rural Asset Accumulation in

Pakistan. Economic Development and Cultural Change, 47, pp. 155-173.
Adams, Jr., R. and Cuecuecha, A. 2010. Remittances, Household Expenditure and

Investment in Guatemala. World Development, 38, pp. 1626-1641.
Adams, Jr, R., Cuecuecha, A. and Page, J. 2008. The Impact of Remittances on Poverty and
Inequality in Ghana. World Bank Policy Research Working Paper 4732. World Bank,
Washington, DC.
Adams, Jr., R. and Page, J. 2005. Do International Migration and Remittances Reduce Poverty
in Developing Countries? World Development 33, pp. 1645-1669.
Bourguignon, F., Fournier, M., and Gurgand, M. 2004. Selection Bias Corrections
Based on the Multinomial Logit Model: Monte-Carlo Comparisons. Unpublished
DELTA working paper, France.
Carrasco, R . 2001. Binary Choice With Binary Endogenous Regressors in Panel Data:
Estimating the Effect of Fertility on Female Labor Participation. Journal of Business and
Economic Statistics, 19 (4), pp. 385-394.
Chami, R., Fullenkamp, C. and Jahjah, S. 2003. Are Immigrant Remittance Flows a Source of
Capital for Development? IMF Working Paper 03/189, International Monetary Fund,
Washington, DC.
Charlier, E., Melenberg, B. and van Soest, A. 2001. An Analysis of Housing Expenditure Using
Semiparametric Models and Panel Data. Journal of Econometrics 101 (1): 71-107.
de la Briere, B., Sadoulet, E., de Janvry, A., and Lambert, S. 2002. The Roles of Destination,
Gender and Household Composition in Explaining Remittances: An Analysis for the Dominican
Sierra, Journal of Development Economics 72, pp. 309-328.
Dubin, J. and McFadden, D. 1984. An Econometric Analysis of Residential Electric Appliance

Holdings and Consumption. Econometrica, 52, pp. 345-362. Greene, W. 2000. Econometric
Analysis. Prentice Hall: New York, NY.
Kyriazidou, E. 1997. Estimation of a Panel Data Sample Selection Model. Econometrica, 65,
pp. 1335-1364.
Lokshin, M., Bontch-Osmolovski, M., and Glinskaya, E. 2010. Work-related Migration and
Poverty Reduction in Nepal. Review of Development Economics 14, pp. 323-332.
51
Lopez-Cordova, E. 2005. Globalization, Migration and Development: The Role of Mexican
Migrant Remittances. Economia, 6, pp. 217-256.
Munshi, K. 2003. Networks in the Modern Economy: Mexican Migrants in the US Labor
Market. Quarterly Journal of Economics, 118, pp. 549-597.
Osili, U. 2004. Migrants and Housing Investments: Theory and Evidence from Nigeria.
Economic Development and Cultural Change, 52, pp. 821-849.
Prais, S. J. and Houthakker, H. S. 1971. The Analysis of Family Budgets. Cambridge:

Cambridge University Press.
Wooldridge, J. 2002. Econometric Analysis of Cross Section and Panel Data. Cambridge, MA:
MIT Press.
Woodruff, C. and Zenteno, R. 2007. Migration Networks and Micro-enterprises in Mexico.

Journal of Development Economics 82, pp. 509-528.
World Bank. 2008. The Malaysia-Indonesia Remittance Corridor. Working Paper No. 149.
Washington, DC.
World Bank. 2011.. Migration and Remittances Factbook, 2011. Washington, DC.
Yang, D. 2008. International Migration, Remittances and Household Investment: Evidence
from Philippine Migrants’ Exchange Rate Shocks. Economic Journal 118, pp. 591-630.
Yang, D. and Choi, H. 2007. Are Remittances Insurance? Evidence from Rainfall Shocks in
the Philippines. World Bank Economic Review 21, pp. 219-248
Yang, D. and Martinez, C. 2006. Remittances and Poverty in Migrants Home Areas: Evidence
from the Philippines, in: C. Ozden and M. Schiff (eds) International Migration, Remittances and
the Brain Drain, (Washington, DC: World Bank), pp. 81-12
52
Chapter 3: More or Less Consumption?
The Effect of Remittances on Filipino Household Spending Behavior
EMILY CHRISTI A. CABEGIN

University of the Philippines
and
MICHAEL ALBA
University of the Philippines
ABSTRACT. Using nationally representative data from the Philippines, this paper
examines the effect of international remittances on household spending behavior in the
Philippines. Estimating a system of demand equations and constructing counterfactuals
that are corrected for selection and endogeneity, the paper finds that international
remittances are invested in human and physical capital at a higher marginal rate than
income from other sources. Compared to a counterfactual situation without migration,
households with remittances spend at the margin 40 percent less on food and 59 percent
more on education and health care and 92 percent more on housing. However, these
selection- and endogeneity corrected findings need to be interpreted with caution as
quantile regression estimates reveal varying migration effects on household spending
behavior across the consumption distribution. There are indications of stronger negative
migration effects for households in the upper food share quantiles and weaker migration
effects for most households below the median food share.
Key words – remittances, consumption, expenditures, investment, Philippines
1. INTRODUCTION
In recent decades international remittances have emerged as the Philippines’ leading

source of external finance. In 2010, international remittances amounted to $18.8 billion or about
10 percent of GDP in the Philippines, and more than ten times the amount of foreign direct
investments of $1.7 billion. While increased interest has spurred a growing literature on the
developmental impact of international remittances, the effect of remittances on household welfare
remains ambiguous.
This paper focuses on the effect of international remittances on household consumption

behavior, an area on which literature has produced mixed results. On the one hand are studies
claiming that remittances are spent largely on consumption rather than on investment goods. For
example, Castaldo and Reilly [2007] found that households receiving international remittances in

We would like to acknowledge gratefully Richard Adams, Trang Nguyen, Daniel Mont and Ahmad Ahsan for very
substantial comments.
53
Albania had a slightly higher marginal propensity to consume food than do non-migrant
households. This corroborates earlier work that found remittances are more likely to be consumed
rather than invested [Rempel and Lobdell, 1978; Lipton, 1980; Keely and Tran, 1989; Durand
and Massey, 1992; Brown and Ahlburg, 1999].
On the other hand there is a growing literature providing evidence on the positive impact
of remittances on investment [Funkhouser, 1992; Glystos, 1993; Brown, 1994; Massey, et al.,
1987; Massey and Parrado, 1998; Adams, 1998; McCormick and Wahba, 2001; Zhao, 2001;
Taylor and Mora, 2006; Woodruff and Zenteno, 2007; Yang, 2008; Adams and Cuecuecha,
2010]. In general, these studies find that international remittances tend to get spent more at the
margin on investment goods like education, health and housing. While the impact of
remittances on education is still being debated, with some studies finding positive effects of
remittances on education (Edwards and Ureta, 2003, Adams, 2006, Mansuri, 2007) and other
studies finding insignificant or negative effects (McKenzie, 2005, Acosta, 2006), there is a
general consensus in the literature that remittances have a positive effect on housing (Osili, 2005,
Adams and Cuecuecha, 2010).
The purpose of this paper is to examine the impact of international remittances on

household consumption and investment behavior by using a demand systems model to analyze
the effects of remittances on household spending behavior. By estimating all the demand
equations simultaneously, the systems model accounts for possible correlation of the equations
through their disturbances while satisfying the adding up and homogeneity criterion in consumer
demand theory by imposing linear restrictions on the fixed parameters [Zellner, 1962; Chambers
and Nowman, 1997].
It is important in any remittances study to address the twin problems of household

selection into migration and the endogeneity of remittances. This paper therefore uses a
theoretically consistent system of demand equations to estimate the parameters of the relevant
household expenditure categories while addressing the endogeneity of remittances and sample
selection into migration. The spending behavior of migrant households on various consumption
items and human capital investment is compared with what this spending behavior would have
been in the absence of migration and remittances.1 The latter outcome is not possible to observe
for migrant households. Counterfactuals are therefore constructed from non-migrant households’
spending behavior, with the estimates corrected for endogeneity in both migration participation
and the level of remittance income. This procedure follows from the more recent strand of
migration and remittances literature that deals with the problem of endogenous migration decision
by formulating a counterfactual scenario of household outcomes (e.g. income, expenditure
patterns) without migration [Adams and Cuecuecha, 2010; Jimenez and Brown, 2008; Acosta,
Fajnzylber and Lopez, 2007].
To provide a better understanding of the sources of remittance effects, the paper employs
an Oaxaca-Blinder decomposition that divides the effects of remittances on the expenditure
budget shares into two components. The first component attributes the differences in budget
shares to the change in household incomes and the second component attributes differences to
changes in spending behavior. The latter provides a test of income fungibility. A significant
difference in spending behavior with and without migration and remittances implies that
remittances affect expenditure patterns independently of their contribution to total disposable
income.
1
In this paper migration and remittances are assumed to be synonymous, since all migrant households receive
international remittances.
54
This study also includes an analysis of the effect of migration and remittances on
different points of the expenditure share distribution. Specifically, the paper uses quantile
regressions to determine whether the effect of migration and remittances on household
consumption patterns is stronger at the median or at the extreme quantiles of the consumption
distribution.
The balance of the paper is organized as follows: section 2 specifies the econometric
model; section 3 describes data and the relevant characteristics of migrant and non-migrant
households; section 4 discusses the empirical findings and section 5 presents the conclusions.
2. EMPIRICAL MODEL
The paper estimates a system of Engel equations following an extended version of the
basic Working-Leser specification that assumes a linear relationship between budget shares and
the logarithm of total household expenditures [Working, 1943; Leser, 1963; Deaton and
Muellbauer, 1980]:
e j   0 j  x kj   j Y   j , k  1,2 ,..., K
Ej [1]
ej   Ej / Y
Ej
j
where ej denotes the share of expenditures in consumption item j to total expenditures. Eight
expenditure items are considered in the study, namely, (a) food, alcoholic beverages and tobacco;
(b) fuel, light and water; (c) education and medical care; (d) durable furniture and clothing; (e)
transportation and communication; and (f) housing; (g) household improvements and non-durable
furnishings; and (i) other consumption items.2 Y is the logarithm of total household expenditures
less taxes.3 In a migrant household, Y includes remittances. The vector x represent a set of
variables that determine household consumption behavior and includes the demographic and
education characteristics of the household head, the number of family members by age and sex
composition and location variables. The unknown parameters to be estimated are α and δ, and μ
is the error term.
It is assumed that households maximize utility in allocating their budgets over

expenditure items, and that the source of income may influence allocation preferences. In
particular, income from remittances may affect consumption behavior differently from domestic
household income. To test whether migrant households behave differently from non-migrant
households in their spending patterns, the study estimates separate regressions for the migrant (m)
and the non-migrant (n) household samples:
2 Durable furniture and equipment includes car and other vehicles, microcomputer, cellphone, audio-visual equipments,
furniture and other appliances; Transportation and communication includes transport fares, gasoline, vehicle
maintenance and repair, telephone bills, phone cards, etc; Housing includes actual and imputed house rent; Household
improvement includes house maintenance and minor repairs and materials for household operation, Non-durable
furnishings include household linen and furnishings, utensils and accessories; Other consumption items include
personal care and effects, recreation, gifts and contributions.
3
Annual income is subject to greater variability than annual expenditures, which makes the latter a better measure of
permanent income [Deaton, 1992].
55
e jm   0 jm  x m kjm   jmYm   jm [2a]
e jn   0 jn  x n kjn   jnYn   jn [2b]
If income is fungible, the differential intercepts and slopes between migrant and non-migrant
households would be equal to zero. Otherwise, remittances can be taken to have a significant
effect apart from total household income4. Given the small-scale of censorship at zero (i.e. at
most 3 percent of the sample) the estimations are derived using all households. For the relevant
expenditure categories of interest, the system of expenditure equations subject to the following
aggregation and homogeneity constraints are simultaneously estimated by three-stage least
squares (3SLS)5:
j
0j 1 [3a]
j
kj  0 ,k  1,2,...K [3b]
j
j 0 [3c]
The marginal budget share (MBS) of the jth expenditure item can be obtained as follows:
MBS j  e j   j [4]
To assess the effect of remittance income on expenditure patterns, we compare the current
spending behavior of migrant households with what that spending behavior would have been in
the absence of international remittances. We do not observe the spending behavior of migrant
households without migration. However, based on the spending patterns of non-migrant
households, we can construct the counterfactual household expenditure of migrant households in
the absence of migration.
Selectivity Bias of Migration Participation
Since migrant households cannot be assumed to be randomly drawn from the household
population, OLS estimates of the counterfactual expenditure shares are biased if the error term in
the migration selection equation is correlated with the error term in the expenditure share
equation. This paper follows more recent remittance work and derives estimates of
counterfactual spending patterns that adjust for self-selection bias.
4
Taylor and Mora [2006] tests for the effect of remittances on consumption behavior by using a migration dummy
variable and interacting this with the logarithm of total household expenditures.
5
The disturbances in the expenditure share equations include common factors and therefore likely to be correlated.
Greater efficiency is attained by joint estimation of the expenditure share equations by 3SLS. Note that the budget
shares sum up to one and at every point the system is singular. To solve the problem of singularity, J-1 equations are
estimated with the last set of parameters derived from using the cross-equation constraints in Equation [3] to satisfy the
conditions of economic theory. The parameters are estimated by iterated seemingly unrelated regression that converges
to maximum likelihood estimates which are invariant to the choice of the numeraire [Greene, 2003].
56
The selection into overseas migration is defined as:
z *j   0 j  r kj  
1 if z *j  0 [5]
zj  
0 if z j  0
*
where z is the actual outcome of a continuous latent variable z* and an indicator of whether the
household is a migrant household ( z *j >0) or a non-migrant household ( z *j =0).
We observe the following expenditure j shares for migrant households (ejm) and non-
migrant households (ejn):
e jm   0 jm  x m kjm   jmYm   jm if z j  1 [6a ]

ej  
e jn   0 jn  x n kjn   jn Yn   jn if z j  0 [6b]
A household is assumed to choose between two regimes, namely, international migration
(z=1) or no migration (z=0). Any given household can be observed in only one of these regimes.
Sample selection bias occurs as a result of the failure to observe en when z=1 or em when z =0.
Equations [5] and [6] comprise an endogenous switching regressions model that addresses the
bias introduced by selection into migration when the correlation of error terms in the selection
and expenditure equations is not equal to zero. Selection into international migration is estimated
on a set of covariates, r. The latter includes some variables in x such as the age and education of
the household head, and the age-sex composition of household members and the identifying
variable of average rainfall level from 2000 to 2003 which is in r but not in x. Given household
incomes, the spending pattern of households is taken to be independent of the level of rainfall, a
natural external factor that is assumed to influence the likelihood of migration. 6
Endogeneity of Remittance Income
Another important estimation problem that needs to be dealt with is the potential bias
introduced by the endogeneity of remittance income. Some household variables (e.g. education
of the household head) may affect expenditure patterns as well as the amount of remittances. If
remittance behavior is stochastic with an error term that is correlated with the errors in the
expenditure equations, then the endogeneity problem yields biased estimates of the effect of
international remittances on expenditures. Remittances are observed only in migrant households.
Hence, the problem of endogeneity of remittances is relevant only for the migrant household
sample.
The log of total expenditures in the migrant sample which includes remittances is
instrumented by the change in the amount of local rainfall. Rainfall variation produces an
exogenous effect on the levels of domestic and remittance incomes in migrant households [Yang
and Choi, 2007]. It is assumed that rainfall variation affects expenditure patterns only indirectly
through its effect on the level and the composition of income.
Rainfall variation is defined as the absolute value of the percent change in 2003 level
from the average of 2000-2002 levels. Rainfall data by province and year are obtained from the
6
Munshi [2003] and Lopez-Cordova [2005] also uses rainfall patterns as an instrument for overseas
migration in Mexico.
57
Philippine Atmospheric, Geographical and Astronomical Services Administration (PAGASA)7.
We contend that while the level of rainfall affects the likelihood of overseas migration, it is the
change in the amount of rainfall that determines the amount of remittance income. Equation [6a]
is adjusted as follows:
e jm   0 jm  x m kjm   jmYˆm   jm
[7]
Yˆm   0 m  x m km   mVm   m
where V is the instrumental variable for Y as defined above.
To correct the estimates for both sample selection bias and endogeneity of remittance
income, we follow the model of Mroz [1997]. In this paper, a three-step approach is employed:
(a) Stage 1 - estimate by probit the migration selection equation on covariates r; (b) Stage 2-
estimate the reduced form regression of the log of remittance income on the exogenous covariates
x, the instrument of rainfall variation (V) and the inverse of the Mills ratio generated from Stage
1; and (c) Stage 3 – estimate the structural equation of expenditure shares on the exogenous
covariates in x, the predicted value of the log of remittance income generated from Stage 2 and
the inverse of the Mills ratio from Stage 1.
The conditional expectation of budget shares for migrant households with migration and for non-
migrant households without migration is given by:
E(e jm | z j  1, Xm )  0 jm  xmkjm   jmYˆm  jmm [8]

E (e jn | z j  0, Xn )   0 jn  x n kjn   jnYn   jnn [9]
 (  0 j  r kj )   (  0 j  r kj )
m  ; n  [10]
(  0 j  r kj ) (1   (  0 j  r kj ))
where λ is Inverse Mills ratio, which when included as an independent variable in the estimation
of expenditure share equations will correct for the problem of selection on unobservable
attributes. The covariance parameter reveals the direction and magnitude of the selection bias.
Adjusting the estimates for sample selection, the conditional expectation of the budget shares can
be derived from equations [8] and [9] minus the selection term:
E(ejm | z j  1, Xm )   0 jm  x m kjm   jmYˆm
[11]
E( ejn | z j  0, X n )   0 jn  x n kjn   jnYn
[12]
The selectivity-adjusted conditional expectation of budget shares for migrant households in the
counterfactual hypothetical case is given by:
E (ejn | z j  1, X m )   0 jn  x m kjn   jnYc
[13]
Since the income of migrant households is not likely to be invariant between the pre-
migration and post-migration regimes, we use Yc in equation [13] which denotes the
counterfactual income of migrant households had migration not occurred:
7
Yang and Choi [2007] also employed changes in local rainfall as an instrument variable to correct for endogeneity of
domestic household incomes. Their findings suggest that remittances serve as insurance to income inadequacy in
migrant households.
58
Yc  E Yn | z j  1, x m , V 
[14]
The selectivity adjusted average treatment effect of migration and remittances on the
marginal spending behavior of the migrant households (ATEm) can be derived from the
conditional expectation of budget shares for migrant households in the observed and
counterfactual cases [Heckman, Tobias and Vytlacil, 2001; Wooldridge, 2002] and the coefficient
of the log of total expenditures for migrant and non-migrant households.
ATEm  E[ MBS m  MBS n | z  1, X m ]  E[ MBS m | z  1, X m ]  E[ MBS n | z  1, X m ]

 E[em | z  1, X m ]   m   E[en | z  1, X m ]   n 
 E[em  en | z  1, X m ]   m   n 
[15]
Selectivity-adjusted ATEm measures the difference in the conditional marginal budget
shares for migrant households relative to what those budget shares would have been in the
absence of migration. Thus, migrant households spend more for expenditure item j than when
migration had not occurred if ATEm for that expenditure item is positive, and they spend less if
ATEm is negative.
Oaxaca-Blinder decomposition of ATEm
Equation [15] is the sum of two differences, namely, the difference in the selectivity-
adjusted conditional expectation of expenditure shares of migrant households in the presence of
migration [Equation 11] and the absence of migration [Equation 13] and the difference in income
effects on spending behavior between migrant and non-migrant households.
The first term in Equation [15] is disaggregated further into a part attributed to higher
incomes of households who choose migration and the behavioral component due to differences in
coefficients. Hence, the selectivity-adjusted average treatment effect on marginal budget shares
for migrant households can be decomposed into three components:
ATE m  E[em  en | z  1, X m ]   m   n 

 E[em | z  1, X m ]  E[en | z  1, X m ]   m   n 
 
 E[em | z  1, x m , Yˆm ]  E[en | z  1, x m , Yc ]   m   n 

 E[em | z  1, x m , Yˆm ]  E[em | z  1, x m , Yc ]
  

a
 E[em | z  1, x m , Yc ]  E[en | z  1, x m , Yc ]   m   n 
  
b c
[16]
where:
a = remittance effect thru a change in household income
b = differences in the effects of household characteristics on spending behavior
c = differences in income effects on spending behavior
The first effect, a, is derived from subtracting the budget shares conditional on
characteristics of households that chose migration (z=1) using predicted migrant household
income, Ym, with that using the counterfactual income, Yc. Hence, it reflects the change in
conditional budget shares due to the increase in household income from migration. The second
effect, b, reflects the differential effect of household characteristics on spending behavior
59
between the two regimes of with and without migration and provides a test of the fungibility of
income. We perform the Wald F-test to determine whether the expenditure functions are
different for migrant and non-migrant households:
( ESS p  ( ESS m  ESS n )) / g
F , F( g , N  2g )
( ESS m  ESS n ) /( N  2 g ) [17]
where ESS refers to the error sum of squares for the pooled (p), migrant (m) and non-migrant
household samples (n), N is the total number of cases in the pooled sample and g is the number of
parameters. This tests the null hypothesis that regression coefficients are the same as against the
alternative hypothesis that some or all of the regression coefficients are different between migrant
and non-migrant households. If the null hypothesis is not rejected, then this indicates fungibility
of income and signifies the irrelevance of income source as a factor for consumption patterns.
Finally, c denotes the differential effect of household income on spending behavior.

While a captures the effect of remittances through its contribution to household income, c
imparts a more direct effect of remittances on the marginal propensity to spend on expenditure
category j and determines whether households spend remittances differently from other sources of
income.
Quantile regression
The conventional analysis of the ATEm using differences in the conditional expectation of
budget shares does not provide any information on the behavior of migrant households with either
very high or very low expenditure shares. The paper allows the determination of migration and
remittance effects across the full range of the expenditure share distribution. This is done by
estimating a quantile regression, a method that expresses quantiles of conditional distribution of a
dependent variable as functions of observed covariates [Koenker and Basett, 1978; Koenker and
Hallock, 2001; Koenker and Xiao, 2002]. Utilizing the entire sample, a quantile regression
circumvents the problem of truncation [Heckman, 1979].
Quantiles are relative measures. A household in the qth quantile of a food share
distribution, for example, means that its food share budget is higher than q proportion of
households and lower than 1-q proportion of households. Quantile regressions will allow us to
identify whether the average treatment effect is significantly larger at the lower, middle or higher
quantiles of the expenditure share distribution. In this paper, 19 regression estimates for each
expenditure item are calculated corresponding to 19 percentiles of the expenditure shares,
namely, 5th, 10th, 15th, 20th,…..,80th, 85th, 90th, 95th percentiles.
Analogous to minimizing the residual sum of squares in OLS, a quantile regression

minimizes the weighted sum of absolute deviations [Chamberlain, 1994; Deaton, 1997; Arias,
Hallock and Sosa-Escudero, 2001],
min 
i
ji   0 ji  x i kji   ji Yi wi
2q, if  ji  0
wi  
2  2q, if  ji  0 , q(0,1) [18]
The quantile treatment effect on the treated or on the migrant households in this
application (QTEm) is given by:
QTE m  Fem1 (q | z  1, X m )  Fen1 (q | z  1, X m )
[19]
60
1
where Fe ( q | z  1, X m ) equals the inverse of the conditional cumulative distribution. In the
paper, this refers to the conditional quantile for percentile q of the expenditure share (e)
distribution, which is distinguished for migrant households in the presence of international
migration (em) and when migration has not occurred (en). In this application of quantile
regression, the test-statistics are based on bootstrapped standard errors with 100 replications.
3. DATA
This study uses the 2003 merged data sets of the Labor Force Survey and the Family Income
and Expenditure Survey. These nationally representative surveys contain information on annual
household income by various sources, annual household expenditure on detailed items, and
characteristics of the household and the household members. Migrant households are defined as
those with a household member that is an overseas contract worker at the time of the survey and
who received international remittances. Non-migrant households are those which did not receive
any remittance (either local or international) and had no overseas contract worker. The total
sample used in the study included 16,259 non-migrant households with no international or
domestic remittances and 1,665 migrant households with positive remittances from abroad.
Table 1 provides the descriptive statistics used in the study including the average budget
shares for the eight expenditure categories. Compared to non-migrant households, households
receiving remittances have a lower average expenditure share for food, just about the same
expenditure share for utilities (fuel, light and water) and a higher share for all other items. On
average, migrant households spend about 42 percent of their total.
Table 1: Descriptive Statistics
Migrant Non-migrant
Variable (N=1,665) (N=16,259)
Mean std dev mean std dev
Dependent variables:
Average share to total expenditures
Food, beverage and tobacco 0.420 0.130 0.576 0.141
Fuel, light and water 0.068 0.028 0.066 0.030
Education and medical care 0.082 0.093 0.038 0.058
Durable furniture 0.069 0.086 0.044 0.058
Transportation and communication 0.085 0.056 0.050 0.045
Housing 0.124 0.083 0.096 0.073
Home improvements and non-durable
furnishings 0.037 0.045 0.027 0.031
Other consumption items 0.114 0.068 0.105 0.059
Independent variables
Age 46.930 12.645 42.947 12.293
Male 0.673 0.469 0.993 0.083
Elementary graduate 0.139 0.346 0.202 0.401
Some high school 0.108 0.311 0.130 0.337
High school graduate 0.253 0.435 0.192 0.394
Some college 0.191 0.393 0.102 0.303
61
Migrant Non-migrant
Variable (N=1,665) (N=16,259)
Mean std dev mean std dev
College graduate 0.213 0.410 0.086 0.280
log of total expenditures 12.068 0.623 11.276 0.706
N 0-6 yrs old, female 0.324 0.644 0.455 0.727
N 7-14 yrs old, female 0.429 0.710 0.555 0.830
N 15-24 yrs old, female 0.589 0.875 0.464 0.740
N 25-64 yrs old, female 1.469 0.870 0.996 0.506
N 65+ yrs old, female 0.126 0.339 0.063 0.248
N 0-6 yrs old, male 0.323 0.601 0.485 0.744
N 7-14 yrs old, male 0.462 0.730 0.587 0.859
N 15-24 yrs old, male 0.585 0.856 0.526 0.833
N 25-64 yrs old, male 1.365 0.809 1.172 0.788
N 65+ yrs old, male 0.153 0.496 0.105 0.543
Rural residence 0.414 0.493 0.628 0.483
expenditures on food, while non-migrant households spend 57 percent of their expenditures on

food. Migrant households spend 4 percentage points more on education and medical care, and 3
percentage points more on housing, and on durable furnitures.
Compared to non-migrant households, migrant households are older, more urban, less
likely headed by males and more likely to be headed by the better educated. They also have a
lower number of very young children and higher number of members in the working ages.
Migrant households also have higher total expenditures than non-migrant households.
4. FINDINGS
The first stage reduced form probit estimates (Appendix 1) use the pooled sample of
migrant and non-migrant households to estimate the probability of households participating in
international labor migration. This first stage equation also estimates the inverse Mill’s ratio
which will be included as a regressor in both the second-stage income regression and the third-
stage expenditure share equation.
Results from the first stage probit equation indicate that households with better educated
heads are more likely to produce international migrants, while those with children in the pre-
school ages are less likely to produce such migrants.
The second-stage OLS estimates (Appendix 2) suggest that households headed by males
and better educated individuals and that households residing in urban areas have higher
household incomes. In this second-stage equation, the instruments in both the selection and
income equations are positive and significant at the .05 level of significance.
The Engel curves for the eight expenditure categories (adjusted for selection into
migration and endogeneity of remittances) are presented in Appendix 3 for non-migrant
households and in Appendix 4 for migrant households. There is evidence that household
participation into migration is determined by unobserved characteristics that also affect household
spending in four expenditure categories for non-migrant households and three expenditure
categories for migrant households. For food, the covariance parameter is significantly positive
for migrant households and positive but not significant for non-migrant households. The presence
62
of selection bias overestimates the food budget share for migrant households and underestimates
the negative ATEm for food.
The selection term coefficients in the housing share equations are negative and significant
for both migrant and non-migrant households, with stronger selectivity bias for migrant
households. Selectivity produces a downward bias in both the observed and the counterfactual
marginal budget share for housing in the migrant households, but to a lesser extent for the
counterfactual case. The differential selection effect is negative and underestimates the positive
ATEm for housing.
For most of the rest of the expenditure items, sample selection appears to have
overestimated the average treatment effect for migrant households. With regard to human capital
expenditures, the selection parameter in the budget share equation is not significant for the
migrant households but significantly negative for the non-migrant households. This results in a
positive differential selection effect that widens the ATEm for education and medical care. Hence,
failure to control for selection into migration in this paper underestimates the magnitude of the
ATEm for food and for housing and overestimates that for education and medical care.
Selectivity and Endogeneity-Corrected OLS estimates
Estimates of the selectivity- and endogeneity-corrected marginal budget shares and

average treatment effects (ATEm) are presented in Table 2. The ATEm figures imply a large
reconfiguration of spending patterns due to migration. Had migration not occurred, migrant
households would have spent much more on food and home improvements. While food remains
the dominant component of the migrant household budget, the marginal budget share devoted to
food in migrant households (0.262) is much lower than what that share would have been without
migration (0.433). Compared to the no migration situation, migrant households with remittances
reduce their marginal budget shares on food and increase their marginal budget shares on
education and medical care, and housing Migrant households with remittances spend at the
margin 40 percent less on food, and 59 percent more on education and medical care and 92
percent more on housing than when migration had not occurred. These are important changes
because they suggest that with the receipt of remittances households spend less at the margin on
consumption goods (like food) and more at the margin on human and capital investment items
(like education and housing)
63
Table 2: Selectivity- and endogeneity-corrected marginal budget shares and average
treatment effect for migrant households by expenditure category, Philippines, 2003†
Migrant household
Expenditure category Non-migrant Estimated ATEm
household Counterfactual MBS level %
MBS change
Food, beverage and 0.454*** 0.433*** 0.262*** -0.171 -39.5

tobacco (0.002) (0.002) (0.041)
Fuel, light and water 0.056*** 0.062*** 0.092*** 0.030 49.5

(0.000) (0.001) (0.010)
Education and medical 0.061*** 0.056** 0.088*** 0.033 58.9

care (0.001) (0.001) (0.032)
Durable goods 0.078*** 0.085*** 0.080*** -0.005 -5.4

(0.001) (0.001) (0.030)
Transportation and 0.073*** 0.083*** 0.092*** 0.009 11.4

communication (0.001) (0.001) (0.020)
Housing 0.115*** 0.119*** 0.228*** 0.109 91.8

(0.001) (0.001) (0.028)
Home improvements and 0.038*** 0.040*** 0.024*** -0.017 -41.4

non-durable furnishings (0.000) (0.001) (0.016)
Other consumption items 0.125*** 0.124*** 0.134*** 0.011 8.6

(0.001) (0.001) (0.024)
†
Standard errors in parenthesis; *Significant at .10 level; ** Significant at .05 level; ***Significant at .01 level.
Table 3 presents the results of the Oaxaca-Blinder type decomposition of the ATEm along
with the Wald statistics of structural change between the migrant and the non-migrant
expenditure functions that provides a test of income fungibility. The Wald test results reject the
equality of expenditure determinants between migrant and non-migrant households at the .01
level of significance for 6 expenditure items, and at the .05 level of significance for the other 2
expenditure items. The results suggest that income is not fungible.
64
Table 3: Decomposition of Average Treatment
Effect for Migrant Households (ATEm) by expenditure category, Philippines, 2003
Behavioral component:
Differential effect on
spending behavior of:
Expenditure category HH
Change in characteristic Wald F-
Income level s HH income statistic†
a b c ATEm
Food, beverage and tobacco -0.107 -0.092 0.028 -0.171 4.68***

Fuel, light and water 0.037 -0.047 0.041 0.030 2.91***
Education and medical care 0.015 0.030 -0.012 0.033 3.42***
Durable goods 0.004 0.020 -0.028 -0.005 3.05***
Transportation and
communication 0.025 -0.014 -0.001 0.009 1.62**
Housing 0.052 0.033 0.024 0.109 3.10***
Non-durable furnishings -0.008 0.008 -0.017 -0.017 1.63**
Other consumption items -0.017 0.064 -0.036 0.011 3.44***
†
Wald F-test with robust variances estimates; *Significant at .10 level; ** Significant at .05 level; ***Significant at .01 level
In Table 3 the change in ATEm is divided into two parts: (a) the part caused by the
change in the income level (a); and (b) the behavioral component that reflects the difference in
effects of characteristics on the marginal propensity to spend for expenditure item j. In the table
the behavioral part is divided further into two components, namely: (a) the difference in the
budget shares that is ascribed to differences in the coefficients of observable characteristics of
migrant and non-migrant households (b); and (b) the difference in the income effect on spending
behavior (c).1
Households who choose migration are estimated to have almost tripled their household
incomes compared to a counterfactual income in the absence of migration. This increase in
income accounts for close to two-thirds (63 percent) of the negative ATEm for food. The rest of
the ATEm for food is due to the unexplained behavioral component. By contrast, close to half of
the positive ATEm for housing and for education and medical care is explained by higher incomes
of households who choose migration. For expenditures on these human capital investments, the
rest of the positive impact is due largely to the unexplained structural effects. The coefficient of
remittance income on human capital expenditures is both positive in both migrant and non-
migrant households but with stronger effects for non-migrant households. The reverse is found
for housing expenditures where migrant households demonstrate larger positive income effects
than that for non-migrant households.
Quantile Regression Estimates
This section relaxes the OLS method assumption of a constant effect of household
characteristics on spending pattern across the expenditure share distribution. Using quantile
1
Refer to Appendices 3 and 4 for the coefficients of the estimated expenditure equations for migrant and
non-migrant households.
65
regressions, the average treatment effects for migrant households are estimated over the various
quantiles of the expenditure share distribution (QTEm). The QTEm is negative throughout the food
expenditure share distribution with more pronounced negative migration effects in the upper
quantiles and less marked effects in the lower quantiles (Figure 1).2 Migrant households with
food shares between the 70th and the 90th percentile have a QTEm of between -0.26 and -0.29.
This is way below the ATEm of -0.17. Migration effects are much less for households with food
shares below the 30th percentile, with a QTEm of between -0.09 to -0.10. Finally, households in
the 95th percentile have the least QTEm of -0.04.
Positive migration effects are generally stable around the ATEm of 0.11 for housing and at
0.03 for utilities (Figures 2 and 3). The migration effects for housing expenditures are slightly
higher for households below than those above the median. There are also indications of stronger
migration effects at the more extreme distribution for utilities expenditures. The reverse appears
to be true for education and medical care. QTEm is positive and well above the ATEm of 0.03 for
households with education shares at the median up to the 75th percentile (Figure 4). Households
with education shares within this middle range spend about 8-13 pesos more for every 100 pesos
increase in household income relative to when migration had not occurred. QTEm is close to zero
for households with education shares below the 30th percentile and is negative for households at
the 95th percentile.
A comparison of the QTEm from the OLS-generated ATEm reveals varying migration
impacts across the different quantiles of the budget share distribution. The OLS results mostly
capture the spending behavior of households in the middle range and are less reflective of the
lower and upper tails of the distribution. Hence, OLS estimates have to be interpreted with a
caveat in mind that the focus on the conditional mean could mask the potential heterogeneity of
migration effects across the distribution.
2
Selectivity-adjusted quantile regression estimates for 19 quantiles,( i.e., .05, 1,…, .95) on expenditures
shares for migrant and non-migrant households are available from the authors upon request.
66
Figure 1a. OLS and Quantile Regression Estimates of Figure 1b. OLS and Quantile Regression Estimates
Marginal Budget Shares (MBS), Food Expenditures, 2003 of Average Treatment Effect for the Migrant
Households (ATEm), Food Expenditures, 2003
0.7
0
0.6
0.5
-0.05
Log of Expenditures
0.4 -0.1
MBS
0.3 -0.15
0.2 -0.2
0.1 -0.25
0.0 -0.3
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Quantile Quantile
Migrant H_Qtile Counterfactual_Qtile
Migrant_OLS Counterfactual_OLS Quantile OLS
Figure 2a. OLS and Quantile Regression Estimates of Figure 2b. OLS and Quantile Regression Estimates of
Marginal Budget Shares (MBS), Average Treatment Effect for the Migrant Households
Utilities Expenditures, 2003 (ATEm), Utilities Expenditures, 2003
0.20
0.12
0.16
0.1
Log of Expenditures
0.12 0.08
MBS
0.06
0.08
0.04
0.04
0.02
0.00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Quantile
Quantile
Migrant OLS Counterfactual OLS Quantile OLS
Marginal Budget Shares (MBS), of Average Treatment Effect for the Migrant
Housing Expenditures, 2003 Households (ATEm), Housing Expenditures, 2003
0.4
0.36 0.20
0.32
Log of Expenditures
0.28 0.16
0.24
MBS
0.2 0.12
0.16
0.08
0.12
0.08 0.04
0.04
0 0.00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Quantile Quantile
Migrant H_OLS Counterfactual_OLS Quantile OLS
67
Education and Medical Care Expenditures, 2003 Households (ATEm), Education and Medical Care
0.31 Expenditures, 2003
0.27 0.14
0.23
MBS
0.10
Log of Expenditures
0.19
0.06
0.15
0.11 0.02
0.07
-0.02
0.03
-0.06
-0.01
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Quantile
Quantile
Migrant OLS Counterfactual OLS Quantile OLS zero
Marginal Budget Shares (MBS), Average Treatment Effect for the Migrant
Transportation Expenditures, 2003 Households (ATEm), Transportation
0.28
Expenditures, 2003
0.24 0.09
0.20
Log of Expenditures
0.07
0.16
MBS
0.05
0.12 0.03
0.08 0.01
0.04 -0.01
0.00 -0.03
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
Quantile Quantile
Migrant H_OLS Counterfactual_OLS Quantile OLS zero
Durable Goods Expenditures, 2003 Households (ATEm), Durable Goods
0.42 Expenditures, 2003
0.36
0.00
0.30
Log of Expenditures
-0.02
0.24
MBS
-0.04
0.18
-0.06
0.12 -0.08
0.06 -0.10
0.00 -0.12
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Quantile Quantile
68
Non-durable goods Expenditures, 2003 (ATEm), Non-Durable Goods Expenditures, 2003
0.02
0.2
0.16 -0.02
Log of Expenditures
0.12 -0.06
MBS
0.08 -0.10
0.04 -0.14
0
-0.18
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
Quantile
Quantile
Other Consumption Expenditures, 2003 (ATEm), Other Consumption Expenditures, 2003
0.28
0.04
0.24 0.03
0.02
0.2
Log of Expenditures
0.01
0.16 0
MBS
-0.01
0.12
-0.02
0.08 -0.03
-0.04
0.04
-0.05
0 -0.06
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Quantile Quantile
5. CONCLUSION
This paper assesses the average treatment effects (ATEs) of migration and remittances on
the spending behavior of international migrant households in the Philippines by testing the
difference in observed spending patterns of migrant households with the counterfactual spending
patterns of these households had migration not occurred. The estimated unobserved
counterfactuals are based on the observed consumption behavior of non-migrant households, with
estimates corrected for self-selection bias in migration participation and the endogeneity of
remittances.
The ATEm for the various expenditure items reveal that when compared to the no
migration situation, households that choose migration spend at the margin 40 percent less on food
and 59 percent more on education and medical care and 92 percent more on housing. The
decomposition of the migration effect into the behavioral component and that part due to higher
incomes show that the higher incomes coming from remittances do not totally account for
changes in the ATEm for most expenditure items. The structural component accounts for close to
two-fifths of the changes in ATEm for food and for more than half of the changes in ATEm for
education and health care, and housing. These findings imply that households spend remittances
differently from other sources of income, thus violating the assumption that income is fungible.
The results of this study buttress findings by Yang [2008], who found that positive
exchange rate shocks in the Philippines lead migrant households to increase their expenditures on
education. The findings of this study are also similar to those of Adams and Cuecuecha [2010],
who found that households receiving international remittances in Guatemala spend less at the
margin on food, and more on education and housing than when migration had not occurred.
69
There are several reasons why international migrant households tend to spend less on
consumption goods (like food) and more on human and physical investments (like education,
health and housing). First, migrant workers abroad may “earmark” remittances for spending on
education, health and housing at home. Second, international migration may shift the preferences
of migrant households for more education as they learn that overseas migration tends to yield
higher returns to schooling. Finally, from a more theoretical perspective, remittances from
international migration may be viewed largely as transitory income that is more likely to be saved
or invested relative to permanent income [Friedman, 1957; Hall and Mishkin, 1982]. Since the
marginal propensity to invest out of transitory income is higher than permanent income, migrant
households may use income from a transitory source like remittances to make more investments
in education, health and housing.
The estimates of the ATEm for expenditure items by the selection- and endogeneity-
corrected OLS regression tend to hide the heterogeneity of the migration effects across the
distribution of budget shares. Results of the QTEm indicate deviation from the ATEm in terms of
magnitude and direction for most of the expenditure items and particularly for households in the
more extreme tails of the budget share distribution. There are indications of stronger negative
migration effects for households in the upper food share quantiles up to the 90th percentile and
weaker migration effects for households in the lower food share quantiles as well as those in the
95th percentile. There appears to be stronger positive migration effects at the more extreme
distribution for utilities expenditures while the opposite seems indicated for education and
medical care. The migration effect is virtually zero for households with the lowest education
shares and is negative for the biggest spenders.
Evidence of migration being associated with increased investments in both housing and
education ascribe to migration an important role in improving not only asset holdings of migrant
households but also increasing the potential for higher productivity of assets through enhanced
human capital. However, findings based on the selection- and endogeneity-corrected OLS
regressions tell an incomplete story. Suggestions for policy reforms need to be cognizant of the
fact that households at different points of the expenditure share distribution respond differently to
migration and remittances.
70
Appendix 1: Probit estimates of migration participation
Philippines, 2003†
Variable Coefficient se
Age -0.0099 0.0087

Age squared 0.0002 * 0.0001
Male -2.3723 *** 0.0654
Elementary graduate 0.2724 *** 0.0556
Some high school 0.4568 *** 0.0604
High school graduate 0.6288 *** 0.0533
Some college 0.6975 *** 0.0589
College graduate 0.7176 *** 0.0613
N 0-6 yrs old, female -0.1043 *** 0.0259
N 7-14 yrs old, female -0.0781 *** 0.0205
N 15-24 yrs old, female 0.1096 *** 0.0193
N 25-64 yrs old, female 0.5281 *** 0.0302
N 65+ yrs old, female 0.4647 *** 0.0574
N 0-6 yrs old, male -0.1340 *** 0.0250
N 7-14 yrs old, male -0.0292 0.0198
N 15-24 yrs old, male 0.0247 0.0189
N 25-64 yrs old, male 0.1006 *** 0.0161
N 65+ yrs old, male 0.0265 0.0240
Rural residence -0.0338 0.0338
Rainfall level 0.00004 ** 0.0000
Constant -0.3217 0.2127
†N=17924; Log L= -3793.05; Pseudo R2=.3155; Wald chi2(20))=2357.25;
*Significant at .10 level; ** Significant at .05 level; ***Significant at .01 level.
71
Appendix 2: OLS regression estimates of logarithm of total expenditures,
Philippines, 2003†
Non-
Variable Migrant Migrant
Household Household
Age 0.020** 0.030***
(0.009) (0.002)
Age squared -0.0002** -.0003***
(0.0001) (0.000)
Male 0.338*** 0.649***
(0.109) (0.146)
High school graduate 0.185*** 0.308***
(0.038) (0.013)
Some college 0.348*** 0.547***
(0.044) (0.018)
College graduate 0.630*** 0.993***
(0.044) (0.021)
N 0-6 yrs old, female 0.049*** -.0002
(0.018) (0.006)
N 7-14 yrs old, female 0.045*** 0.045***
(0.017) (0.005)
N 15-24 yrs old, female 0.079*** 0.091***
(0.015) (0.007)
N 25-64 yrs old, female 0.031 0.130***
(0.024) (0.017)
N 65+ yrs old, female 0.020 0.073***
(0.045) (0.023)
N 0-6 yrs old, male 0.037* 0.008
(0.020) (0.006)
N 7-14 yrs old, male 0.048*** 0.035***
(0.017) (0.005)
N 15-24 yrs old, male 0.087*** 0.092***
(0.015) (0.005)
N 25-64 yrs old, male 0.059*** 0.083***
(0.018) (0.007)
N 65+ yrs old, male 0.109*** 0.039***
(0.024) (0.011)
Rural residence -0.255*** -0.387***
(0.027) (0.009)
Change in Rainfall 0.005*** 0.001***
(0.0009) (0.000)
Inverse of the Mills ratio -0.331*** -0.603***
(0.072) (0.112)
Constant 11.184*** 9.484***
(0.211) (0.160)
2
†(Migrant) N=1665; R =0.4027; F(19,1645)=59.5;
(Non-migrant) N=16259; R2=0.506; F(19,16239)=781.5;
*Significant at .10 level;** Significant at .05 level; ***Significant at .01 level.
72
Appendix 3: Selectivity-adjusted 3SLS estimates on expenditure shares, by category,
Non-migrant households, Philippines, 2003†
Food, alcohol and Education and Furniture and

tobacco Fuel, light and water medical care clothing
Variables [1] [2] [3] [4]
Coeff se coeff se coeff se coeff se
Constant 1.9634 *** 0.0277 0.1512 *** 0.0085 -0.3276 *** 0.0155 -0.2464 *** 0.0163
Age 0.0006 0.0005 0.0002 0.0001 0.0020 *** 0.0003 -0.0014 *** 0.0003
Age squared 0.0000 *** 0.0000 0.0000 0.0000 0.0000 *** 0.0000 0.0000 *** 0.0000
Male -0.0200 0.0201 -0.0009 0.0062 0.0254 ** 0.0113 -0.0244 ** 0.0118
Elementary graduate -0.0080 *** 0.0022 0.0035 *** 0.0007 -0.0007 0.0013 0.0030 ** 0.0013
Some High School -0.0095 *** 0.0027 0.0016 ** 0.0008 0.0013 0.0015 0.0032 ** 0.0016
High school graduate -0.0206 *** 0.0028 0.0067 *** 0.0009 0.0027 * 0.0016 0.0012 0.0016
Some college -0.0415 *** 0.0035 0.0072 *** 0.0011 0.0086 *** 0.0019 0.0020 0.0020
College graduate -0.0712 *** 0.0040 0.0050 *** 0.0012 0.0157 *** 0.0022 -0.0071 *** 0.0024
ln (disposable income) -0.1263 *** 0.0015 -0.0077 *** 0.0005 0.0252 *** 0.0008 0.0313 *** 0.0009
HH members 0-6,
female 0.0150 *** 0.0011 -0.0002 0.0003 -0.0029 *** 0.0006 -0.0024 *** 0.0006
HH members 7-14,
female 0.0152 *** 0.0010 -0.0002 0.0003 0.0017 *** 0.0005 -0.0026 *** 0.0006
HH members 15-24,
female 0.0087 *** 0.0012 -0.0017 *** 0.0004 0.0054 *** 0.0007 -0.0010 0.0007
HH members 25-64,
female 0.0083 *** 0.0030 0.0007 0.0009 -0.0067 *** 0.0017 0.0018 0.0018
HH members 65+,
female 0.0036 0.0041 0.0013 0.0013 0.0039 * 0.0023 -0.0035 0.0024
HH members 0-6, male 0.0177 *** 0.0011 -0.0005 0.0003 -0.0026 *** 0.0006 -0.0031 *** 0.0006
HH members 7-14, male 0.0192 *** 0.0009 -0.0008 *** 0.0003 0.0009 * 0.0005 -0.0024 *** 0.0005
HH members 15-24,
male 0.0153 *** 0.0010 -0.0018 *** 0.0003 0.0033 *** 0.0005 -0.0013 ** 0.0006
73
Food, alcohol and Education and Furniture and
tobacco Fuel, light and water medical care clothing
Variables [1] [2] [3] [4]
Coeff se coeff se coeff se coeff se
HH members 25-64,
male 0.0184 *** 0.0011 -0.0012 *** 0.0003 -0.0018 *** 0.0006 -0.0009 0.0006
HH members 65+, male 0.0188 *** 0.0017 -0.0027 *** 0.0005 -0.0005 0.0009 -0.0009 0.0010
Rural -0.0052 *** 0.0017 -0.0131 *** 0.0005 0.0106 *** 0.0010 0.0142 *** 0.0010
Inverse of the Mills ratio 0.0075 0.0146 -0.0036 0.0045 -0.0160 ** 0.0082 0.0279 *** 0.0086
†
Parameters are estimated by iterated SUR which converges to ML estimates; N=16,259; [1] R2=.975; [2] R2=.836; [3] R2=.415; [4] R2=.418;
* Significant at .10 level; ** Significant at .05 level; *** Significant at .01 level
74
Appendix 3: Selectivity-adjusted 3SLS estimates on expenditure shares, by category,
Non-migrant households, Philippines (cont’n), 2003†
Home operations and

improvements, and
Transportation and non-durable Other consumption
Variables communication Housing furnishing goods
[5] [6] [7] [8]
coeff se coeff se coeff se coeff se
Constant -0.2174 *** 0.0112 -0.1133 *** 0.0195 -0.0558 *** 0.0090 -0.1541 *** 0.0162
Age 0.0001 0.0002 -0.0010 *** 0.0003 -0.0004 *** 0.0002 0.0000 0.0003
Age squared 0.0000 0.0000 0.0000 *** 0.0000 0.0000 *** 0.0000 0.0000 0.0000
Male 0.0004 0.0082 0.0081 0.0142 -0.0086 0.0066 0.0200 * 0.0118
Elementary graduate 0.0006 0.0009 0.0018 0.0016 -0.0019 ** 0.0007 0.0018 0.0013
Some High School 0.0042 *** 0.0011 -0.0016 0.0019 -0.0010 0.0009 0.0018 0.0016
High school graduate 0.0055 *** 0.0011 0.0043 ** 0.0020 -0.0038 *** 0.0009 0.0040 ** 0.0016
Some college 0.0109 *** 0.0014 0.0061 ** 0.0024 -0.0030 *** 0.0011 0.0099 *** 0.0020
College graduate 0.0211 *** 0.0016 0.0034 0.0028 0.0034 *** 0.0013 0.0298 *** 0.0023
ln (disposable income) 0.0241 *** 0.0006 0.0231 *** 0.0011 0.0093 *** 0.0005 0.0209 *** 0.0009
HH members 0-6,
female -0.0038 *** 0.0004 -0.0041 *** 0.0008 0.0004 0.0004 -0.0020 *** 0.0006
HH members 7-14,
female -0.0040 *** 0.0004 -0.0051 *** 0.0007 -0.0010 *** 0.0003 -0.0040 *** 0.0006
HH members 15-24,
female 0.0004 0.0005 -0.0083 *** 0.0008 -0.0018 *** 0.0004 -0.0017 ** 0.0007
HH members 25-64,
female 0.0024 ** 0.0012 -0.0078 *** 0.0021 -0.0003 0.0010 0.0016 0.0017
HH members 65+,
female -0.0023 0.0016 -0.0014 0.0029 0.0004 0.0013 -0.0020 0.0024
HH members 0-6,
male -0.0049 *** 0.0004 -0.0052 *** 0.0008 -0.0001 0.0004 -0.0013 ** 0.0006
HH members 7-14, -0.0036 *** 0.0004 -0.0062 *** 0.0006 -0.0012 *** 0.0003 -0.0058 *** 0.0005
75
Home operations and
improvements, and
Transportation and non-durable Other consumption
[5] [6] [7] [8]
male
HH members 15-24,
male -0.0016 *** 0.0004 -0.0083 *** 0.0007 -0.0020 *** 0.0003 -0.0036 *** 0.0006
HH members 25-64,
male -0.0021 *** 0.0004 -0.0078 *** 0.0007 -0.0005 0.0003 -0.0041 *** 0.0006
HH members 65+,
male -0.0010 0.0007 -0.0087 *** 0.0012 -0.0015 *** 0.0005 -0.0036 *** 0.0010
Rural -0.0030 *** 0.0007 -0.0233 *** 0.0012 0.0031 *** 0.0006 0.0167 *** 0.0010
Inverse of the Mills
ratio 0.0006 0.0059 -0.0264 *** 0.0103 0.0095 ** 0.0047 0.0005 0.0085
†
Parameters are estimated by iterated SUR which converges to ML estimates; N=16,259; [5] R2=.672; [6] R2=.693; [7] R2=.440;
* Significant at .10 level; ** Significant at .05 level; *** Significant at .01 level
76
Appendix 4: Selectivity-adjusted IV 3SLS estimates on expenditure shares, by category,
Migrant households, Philippines, 2003†
Food, alcohol and tobacco Education and Furniture and

[1] Fuel, light and water medical care clothing
Variables [2] [3] [4]
Constant 1.5073 *** 0.4653 -0.3222 *** 0.1095 -0.1964 0.3604 0.1146 0.3425
Age -
0.0016 0.0019 -0.0002 0.0005 0.0021 0.0015 0.0038 *** 0.0014
Age squared 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 *** 0.0000
Male -0.0106 0.0357 -0.0236 *** 0.0084 -0.0254 0.0277 0.0166 0.0263
Elementary graduate -0.0132 0.0129 0.0035 0.0030 0.0184 * 0.0100 0.0055 0.0095
Some High School -0.0261 * 0.0148 0.0007 0.0035 0.0349 *** 0.0114 0.0267 ** 0.0109
High school graduate -0.0330 ** 0.0168 -0.0033 0.0040 0.0394 *** 0.0131 0.0112 0.0124
Some college -0.0423 ** 0.0212 -0.0101 ** 0.0050 0.0501 *** 0.0164 0.0167 0.0156
College graduate -0.0697 ** 0.0295 -0.0260 *** 0.0070 0.0584 *** 0.0229 0.0190 0.0217
ln (disposable income) -0.0979 ** 0.0409 0.0335 *** 0.0096 0.0135 0.0317 0.0032 0.0301
HH members 0-6, -
female 0.0204 *** 0.0051 -0.0008 0.0012 -0.0065 * 0.0040 0.0030 0.0038
HH members 7-14, -
female 0.0262 *** 0.0046 -0.0024 ** 0.0011 -0.0021 0.0035 0.0052 0.0034
HH members 15-24, -
female 0.0131 *** 0.0049 -0.0037 *** 0.0012 0.0118 *** 0.0038 0.0063 * 0.0036
HH members 25-64, -
female 0.0226 *** 0.0073 0.0014 0.0017 -0.0038 0.0056 0.0059 0.0053
HH members 65+, -
female 0.0249 ** 0.0117 0.0036 0.0028 0.0137 0.0091 0.0124 0.0086
HH members 0-6, -
male 0.0186 *** 0.0054 0.0018 0.0013 -0.0101 ** 0.0042 0.0033 0.0040
HH members 7-14, 0.0267 *** 0.0043 -0.0037 *** 0.0010 -0.0004 0.0034 0.0029 0.0032
77
Food, alcohol and tobacco Education and Furniture and
[1] Fuel, light and water medical care clothing
Variables [2] [3] [4]
male
HH members 15-24,
male 0.0106 ** 0.0049 -0.0050 *** 0.0011 0.0089 * 0.0038 0.0007 0.0036
HH members 25-64, -
male 0.0106 ** 0.0047 -0.0023 ** 0.0011 0.0013 0.0036 0.0019 0.0034
HH members 65+, -
male 0.0267 *** 0.0089 -0.0012 0.0021 -0.0135 * 0.0069 0.0035 0.0066
Rural -0.0221 * 0.0132 -0.0012 0.0031 0.0124 0.0103 0.0296 *** 0.0097
Inverse of the Mills -
ratio 0.0531 ** 0.0214 0.0077 0.0050 0.0008 0.0166 0.0098 0.0157
†
Parameters are estimated by iterated SUR which converges to ML estimates; N=1,665; [1] R2=.933 [2] R2=.868; [3] R2=.495; [4] R2=.420;
* Significant at .10 level; ** Significant at .05 level; *** Significant at .01 level;
78
Appendix 4: Selectivity-adjusted IV 3SLS estimates on expenditure shares, by category,
Migrant households, Philippines (cont’n), 2003†
Home operations
and improvements,
Transportation and and non-durable Other consumption
[5] [6] [7] [8]
Coeff se Coeff se Coeff se Coeff se
Constant -0.2044 0.2213 -0.3275 0.3160 0.1388 0.1812 0.2898 0.2699
Age -0.0002 0.0009 0.0001 0.0013 -0.0006 0.0008 0.0011 0.0011
Age squared 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Male -0.0280 * 0.0170 0.0409 * 0.0242 -0.0102 0.0139 0.0402 * 0.0207
Elementary graduate -
0.0089 0.0061 -0.0113 0.0088 0.0006 0.0050 0.0126 * 0.0075
Some High School -
0.0159 ** 0.0070 -0.0312 *** 0.0100 0.0022 0.0058 0.0232 *** 0.0086
High school graduate -
0.0175 ** 0.0080 -0.0269 ** 0.0114 0.0061 0.0066 0.0110 0.0098
Some college -
0.0214 ** 0.0101 -0.0412 *** 0.0144 0.0063 0.0083 0.0010 0.0123
College graduate 0.0238 * 0.0141 -0.0338 * 0.0201 0.0171 0.0115 0.0113 0.0171
ln (disposable income) -
0.0229 0.0195 0.0475 * 0.0278 -0.0072 0.0159 0.0154 0.0237
HH members 0-6,
female -0.0068 *** 0.0024 -0.0049 0.0035 0.0012 0.0020 0.0005 0.0030
HH members 7-14, -
female -0.0015 0.0022 -0.0125 *** 0.0031 -0.0014 0.0018 0.0013 0.0027
HH members 15-24,
female 0.0003 0.0023 -0.0156 *** 0.0033 -0.0013 0.0019 0.0018 0.0028
HH members 25-64,
female 0.0035 0.0035 -0.0200 *** 0.0049 -0.0004 0.0028 0.0026 0.0042
HH members 65+, -
female 0.0052 0.0056 -0.0164 ** 0.0080 -0.0004 0.0046 0.0183 *** 0.0068
79
Home operations
and improvements,
Transportation and and non-durable Other consumption
[5] [6] [7] [8]
Coeff se Coeff se Coeff se Coeff se
HH members 0-6,
male -0.0067 *** 0.0026 -0.0082 ** 0.0037 0.0013 0.0021 0.0067 ** 0.0031
HH members 7-14, -
male -0.0059 *** 0.0021 -0.0125 *** 0.0030 -0.0003 0.0017 0.0068 *** 0.0025
HH members 15-24, -
male -0.0010 0.0023 -0.0088 *** 0.0033 -0.0014 0.0019 0.0041 0.0028
HH members 25-64, -
male 0.0024 0.0022 -0.0064 ** 0.0032 -0.0021 0.0018 0.0017 0.0027
HH members 65+,
male -0.0058 0.0042 -0.0072 0.0061 -0.0012 0.0035 0.0057 0.0052
Rural -0.0058 0.0063 -0.0233 *** 0.0090 0.0023 0.0052 0.0081 0.0077
Inverse of the Mills -
ratio 0.0107 0.0102 -0.0417 *** 0.0145 0.0048 0.0083 0.0257 ** 0.0124
† 2 2 2
Parameters are estimated by iterated SUR which converges to ML estimates; N=1,665; [5] R =.718; [6] R =.732; [7] R =.421;
* Significant at .10 level; ** Significant at .05 level; *** Significant at .01 level;
80
REFERENCES
Acosta, P., P. Fajnzylber and J. Lopez. 2007. “The Impact of Remittances on Poverty and
Human Capital: Evidence from Latin American Household Surveys.” In International Migration,
Economic Development and Policy. C. Ozden and M. Schiff (eds). Washington D.C.: World
Bank and Palgrave Macmillan. pp. 59-98.
Acosta, P. 2006. “Labor Supply, School Attendance, and remittances from International
Migration: The Case of El Salvador.” World Bank Policy Research Paper 3903.
Adams, R., A. Cuecuecha and J. Page. 2008. “Remittances, Consumption and Investment in
Ghana.” World Bank Policy Research Working Paper 5433.
Adams, R. 1998. “Remittances, Investment and Rural Accumulaiton in Pakistan.” Economic

Development and Cultural Change 47(1):155-173.
Adams, R. 2006. “Remittances, Poverty and Investment in Guatemala.” International Migration,

Remittances, and the Brain Drain. in C. Ozden and M. Schiff (edO. Washington DC: World
Bank.
Adams, R. and A. Cuecuecha. 2010. “Remittances, Household Expenditure and Investment in

Guatemala.” World Development (In press).
Barham, B. and S. Boucher. 1998. “Migration, Remittances and Inequality: Estimating the Net
Effects of Migration and Income Distribution.” Journal of Development Economics 55(2): 307-
31.
Blinder, A. 1973. “Wage Discrimination: Reduced Form and Structural Estimates.” Journal of
Human Resources 8:436-455.
Brown, R. 1994. “Migrants’ Remittances, Savings, and Investment in South Pacific.”

International Labor Review 133(3): 347-367.
Brown, R. and D. Ahlburg. 1994. “\Remittances in the South Pacific.” International Journal of
Social Economics 26(1/2/3): 325-344.
Castaldo, A. and B. Reilly. 2007. “Do Migrant Remittances Affect the Consumption Patterns of
Albanian Households?” South-Eastern Europe Journal of Economics 1:25-45.
Chambers, M. and K. Nowman. 1997. Forecasting with the Almost Ideal Demand System:
Evidence from Some Alternative Dynamic Specifications. Applied Economics. 29 (7).
Deaton, A. 1992. Understanding Consumption. New York: Oxford University Press.

Deaton, A. 1997. The Analysis of Household Surveys: A Microeconometric Approach to
Development Policy. Baltimore: Johns Hopkins University Press for the World Bank.
Deaton, A. and J. Muellbauer. 1980. Economics and Consumer Behavior. Cambridge:
Durand, J. and D. Massey. 1992. “Mexican Migration to the United States: A Critical Review.”
Latin American Research Review 27:3-43.
81
Edwards, A.C. and M. Ureta. 2003. “International Migration, Remittances and Schooling:
Evidence from El Salvador.” Journal of Development Economics 72(2): 429-461.
Friedman, M. 1957. A Theory of the Consumption Function. Princeton: Princeton University

Press.
Funkhouser, E. 1992. “Migration from Nicaragua: Some Recent Evidence.” World Development
20(8): 1209-1218.
Glystos, N. 1993. “Measuring the Income Effects of Migrant Remittances: A Methodological

Approach Applied to Greece.” Economic Development and Cultural Change 42(1): 131-168.
Gould, W. 1997. “Interquantile and Simultaneous-Quantile Regression.” Stata Technical Bulletin

38: 14-22.
Hall, R. and F. Mishkin. 1982. “The Sensitivity of Consumption to Transitory Income:

Estimates from Panel Data on Households”. Econometrica 50(2): 462-481.
Heckman, J. 1979. “Sample Selection Bias and Specification Error.” Econometrica 47: 153-161.
Heckman, J., J. Tobias and E. Vytlacil. 2001. “Four Parameters of Interest in the Evaluation of
Social Programs”. Southern Economic Journal 68(2): 210-233.
Jimenez, E. and R. Brown. 2008. “ Assessing Poverty Impact of Remittances with Alternative
Counterfactual Income Estimates.” Discussion Paper Series 375. School of Economics,
University of Queensland, Australia. Available at: http://ideas.repec.org/p/qld/uq2004/375.html.
Keeley, C. and B. Tran. “Remittances from Labor Migration: Evaluations, Performance, and
Implications.” International Migration Review 23(3):500-525.
Koenker, R. and G. Basett. 1978. “Regression Quantiles.” Econometrica 43: 33-50.

Koenker, R. and K. Hallock. 2001. “Quantile Regression.” Journal of Economic Perspectives 15:
143-156.
Koenker, R. and Z. Xiao. 2002. “Inference on the Quantile Regression Process.” Econometrica
70: 1583-1612.
Lauby, J. and O. Stark. 1988. “Individual migration as a family strategy: Young women in the
Philippines.” Population Studies 42:473-486.
Leser, C. 1963. “Forms of Engel Functions”. Econometrica 31(4): 694-703.
Lopez-Cordova, E. 2005. “Globalization, Migration, and Development: The Role of Mexican

Migrant Remittances”. ECONOMIA Fall 2005. pp 217-256.
Mansuri, G. 2007. “Does Work Migration Spur Investment In Origin Communities?

Entrrepreneurship, Schooling and Child Health in Rural Pakistan.” International
Migration,Development and Policy. C. Ozden and M. Schiff (eds). Washington DC: World
Bank.
Massey, D. and E. Parrado. 1998. “International Migration and Business Formation in Mexico.”
Social Science Quarterly 79(1): 1-19.
82
McKenzie, D. 2005. “Beyond Remittances: The Effect of Migration on Mexican Households.”
In International Migration, Economic Development and Policy. C. Ozden and M. Schiff (eds).
Washington D.C.: World Bank. pp. 123-148.
Mroz. T. 1997. “The Sensitivity of an Empirical Model of Married Women’s Hours of Work to
Economic and Statistical Assumptions.” Econometrica 55(4):765-799.
Munshi, K. 2003. “Networks in the Modern Economy: Mexican Migrants in the United States
Labor Market.” Quarterly Journal of Economics 118(2): 549-597.
Oaxaca, R. 1973. “Male-Female Differentials in Urban labor Markets.” International Economic

Review. 14(3): 693-709.
Orbeta, A. 2008. “Economic Impact of International Migration and Remittances on Philippine

Households: What We Thought We Knew, What We Need To Know”. PIDS Discussion Paper
Series No. 2008-32. Manila: Philippine Institute for Development Studies.
Osili, U. 2005. “Migrants and Housing Investments: Theory and Evidence from Nigeria. “
Economic and Cultural Change 52(4): 821-839.
Rapoport, H. and F. Docquier. 2005. “The Economics of Migrants’ Remittances.” IZA

Discussion Paper 1531.
Stark, O. 1991. “Migration in LDCs: Risk, Remittances, and the Family.” Finance and
Development 28(4): 39–41.
Tabuga, A. 2007. “International Remittances and Household Expenditures”. PIDS Discussion
Paper Series 2007-18. Manila: Philippine Institute for Development Studies.
Taylor, J.E,,et al. 1996. “International Migration and Community Development”. Population
Index 62(3): 397-418.
Taylor, J.E, and J. Mora. 2006. “Does Migration Reshape Expenditures in Rural Households?
Evidence from Mexico”, World Bank Policy Research Working Paper 3842. Washington, DC:
World Bank.
Woodruff, C. and R. Zenteno. 2007. “Migration Networks and Microenterprises in Mexico.”

Journal of Development Economics 82: 509-528.
Wooldridge, J. 2002. Econometric analysis of cross section and panel data. Cambridge, Mass.:
The MIT Press.
Working, H. 1943. “Statistical Laws of Family Expenditure”. Journal of the American

Statistical Association. 38(221): 43-56.
Yang, D. and H. Choi 2007. “Are Remittances Insurance? Evidence from rainfall shocks in the
Philippines”. World Bank Economic Review 21(2): 219-248.
Yang, D. 2008. “International Migration, Remittances, and Household Investment: Evidence

from Philippine Migrants’ Exchange Rate Shocks.” Economic Journal 118(528).
83
Chapter 4: Impacts of International Migration and Remittances on
Child Outcomes and Labor Supply in Indonesia: How Does Gender
1
Matter?
TRANG NGUYEN
The World Bank
TNguyen16@worldbank.org
and
RIRIN PURNAMASARI
The World Bank
*
rpurnamasari@worldbank.org
ABSTRACT: This paper aims to investigate empirically how international migration and
remittances in Indonesia, particularly female migration, affect child outcomes and labor supply
behavior in sending households. We analyze the Indonesia Family Life Survey (IFLS) data set
and apply an instrumental variable estimation method, using historical migration networks as
instruments for migration and remittance receipts. This study finds that, in Indonesia, the
impacts of international migration on sending household are likely to vary depending on the
gender of the migrants. On average, migration reduces the working hours of remaining
household members, but this effect is not observed in households with female migrants. At the
same time, female migration and their remittances tend to reduce child labor. The estimated
impacts of migration and remittances on school enrollment are not statistically significant, but
this result is interesting in that the directions of the effects can be the opposite when the migrant
is male or female.
1. INTRODUCTION
The number of international migrants has been increasing over time. In parallel with that,
participation of women in international labor migration also rose. In 2005, the share of female
migrants in the world’s migrant stock was getting closer to half (World Bank 2008a). Following
an upward, global trend in international migration, Indonesia is now one of the largest migrant
labor exporting countries in the world. International migration from Indonesia has been strongly
dominated by women. According to official records, the number of documented female migrant
workers make up to approximately 80 percent of the Indonesia’s migrant workers in 2007 (Figure
1).
1
We thank Ahmad Ahsan, Richard Adams, Jr.,Daniel Mont, and participants at the June 2010 conference on “Cross-
Border Labor Mobility” in Singapore for their very helpful comments and feedback. We are grateful to Survey Meter in
Indonesia for constructing the IFLS panel data and for providing data support. Kalpana Mehra and Aaron Szott
provided great research assistance. The opinions expressed here are those of the authors and do not necessarily
represent the view of the World Bank, its Executive Directors, or the countries they represent.
84
Figure 1
Flow of num ber of m igrant w orkers by sex
600000
500000
400000
Male
300000
Female
200000
100000
0
94
96
98
00
02
04
06
19
19
19
20
20
20
20
Source: BNP2TKI
The rise in migration has also contributed to the increase in global remittance flows.
Recorded remittances sent home to developing countries were nearly US$ 222 billion in 2006.
About 17 percent of global remittances flow back to the East Asia and Pacific region, including
Indonesia. According to the IMF Balance of Payments data, remittances to Indonesia totaled
approximately US$5.7 billion in 2006 (World Bank, 2008b). Although the amount of remittances
appears small relative to total GDP (US$364.5 billion), these inflows may be quite significant in
the specific regions of the migrants’ origin.
While international migrant work has become an increasingly important part of

Indonesian economy, not much research has been done to understand to what extent migration
and remittances may affect the livelihoods of the migrant’s origin household. The impacts on
adult and child labor supply as well as human capital accumulation can manifest through the
income effects of remittances, the impacts on work incentives, exposure to new information, and
consequences of family disconnects. Previous literature has shown evidence of impacts on the
above outcomes in various developing countries (Adams 2010), though the gender dimension of
the impacts has been less explored. The existing research on international migration from
Indonesia mostly focuses on issues related to the financial literacy and vulnerability of migrant
workers. Such research to date is largely based on qualitative assessments and anecdotal
evidence rather than rigorous quantitative analysis.
This paper fills the gap by quantitatively investigating the development impact of
international migration and remittances in Indonesia, particularly how the migrant’s gender
matters. Female migration is of interest since its influence on labor supply and child outcomes at
home might be very different from that of male migration. World Bank (2008a) argues that men
and women show important differences in the determinants of their decision to migrate as well as
their opportunity cost of migration. Furthermore, the report finds gender differences in the
85
patterns of remittances, budget allocation of remittance income, and hence gender differences in
the impact of migration or remittances on household decisions and welfare.
Predictions from economic theory regarding the differential impact by the migrant’s
gender on household decisions and outcomes, such as labor supply and education investment, are
ambiguous. With migration and remittances, shifts in income sources may affect intra-household
decision making, which suggests potential gender dimensions in the impacts on human capital
decisions. When a woman moves abroad to work, increased income from remittances of a female
migrant may increase her bargaining power and her influence over investment choices in the
household. However, the physical absence from the household is expected to create disconnects
and loss of control over the decisions and activities at home. Family disruption can have negative
consequences for children’s welfare, and whether the mother or the father goes away may matter.
Since the net impact is a priori ambiguous, it becomes an important empirical question.
This study empirically estimates how female and male migration and remittances in
Indonesia affect sending households’ child outcomes and labor supply behavior.2 The analysis
uses large-scale household survey data from the Indonesia Family Life Survey (IFLS) 2000 and
2007. Estimating the causal impact of migration and remittances is usually challenging since the
decision to migrate and to remit is likely to be endogenous. To account for endogeneity, this
study applies an instrumental variable method using historical migration networks as instruments
for migration and remittance receipts.3 Being the first quantitative research, based on a large
survey data, in this topic in Indonesia, this study contributes to a broader view and understanding
of the potential gains and losses from migration for sending households. It also contributes to the
limited existing research on the gender dimensions associated with these gains and losses.
Our results suggest important gender differences in the impacts of international migration
on sending households. In Indonesia, migration reduces the working hours of remaining
household members, but this effect is mainly driven by what happens in households with male
migrants. This negative relationship was not observed for households with female migrants. The
results about child outcomes are also divided along the gender line. Female migration and their
remittances tend to reduce child labor outside the home but not necessarily boost schooling
activities. Though not born out statistically significant, the direction of the estimates suggest that
migration may have a slightly positive impact on school enrollment among households with male
migrants, but this impact disappears among those with female migrants. The lack of oversight
associated with the mother’s absence is likely to make it difficult to ensure sufficient schooling
activities for children at home.
The following section reviews the relevant literature. Section 3 describes the IFLS data, and
Section 4 provides a descriptive examination of migrants and migrant-sending households in
Indonesia. The empirical strategy and results are presented in the subsequent two sections. The
last section concludes.
2
The impact on poverty is explored in a parallel paper. Adams and Cuecuecha (2010) find that receiving remittances
reduces the probability of an Indonesian household being poor by 27.8 percent. Other possible socio-economic effects
of migration, such as social impacts, macroeconomic impacts and transfer of knowledge and skills, are beyond the
scope of this paper.
3
McKenzie and Sasin (2007) provide a detailed discussion on the empirical challenges of estimating the causal impact
of migration and possible instruments that have been used in various research questions. Other papers that have used
migration networks as instruments to estimate the impact on outcomes in the home country include Hildebrandt and
McKenzie (2005), Mansuri (2006b,c), Acosta (2006) and Beaudouin (2005).
86
Literature Review
While there has not been any quantitative analysis, before this paper, about the
development impacts of migration with a gender focus in the Indonesia context, there is a general
literature on different country experiences of impacts of migration4 and some indicative evidence
on the differential impact by gender. Most studies find that migration and remittances tend to
reduce the labor supply and participation of non-migrating family members (Adams 2010). With
regard to impacts of migration and remittances on education investment and outcomes, findings
in the literature are mixed.5
On differential impacts by gender of the migrant, Pfeiffer and Taylor (2008) find that, in
Mexico, households with female migrants are associated with less spending on education than
those without female migrants, while it is not the case for households with or without male
migrants. The authors interpret this result as possibly be due to the migrant women’s limited
monitoring over household budget allocations, or also low-skilled jobs abroad send a signal of
low returns to migration work. In Ghana, Guzman et al. (2008) find that households with female
remitters have a higher expenditure share on health but a lower share on education and on food.
The authors give two possible reasons. First, the husband, in the wife’s absence and lack of
monitoring, is likely to spend less on education. Second, some children might leave with the
migrant wife, resulting in less demand on education expenditure in the origin household. The
above two papers, however, do not account for endogeneity in estimating the impacts. Using
panel data and controlling for household fixed effects in rural El Salvador, Acosta (2011) finds
that male migration has null to slightly positive effect on children’s school enrollment while
female migration appears to have the opposite effect. At the same time, female migration tends to
reduce child labor, the opposite to the effect of male migration.
On differential impacts by gender of the remittance recipient, Acosta (2006) finds that in
El Salvador, labor force participation in households with remittance income decreases for women
but not for men. The hours worked, however, reduced for both genders. These links are
interpreted in the paper as causal impacts since the author attempts to control for selection into
migration. Cabegin (2006), employing a two-stage probit OLS regression, finds that for the
Philippines, on average, higher remittance income reduces the probability to work full time for
both married men and women. However, for women, the effect operates mostly through time
spend at home while for men, the main mechanism is the income effect of remittances, i.e. more
income leading to consuming more leisure. Women in migrant households with school-age
children are less likely to have a full time job than those in non-migrant households.
This paper builds on the existing literature in two important aspects. First, this study is the first to
identify causal impacts of migration and remittances on child outcomes and labor supply in
sending households in Indonesia. In doing so, it explicitly deals with the endogeneity of
migration. Second, it identifies the gender dimensions of these impacts.
Data
The data used in this paper comes from the Indonesia Family Life Survey (IFLS). This
survey started in 1993 and represents a large part of the population. It covers 13 out of 27 (now
4
See Adams (2010) and Hanson (2008) for an extensive review
5
For example, positive impacts have been documented by Yang (2008) for the Philippines, Acosta (2006)
for El Salvador, Mansuri (2006a) for Pakistan. Null or negative impacts have been documented by
McKensie and Rapoport (2006) for Mexico and Acosta et al. (2006) for some countries in Latin America.
87
33) provinces in Indonesia, and these provinces were chosen to maximize representation of the
population. These provinces contained 83% of the national population in 1993, and the survey
used a nationally-representative sample frame within these 13 provinces. The first wave of IFLS
covered 7216 households, and later waves of the survey attempted to capture as many of these
original 7216 households as possible. Unique household-level and individual-level panel data
sets were constructed. The final data set used for analysis in this paper covers 6128 households
tracked in all four waves, excluding those few households with zero expenditure or zero
household size reported. Data was also collected about the communities in which these
households lived.6
In addition to basic household characteristics, the IFLS includes a consumption module

and also collects a few questions related to international migration. Some information on human
development outcomes (education, health), labor supply (hours worked), and household assets is
available. While information collected on income is weak, there is detailed expenditure data on
food and non-food items, such as health, education and durables. We will use per capita
expenditure, constructed in the IFLS, as a proxy measure of welfare, and housing and land
ownership as measures of asset ownership.
The regression analysis in this paper focuses on the 2007 and 2000 IFLS rounds for the
following reasons. The survey was not designed to focus on international migrants and
remittances, and only limited information were collected about migrant characteristics and
remittances sent by these migrants. The earlier years of the IFLS have a very small sample size
of migrants, unlikely to provide sufficient statistical power to answer the research questions of
interest. Moreover, the type of migrants captured in the data can vary from round to round.
Therefore, empirical analysis is restricted to the 2007 and 2000 IFLS rounds for a consistent
definition of migration. In the household-level data, the definition of households with
international migrants refers to those families with parents, children, or spouses who are not co-
residents and are abroad.7 Given the way the questions about remittances are asked, in the data,
households receiving international remittances are a strict subset of households with international
migrants.
The individual-level panel, however, is less useful for this paper because not all
individuals abroad are surveyed. An international migrant can only be defined as an individual
who was present in the household in the previous IFLS round, but is abroad in the following
round. The rate of migrants captured in this data is very small. Only descriptive statistics,
without regression analysis, about individual migrants are reported.
Among households with migrants, the data allows further disaggregation by gender of the
migrants. We can classify those families into (i) those with only female migrants, (ii) those with
only male migrants, and (iii) those with both male and female migrants. The third category has
very few households. In the regression analysis, we will combine categories (ii) and (iii) as
households with mainly male migrants, for easy interpretation. Alternatively, the data also allows
calculation of the female share among migrants, that is, for each household with migrants, what is
the fraction of migrants that are female (mother, daughter, and wife).
6
Community data is available for slightly fewer households than the full sample since only original IFLS
communities were surveyed. As households moved to new communities, data is not available for the new
ones.
7
The 2000 data also identifies a fourth source—siblings—but for consistency purposes, we keep the same
definition between the 2000 and 2007 data.
88
Descriptive Statistics
Analyzing the differential impacts of female migrants requires a good understanding of

the socio-economic factors of the migrants and migrants’ origin households in determining their
international labor migration. This section presents descriptive examinations of the profiles of
migrants and migrants’ origin households, including their trend over times. Some key
characteristics of the migrants and migrants’ households are analyzed to identify who among
migrants are likely to search for jobs overseas, and from which households they are,
differentiating between male and female migrants. This section also presents a descriptive
examination of which type of household is likely to receive international remittances. Various
characteristics among individual migrants as well as migrant and non-migrant households are
presented in Table 1 and 2 (following pages). A simple regression to identify correlates of
migration is presented in Table 3.
The majority of Indonesian migrant workers are married and from the age group of 21 to
30 years old. On average, male migrants are older than female migrants. Interestingly, more
recent female and male migrants are both older than in previous years. The average age of female
(male) migrants in 2000-2007 was 28.7 (31.2) years old, while in 1997-2000 was 26.8 (30.5)
years old.
Migrant workers tend to be more educated than before. During the period of 1993-2000,
about half of migrant workers had primary education. However, during the subsequent period of
2000-2007, more workers with higher levels of education found work abroad. The proportion of
migrants with junior and high school education during 2000-2007 was 9 and 6 percentage points
higher than that proportion during 1993-1997 and 1997-2000, respectively. For female migrants,
the proportion of senior high education workers increased more, particularly for the last period,
although the proportion of migrants who completed senior high school was still lower than those
completed junior high school. The trend to send more educated female workers is possibly related
to demand from new destination countries for slightly more skilled work, such as baby sitter and
care taker for the elderly rather than the demand for domestic worker. In contrast, male migrant
workers experienced higher increase in the level of junior high education over time; however, the
proportion of workers having senior high education is higher in a given year.
Indonesia migrants generally find employment in neighboring Asian countries and in the
Middle East, of which Malaysia and Saudi Arabia are the two main destinations. While demand
from these two countries continues to stay strong, the number of destination countries in Asia is
expanding. Almost 90 percent of all migrant workers moved to Malaysia and Saudi Arabia for
work during 1997-2000. This fraction decreased to about 80 percent of all Indonesian migrant
flows by 2000-2007. Female and male migrants, however, are choosing different destinations.
Although most female migrant workers work in Saudi Arabia, they are increasingly finding
employment in other Asian countries, such as Malaysia, Singapore, Taiwan and Hong Kong. The
trend is the opposite for men. Although the majority of male migrants find work in Malaysia, they
are now increasingly shifting to Saudi Arabia.
About 4 to 5 percent of all sample households in the IFLS survey reported having
international migrants. The migrant sending households, however, are concentrated in some
provinces such as West Nusa Tenggara, West Java and East Java. In these provinces, the share of
sample households with migrants in 2007 was as high as 17 percent (Nusa Tenggara) and 6
percent (West and East Java). While female migrants are predominantly from West Java, male
migrants mostly come from West Nusa Tenggara. The shares of migrants from these two main
regions are increasing over time.
89
Table 1: Migrants' Profiles
1993-1997 1997-2000 2000-2007
Male Female Total Male Female Total Male Female Total
Origin region
West Java 0.02 0.08 0.10 0.03 0.09 0.12 0.04 0.13 0.16
East Java 0.22 0.06 0.28 0.15 0.08 0.22 0.08 0.10 0.17
West Nusa Tenggara 0.25 0.04 0.29 0.22 0.06 0.28 0.18 0.09 0.27
Central Java 0.07 0.04 0.10 0.06 0.06 0.13 0.06 0.05 0.11
Other provinces 0.15 0.09 0.23 0.13 0.12 0.25 0.10 0.19 0.29
Education attainment
Elementary and lower 0.39 0.17 0.56 0.29 0.20 0.49 0.18 0.25 0.43
Junior High School 0.17 0.07 0.24 0.13 0.10 0.22 0.12 0.13 0.25
Senior High School 0.13 0.05 0.18 0.16 0.07 0.23 0.13 0.13 0.26
University 0.01 0.01 0.02 0.02 0.02 0.04 0.02 0.02 0.04
Don't know 0.01 0.00 0.01 0.00 0.01 0.01 0.00 0.01 0.02
Age
< 20 years 0.08 0.07 0.15 0.05 0.04 0.09 0.03 0.06 0.08
20-25 years 0.21 0.09 0.30 0.20 0.17 0.37 0.11 0.18 0.29
26-30 years 0.07 0.05 0.12 0.09 0.08 0.17 0.11 0.11 0.22
31-35 years 0.10 0.04 0.14 0.10 0.04 0.15 0.06 0.09 0.15
35-40 years 0.12 0.04 0.16 0.07 0.05 0.12 0.07 0.07 0.13
Above 40 years 0.12 0.01 0.13 0.10 0.01 0.11 0.07 0.05 0.12
Marital status
Single 0.29 0.14 0.42 0.26 0.14 0.39 0.19 0.19 0.38
Married 0.30 0.10 0.40 0.32 0.21 0.53 0.23 0.29 0.52
Widow/Divorcee 0.12 0.06 0.18 0.02 0.05 0.07 0.02 0.07 0.09
Origin HH Percapita expenditure
Quintile 1 0.18 0.07 0.25 0.13 0.08 0.22 0.14 0.12 0.26
Quintile 2 0.16 0.05 0.21 0.14 0.08 0.22 0.10 0.14 0.24
Quintile 3 0.18 0.08 0.26 0.14 0.07 0.22 0.08 0.09 0.17
Quintile 4 0.14 0.06 0.20 0.11 0.06 0.17 0.05 0.08 0.13
Quintile 5 0.04 0.04 0.08 0.07 0.11 0.18 0.06 0.08 0.14
Destination country
Malaysia N/A N/A N/A 0.52 0.16 0.68 0.34 0.17 0.51
Saudi Arabia N/A N/A N/A 0.03 0.18 0.21 0.05 0.23 0.27
Singapore N/A N/A N/A 0.00 0.02 0.03 0.00 0.04 0.04
Taiwan N/A N/A N/A 0.00 0.03 0.03 0.01 0.03 0.04
Hong Kong N/A N/A N/A 0.00 0.00 0.00 0.00 0.03 0.03
Other countries 0.04 0.01 0.05 0.05 0.05 0.11
N 291 446 826
Note: Table presents proportion of sub-sample from total sample for each year
Migrant 2000-2007 is defined as household member present in 2000 but abroad in 2007
90
Table 2: Basic characteristics of households with and without migrants
Panel A: By migrant status Panel B: By migrant gender*
2007 2000 2007 Migrants 2000 Migrants
Male and Male and
Non-migrant Migrant Non-migrant Migrant Male Female Female Male Female Female
Urban (%) 49.4 37.0 44.3 34.2 40.9 37.9 22.9 31.4 35.4 37.0
Household size 4.1 3.9 4.6 4.5 4.3 3.7 4.0 4.6 4.2 4.9
Age of Head 53.0 57.0 50.1 58.3 56.9 55.7 63.1 64.2 55.6 53.0
Male head (%) 78.6 69.8 83.2 72.0 63.4 76.5 57.1 62.8 76.8 79.6
Religion of head - Islam 88.0 96.4 88.0 95.0 96.8 96.7 94.3 95.3 96.3 92.6
Head with no schooling (%) 12.6 26.3 14.5 27.0 26.9 20.9 48.6 31.4 26.8 20.4
Head with elementary (%) 50.6 51.6 52.7 52.7 45.2 57.5 42.9 41.9 62.2 55.6
Head employed (%) 84.3 74.7 88.4 81.1 73.1 75.8 71.4 79.1 82.9 81.5
Per Capita expenditures (000 Rupiah/year - real) 1,149 878 769 663 747 1,028 568 574 745 682
Food Expenditures (000 Rupiah/year - real) 1,659 1,456 1,608 1,514 1,565 1,451 1,187 1,428 1,421 1,792
Non-food expenditures (000 Rupiah/year - real) 2,270 1,330 1,567 1,134 1,129 1,533 976 874 1,198 1,452
Housing expenditures (000 Rupiah/year - real) 646 410 342 262 298 522 219 196 320 279
Education expenditures (000 Rupiah/year - real) 44 28 35 23 25 31 24 19 28 22
Health expenditures (000 Rupiah/year - real) 25 25 21 14 17 28 37 13 15 13
Males over 15 years 1.5 1.4 1.5 1.4 1.3 1.4 1.3 1.3 1.5 1.5
Females over 15 years 1.6 1.5 1.7 1.7 1.8 1.3 1.5 1.8 1.4 1.9
Children under 5 years 0.3 0.3 0.4 0.4 0.4 0.3 0.4 0.5 0.3 0.5
Members 6-18 years 1.0 1.0 1.3 1.3 1.0 0.9 1.2 1.2 1.3 1.4
Wage earners in HH 1.2 0.9 1.2 1.0 0.9 0.9 0.8 1.0 1.0 1.3
Hours worked last week by all HH members 74.6 62.5 81.6 81.3 60.6 65.4 55.4 73.0 79.5 97.1
Hours worked last week by HH head 30.4 27.2 33.8 36.8 23.6 30.5 22.7 36.8 36.0 37.9
Members over 15 years with elementary 1.2 1.2 1.4 1.3 1.2 1.2 1.1 1.2 1.4 1.4
Members over 15 years with junior high 0.5 0.5 0.5 0.5 0.4 0.5 0.5 0.5 0.3 0.7
Members over 15 years with senior high 0.8 0.5 0.7 0.5 0.6 0.5 0.4 0.4 0.5 0.7
Members over 15 years with university 0.3 0.1 0.2 0.1 0.1 0.1 0.1 0.1 0.0 0.1
Males 6-18 years still in school 0.4 0.4 0.7 0.7 0.5 0.4 0.4 0.7 0.7 0.8
Females 6-18 years still in school 0.4 0.4 0.7 0.5 0.4 0.4 0.5 0.5 0.5 0.6
Males 6-18 years working in last 12 months 0.1 0.1 0.6 0.6 0.2 0.0 0.1 0.7 0.6 0.7
Females 6-18 years working in last 12 months 0.1 0.0 0.5 0.4 0.1 0.0 0.1 0.4 0.4 0.4
Used health facility when sick (%) 47.8 39.9 47.8 41.9 40.9 38.6 42.9 45.3 39.0 40.7
Own house (%) 91.2 95.7 91.6 95.9 98.9 93.5 97.1 100.0 93.9 92.6
Own land (%) 49.2 51.2 54.8 54.1 51.6 45.8 74.3 54.7 53.7 53.7
Borrow money in last 12 months (%) 17.9 8.2 23.9 16.7 8.6 8.5 5.7 17.4 18.3 13
N 5,847 281 5,906 222 93 153 35 86 82 54
Note: The figures are sample mean in each sub-sample, otherwise indicated. ** n < 10 households
* Mutually exclusive classification. Male = households with only male migrants, Female = households with only female migrants,
Male and Female = households with male and female migrants
91
Table 3: Determinants of Households with International Migrants (First Stage)
Dependent variable: Dummy if household
has an international migrant 2007 2000
OLS OLS Probit Probit OLS OLS Probit Probit
(1) (2) (3) (4) (5) (6) (7) (8)
Urban status indicator -0.015** -0.014* -0.012** -0.011* -0.005 -0.009 -0.004 -0.005
(0.006) (0.008) (0.005) (0.006) (0.005) (0.007) (0.004) (0.006)
Household size 0.002 0.002 0 0.001 0.004* 0.004* 0.003 0.003**
(0.002) (0.003) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
Age of household head 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.000***
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Household head is male -0.020** -0.020** -0.019** -0.018** -0.025*** -0.025*** -0.022*** -0.023***
(0.008) (0.009) (0.008) (0.008) (0.009) (0.009) (0.008) (0.008)
Islamic household head indicator 0.020*** 0.020*** 0.023*** 0.024*** 0.013** 0.011** 0.013** 0.014**
(0.005) (0.006) (0.005) (0.006) (0.005) (0.005) (0.006) (0.006)
Household head's highest educational level is primary -0.030*** -0.028** -0.016*** -0.015** -0.015 -0.016* -0.009 -0.01
(0.011) (0.011) (0.006) (0.007) (0.009) (0.009) (0.006) (0.006)
Household head's highest educational level is higher than
primary -0.035*** -0.034*** -0.022*** -0.021*** -0.016 -0.015 -0.014** -0.009
(0.012) (0.012) (0.007) (0.007) (0.010) (0.010) (0.006) (0.007)
Number of household males over age 15 -0.013** -0.014** -0.009** -0.009** -0.009** -0.009** -0.008** -0.006*
(0.005) (0.005) (0.004) (0.004) (0.004) (0.004) (0.003) (0.003)
Number of household children under age 5 0.014** 0.015** 0.014*** 0.014** 0.006 0.006 0.004 0.004
(0.006) (0.007) (0.005) (0.005) (0.005) (0.006) (0.004) (0.004)
Proportion of household members that are female -0.079*** -0.085*** -0.061*** -0.063*** -0.060*** -0.062*** -0.048*** -0.047***
(0.019) (0.021) (0.013) (0.014) (0.017) (0.018) (0.013) (0.013)
Log total household expenditure in lag (2000 or 1997) 0.001 0 0 0 -0.003 -0.003 0 -0.004
(0.003) (0.004) 0.000 (0.003) (0.003) (0.003) 0.000 (0.003)
Fraction of hh in community that has a migrant in 1997 0.200** 0.183* 0.122*** 0.101**
(0.086) (0.095) (0.041) (0.045)
Fraction of hh in community that has a migrant in 1993 0.731*** 0.761*** 0.283*** 0.282*** 0.624*** 0.629*** 0.244*** 0.239***
(0.087) (0.093) (0.029) (0.032) (0.081) (0.085) (0.024) (0.025)
No. Elementary schools per capita in village -0.78 -4.201 -4.713 -5.452
(0.736) (5.813) (3.231) (4.403)
No. Junior high schools per capita in village -2.446** -5.998 0.792 0.134
(1.132) (5.385) (2.372) (3.273)
Farming is the village's major economic enterprise 0 0.004 -0.003 0.001
(0.008) (0.007) (0.007) (0.006)
Constant 0.056 0.061 0.061** 0.070**
(0.035) (0.040) (0.029) (0.032)
Observations 6126 5343 6126 5343 6078 5535 6116 5535
R-squared 0.068 0.072 . . 0.054 0.055 . .
F-statistic for test of joint significance of excluded instruments 41.240 37.990 58.830 55.180
Robust standard errors in parentheses. Probit results are presented as marginal effects.
*** p<0.01, ** p<0.05, * p<0.1
92
Although Indonesia migrant workers are increasingly coming from urban areas, the
majority of them still come from rural areas. More than 60 percent of migrant workers, during the
2000-2007 period, still came from rural households. The new urbanizing trend is continuing
possibly as the Indonesian population is urbanizing, and increasingly more workers from urban
areas are finding work abroad. The share of households having migrant workers from urban areas
increased from 34 percent in 2000 to 37 percent in 2007. The urbanizing trend is observed
stronger for households with only male migrants, with the share from urban area rising from 31
percent in 2000 to 41 percent in 2007.
Households with migrants have, on average, significantly less per capita total expenditure
than those without migrants. From 1993 until 2000, almost equal shares of migrant workers came
from quintiles 1 to 4. Recently, the highest share of migrant workers came from the poorest
quintile of households. Migrant workers from these households look for work outside of
Indonesia in order to better financially support their families back home. Households having only
male migrants are generally worse off but, interestingly, on average, have higher food
expenditure than female migrant households. Meanwhile, households with female migrants tend
to have higher expenditure on non-food items, particularly housing, education and health.
Interestingly, when household welfare is assessed using assets owned by the household,
the proportion of households owning house and land is higher among households sending
migrants than among non-migrant households. Similarly, the proportion of migrant households
that borrow money in last 12 months is less than those of non-migrant households. Such welfare
measures, when not assessed in lags, may indicate the use of remittances sent by household
members currently overseas.
Relative to non-migrant households, households sending migrants are typically headed by

those of older age and having lower levels of education. Female migrants come from households
with younger heads, compared to male migrants. Both female and male migrants, however,
mostly come from households whose head has primary school education. As can be seen in Table
3, older household heads significantly increase the probability of the corresponding household to
have an international migrant, while higher educational attainment of the household head makes it
less likely to send member to work overseas.
Although most household heads are employed, the proportion of household heads with
employment is found less in migrant households. Among migrant households, this proportion is
always higher for female than male migrant households. Looking at the number of wage earners
in each household, non-migrant households have a higher average number of wage earners than
migrant households. As a result, the average number of hours worked (during the last week) by
the head and other members of households with migrants is likely to be lower than that in
households without migrants. Female migrant households have higher average hours worked
than those with male migrants.
Other demographic characteristics of household members, such as household size,

number of household members above 15 years old and their education attainment, tend to be
different between households with and without migrants. On average, household migrants tend to
have fewer household members with education, particularly senior high. Households sending
migrants, both male and female migrants, also tend to have smaller household size and a smaller
number of household members above 15 years old. The regression results in Table 3 show that an
additional male household member above age 15 significantly reduces the probability of the
93
household to have an international migrant by about 0.9 percentage points. The higher the
proportion of female household members, the less likely that the household has an international
migrant.
In general, rural areas are often lack of information, including that related to job
opportunities. Yet, most international migrants are from rural areas. This is because the decision
to take overseas jobs is partly due to networks, i.e. it is driven by success stories of returning
migrant workers who benefit from higher salaries offered in migrant recipient countries. The
regression results shown in Table 3, controlling for multivariate correlations, show that an
increase in the proportion of household with migrants in a village from previous periods (as a
proxy for network) significantly increase the likelihood of household sending migrants in the
current period.
Remittances from migrant workers are an additional source of external financing for
recipient families. The data shows that poorer households are more likely to receive remittances.
Consistent with the profiles of households sending migrants, households receiving remittances
tend to be mostly rural, having more assets, and headed by those of older age, lower education
level and less likely to have jobs.1
Over time, remittances became increasingly important, particularly in households with

female migrants. The fraction of households receiving remittances from female migrants
increased, but the fraction of households receiving remittances from male migrants stayed
roughly the same. Moreover, the average share of remittances to household expenditure from
females has also increased, while the average share from male migrants decreased.
Empirical Strategy
This section discusses the methodological challenges and our approach used to estimate
the extent to which migration and remittances, differentially by gender, affect sending
households’ child outcomes and labor supply behavior.
Endogeneity
A problem commonly faced in estimating the causal impact of migration and remittances
is endogeneity. Running a simple OLS regression of household outcomes with migration status
or remittance receipts as explanatory variables could give a biased estimate of the impact. The
error term and the explanatory variable are likely correlated due to several reasons such as
reversed causality, omitted variables or selection bias. Unobservable characteristics that are
omitted from the analysis such as ability or well-connectedness may be correlated with both the
explanatory variables—migration and remittances—and the outcomes of interest. The direction
of the bias is unclear a priori.
Ideally, an unbiased estimate of the causal impact would be the difference between
outcomes of households with migrants (and/or remittances) and their outcomes in the
counterfactual scenario when these same households do not have migrants. However, households
with migrants tend to be “selected” based on unobservable characteristics. Therefore, households
without migrants will not be a good counterfactual for them.
1
The tables of descriptive statistics and determinants of households receiving remittances are not presented as they are
very similar and consistent with those households having migrants. The tables can be obtained from the authors.
94
One method to potentially account for such possible biases is the use of panel data to
perform fixed-effects or first-difference estimation. The panel data structure allows us to control
for unobserved fixed heterogeneity using household fixed effects. However, the identification
assumption would be that there are no time-varying unobservable determinants of the outcome
variables. We are concerned that there might be unobserved shocks, such as changes in the
structure of the economy or weather shocks, that correlate with the migration decision as well as
child outcomes and labor supply of the origin household.
Another method, the instrumental variable approach, will be used to separate the impact
of migration from selection effects. Historical migration networks, defined as the percentage of
households in the village with migrants in the past, will be used as instruments for migration and
remittances in estimating their impact on the sending households. We expect that larger initial
migration networks would lower the cost of subsequent migration, through information or
through financing, and thus induce more migration. The first-stage regression reported in Table 3
confirms this relationship. The bottom row shows that 1993 and 1997 migration rates are very
strong determinants of whether a household has a migrant in 2007 (as well as whether a
household receives remittances). The F-statistic of joint significance of the instruments is
reported at 38 or higher. By living in a community with high levels of migration in the 1990s, a
household has a higher probability of having an international migrant than similar households in
community with lower initial migration rates. For the 2000 regressions, we use only 1993
network as the instrument since it can be difficult to argue for using the 1997 migration rate as
historical networks in the year 2000.
The identification assumption in this instrumental method is that past migration networks
do not influence household outcomes directly other than through their likelihood of having a
migrant member. While the validity of instruments is usually argued and not easy to verify, we
support the identification strategy in this paper in two ways. First, tests for over-identification in
the 2007 analysis fail to reject that excluded instruments are exogenous. P-values of Sargan’s
over-identification tests are reported. Second, it is important to consider possible threats to the
exclusion restriction. Past migration with remittance inflows might have affected local levels of
human and physical capital intensity, possibly affecting labor demand or infrastructure level and
local economic development in general. These factors in turn might affect schooling and work
decisions now. To gain insights about these possible channels, we investigated the raw and
conditional correlations between various measures of local development in 2007 with 1993 and
1997 migration rates. Since correlations exist for some variables of local development but not
others, and since they do not tell a coherent story, we control for these variables in our
regressions. However, if current migration also results in greater development, such over-
controlling might not capture the full impacts of migration. In the end, our analysis checks for
robustness with and without these various controls.
Estimating Equations
The base estimating equation is, for each household i at time t = 2000 or 2007
(1) Y_it = alpha*X_it + beta*M_it + gamma*female_it*M_it + delta*
female_it + μ_it
The outcomes of interest, Y_it, include hours worked for remaining household members,
household head employment status, children’s school enrollment, and child labor supply. X is a
set of household characteristics, which can include household composition, log of per capita
expenditure (lagged), asset information (land and house ownership), and community
95
characteristics (fraction of population working in agriculture, access to schools and roads, and so
on2). μ is the unobserved component for household i
M is an indicator for having a migrant or receiving remittances. To account for potential
endogeneity, in the two-stage least squares regressions, 1993 and 1997 migration networks,
defined as the percentage of households in the village with migrants, will be used as instruments
for M in 2007. Only 1993 network will be used as instrument for M in the 2000 equation.
As female migration may have a differential impact on the outcomes, we include “female_it”—a
dummy variable which denotes households with only female migrants, the omitted category being
those households with at least some male migrants. Alternatively, this variable can refer to the
female share among migrants, for each household with migrants. The coefficients of interest
related to the impacts of migration are beta and gamma. Gamma, in particular, captures the
differential impact of female migrants.
For robustness check, we will present estimations of probit model in addition to linear probability
for binary variables and also estimate a non-linear relationship in the form of IV probit in
estimating the probability of household head being employed, which is a binary outcome. We
will discuss the robustness of the results with respect to various control variables.
Results
This section first presents the estimated impacts on labor supply and child outcomes, and
how the impacts vary with the gender of the migrant. The results for migration and remittances
are interpreted together throughout this section.3 This is because the results for migration and
remittances are qualitatively very similar. Also, households receiving international remittances
identified in the data are a strict subset of households with international migrants, therefore
constituting a smaller number of observations. Subsequently, a discussion of robustness of the
base results follows. The main results are reported for 2007, the year for which we could use two
instruments, test for over-identification, and interpret the results with more confidence. The
results for 2000, the year for which only 1993 network was used as an instrument for migration,
are discussed in the robustness sub-section. The last sub-section provides a discussion of the
results.
Do Migration and Remittances Affect Labor Supply?
Migration and remittances are likely to have consequences for the welfare of sending
households through their impacts on wages and labor supply decisions. When a large proportion
of the working population migrates, this can exert an upward pressure on wages and create work
incentives for certain sectors of the labor force in sending countries. However, impacts on the
equilibrium wage are unlikely in Indonesia since the rate of migration as a share of the large labor
force is still modest, despite recent increases in migrant outflows. Even without a change in the
equilibrium wage, migration and remittances can still affect the work decision of non-migrating
family members. For example, with the net additional income from remittances, they may opt to
work less and consume more leisure.
2
Other proxies for community characteristics that have been controlled for in checking for robustness in
the estimations are village home ownership rate, access to clean water, existence of a bank in the village,
existence of a slum area in the village, and electricity coverage.
3
The full set of remittance results, though not all presented, is available from the authors upon request.
96
Table 4 presents the results for hours worked last week by all household members in
2007. OLS regression results, with or without controlling for various local development
variables, as shown in the first two columns, suggest that migrant-sending households tend to
work less than non-migrant households (also controlling for household size).
When we use historical migration networks as instruments for having a migrant in 2007,
the above negative effect found in the OLS still hold (Columns 3 and 4), although the size of the
effect detected under the IV is larger. Migrant-sending household members work 26 hours less
per week, compared to the average 75 hours worked among households without migrants.4 The
impact of receiving remittances in 2007 is similar. This estimate of the effect of migration and
remittances on household labor supply has both statistical and economic significance. As also
shown in Table 4, the Sargan’s test for over-identification of the instruments in the 2007 IV
estimation does not reject the null hypothesis that both instruments are exogenous (p-values range
from 0.619 to 0.976).
Considering the different effect by gender, we find that households with female migrants
do not reduce their work efforts while households with male migrants do. Columns 5 to 10 of
Table 4 show the differential impacts when households have more female migrants compared to
male migrants. As mentioned previously, there are two ways we can define “female migrant
households”: (i) dummy variable which denotes households with only female migrants, the
omitted category being those households with at least one male migrant5 (columns 5-7); and (ii)
the female share among migrants for each household with migrants (columns 8-10).
The coefficient of interest is that of the interaction terms, interpreted as the differential
impact of migration between when the household has mostly female migrants versus when it has
mainly male migrants. This number is small and insignificant under OLS regressions but rather
large and significant under IV regressions. In column 7, for example, the estimated coefficient is
-33.402 for the Migrant Households term, and 31.939 for the interaction term “Households has
international migrants, and they are all female.” Both of them are statistically significant and
large. That means, while households with predominantly male migrants work 33 hours less per
week than non-migrant households, this effect is reverted toward zero among households with
female migrants. Column 10 also tells a similar story. Households with all male migrants work
45 hours less than non-migrant households, but this effect is reverted toward zero as the share of
female migrants from the household increases. Again, the influence of receiving remittances is
very similar to the influence of having a migrant.
4
The censored regression (tobit), under the assumption that hours worked are censored below zero, gives similar
results.
5
There are very few households with migrants of both genders. We also have the results for all three categories:
female migrants only, male migrants only, and both genders. The results remain the same: any differential impacts are
between households with only female migrants and those with only male migrants.
97
Table 4: Impact of Migration, by Gender, on Household Labor Supply -- Cross section 2007
Dependent variable: #hours work ed last week by all household female= Hh with ONLY female
members migrants female= share of female migrants
OLS OLS IV IV OLS IV IV OLS IV IV
(Sample mean = 74) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Urban status indicator 3.791*** 2.232 3.442** 2.128 2.256 3.551** 2.262 2.27 3.599*** 2.378
(1.380) (1.842) (1.404) (1.852) (1.842) (1.392) (1.843) (1.842) (1.394) (1.852)
Household size 8.535*** 8.495*** 8.595*** 8.556*** 8.529*** 8.681*** 8.640*** 8.526*** 8.725*** 8.692***
(0.681) (0.701) (0.686) (0.705) (0.701) (0.689) (0.708) (0.701) (0.694) (0.713)
Age of household head -0.402*** -0.376*** -0.382*** -0.355*** -0.375*** -0.384*** -0.360*** -0.375*** -0.384*** -0.360***
(0.052) (0.056) (0.054) (0.058) (0.056) (0.053) (0.057) (0.056) (0.054) (0.057)
Household head is male 8.895*** 9.353*** 8.452*** 8.923*** 9.279*** 8.405*** 8.896*** 9.301*** 8.365*** 8.845***
(1.890) (2.007) (1.900) (2.016) (2.006) (1.902) (2.017) (2.006) (1.911) (2.026)
Islamic household head indicator -1.699 -3.494 -0.932 -2.762 -3.508 -1.319 -3.194 -3.501 -1.298 -3.145
(2.151) (2.316) (2.202) (2.365) (2.315) (2.169) (2.330) (2.315) (2.168) (2.328)
Household head's highest educational level is primary -1.079 -0.863 -1.823 -1.494 -0.948 -1.774 -1.454 -0.905 -1.696 -1.424
(1.923) (2.012) (1.980) (2.058) (2.012) (1.965) (2.044) (2.011) (1.965) (2.048)
Household head's highest educational level is higher than primary -11.577*** -12.213*** -12.475*** -12.968*** -12.248*** -12.178*** -12.694*** -12.203*** -12.017*** -12.563***
(2.365) (2.516) (2.428) (2.568) (2.515) (2.394) (2.537) (2.515) (2.392) (2.535)
Number of household males over age 15 20.441*** 20.548*** 20.139*** 20.253*** 20.524*** 20.189*** 20.326*** 20.521*** 20.125*** 20.247***
(1.375) (1.439) (1.391) (1.455) (1.439) (1.385) (1.447) (1.439) (1.389) (1.451)
Number of household children under age 5 -7.552*** -7.455*** -7.220*** -7.132*** -7.460*** -7.364*** -7.318*** -7.464*** -7.384*** -7.333***
(1.695) (1.798) (1.707) (1.810) (1.797) (1.697) (1.797) (1.797) (1.699) (1.796)
Proportion of household members that are female 29.957*** 30.367*** 28.309*** 28.675*** 30.419*** 29.245*** 29.735*** 30.430*** 29.338*** 29.810***
(3.631) (3.945) (3.759) (4.084) (3.944) (3.667) (3.980) (3.945) (3.671) (3.982)
Log total household expenditure in lag (2000) 5.036*** 5.623*** 5.040*** 5.650*** 5.620*** 4.987*** 5.625*** 5.617*** 4.945*** 5.604***
(1.140) (1.267) (1.143) (1.270) (1.267) (1.140) (1.266) (1.267) (1.140) (1.266)
Migrant Households in 2007 -4.858* -5.993* -26.450** -26.341** -10.527** -33.402** -31.166** -11.250** -45.582** -45.102**
(2.950) (3.093) (12.310) (12.607) (4.476) (15.019) (14.945) (5.113) (21.099) (21.692)
Household has international migrants, and they are all female --
2007 8.407 31.939** 28.631*
(5.939) (15.253) (15.263)
Migrant Households * Share of female migrants, 2007 8.707 48.366** 46.859*
(6.584) (24.376) (25.432)
Constant -24.682*** -25.911** -23.276** -25.156** -26.060** -23.913** -26.056** -26.046** -23.839** -26.236**
(9.513) (10.708) (9.538) (10.727) (10.708) (9.506) (10.714) (10.710) (9.518) (10.746)
Other controls Yes Yes Yes Yes Yes Yes
Observations 6126 5343 6126 5343 5343 6126 5343 5343 6126 5343
R-squared 0.297 0.306 0.292 0.300 0.306 0.294 0.303 0.306 0.291 0.300
Sargan's test for overidentification (p-value) 0.619 0.873 0.779 0.908 0.862 0.976
Robust standard errors in parentheses. Instruments for Migrant Households are Village networks in 1993 and 1997.
*** p<0.01, ** p<0.05, * p<0.1. Other controls: Farming is the village's major economic enterprise, No. Junior high schools and No. Elementary schools per capita in village
98
Does migration also affect the extensive margin of the decision to work, in addition to
how much to work? Table 5 analyzes factors affecting the probability that the household head is
currently working. The head is more likely to be working in rural households (probably self-
employed in agriculture), when the head is male, and when the household is previously poor.
Unconditional averages suggest that household heads with higher education are slightly more
likely to be working. However, this observation does not hold in the regression results.
Regarding the impact of migration, while the OLS and probit regressions for this
outcome variable suggest that household heads in migrant-sending households are significantly
less likely to be employed, the instrumental variable approach shows that this coefficient is
insignificant. Column 4 reports the IVprobit results since using 2SLS for a binary outcome
variable and binary endogenous variable might yield inconsistent estimates. The zero effect on
employment by the household head is observed regardless of the migrant’s gender, as female
interaction terms are small and indistinguishable from zero. Migration does not appear to
influence the household head’s likelihood of working, even though it reduces the household
head’s hours worked (not reported), similar to the result on hours worked for all members.
Do Migration and Remittances Affect Children’s Schooling and Work Behavior?
The next set of outcome variables relates to children’s schooling and work behavior.
These regression analyses refer to children 6 to 18 years old, thus a smaller sample of households
with children.
First, as presented in Table 6, the dependent variable is the fraction of children 6-18 years
old in the household that are in school in 2007. Children in urban and richer households with
more educated household head, unsurprisingly, have higher engagement in education.
While the OLS regression for this outcome variable (column 1) suggests that children in
migrant-sending households are more likely to be in school, the instrumental variable approach
shows that this coefficient is much smaller and insignificant. Migration has no statistically
significant effects on children’s school enrollment in Indonesia.1 The point estimates of the
“migrant households” coefficient in the IV regressions (columns 3 and 4) are closer to zero and
with larger standard errors. This finding holds whether or not we control for log of per capita
expenditure or community-level development (including the village school supply), and whether
or not the household is in urban or rural areas. Thus, only looking at OLS regressions would
give a biased estimate of the effect of migration. Even though the OLS regressions control for
wealth and other pre-determined characteristics, migrant-sending households might correlate with
some unobservables such as connectedness, which also affects investment in education.
1
The results do not depend on the gender of the children. They are similar for boys’ and girls’ enrollment.
99
Table 5: Impact of Migration, by Gender, on Household Head's Employment Status -- Cross section 2007
Dependent variable: Indicator that household head is working female= Hh with ONLY female migrants female= share of female migrants
OLS IV2SLS Probit IVProbit OLS IV2SLS IV2SLS OLS IV2SLS IV2SLS
(Sample mean = 0.84) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Urban status indicator -0.074*** -0.074*** -0.067*** -0.067*** -0.043*** -0.073*** -0.043*** -0.043*** -0.073*** -0.043***
(0.009) (0.009) (0.009) (0.009) (0.012) (0.009) (0.012) (0.012) (0.009) (0.012)
Household size -0.006 -0.006 -0.006 -0.006 -0.010** -0.006 -0.010** -0.010** -0.006 -0.010**
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
Age of household head -0.010*** -0.010*** -0.008*** -0.008*** -0.009*** -0.010*** -0.009*** -0.009*** -0.010*** -0.009***
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Household head is male 0.228*** 0.229*** 0.246*** 0.246*** 0.219*** 0.229*** 0.219*** 0.219*** 0.230*** 0.219***
(0.016) (0.016) (0.018) (0.018) (0.017) (0.016) (0.017) (0.017) (0.016) (0.017)
Islamic household head indicator -0.021 -0.023 -0.018 -0.018 -0.013 -0.023* -0.015 -0.014 -0.024* -0.015
(0.013) (0.014) (0.011) (0.011) (0.015) (0.014) (0.015) (0.015) (0.014) (0.015)
Household head's highest educational level is primary -0.009 -0.007 -0.009 -0.009 -0.003 -0.007 -0.003 -0.003 -0.006 -0.002
(0.016) (0.016) (0.012) (0.012) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016)
Household head's highest educational level is higher than primary -0.059*** -0.057*** -0.059*** -0.059*** -0.045** -0.057*** -0.044** -0.045** -0.056*** -0.043**
(0.018) (0.018) (0.016) (0.016) (0.018) (0.018) (0.018) (0.018) (0.018) (0.018)
(0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008)
Number of household children under age 5 -0.022** -0.023** -0.008 -0.008 -0.017* -0.023** -0.017* -0.017* -0.023*** -0.017*
(0.009) (0.009) (0.010) (0.010) (0.010) (0.009) (0.010) (0.010) (0.009) (0.010)
Proportion of household members that are female 0.116*** 0.119*** 0.108*** 0.108*** 0.126*** 0.120*** 0.129*** 0.126*** 0.122*** 0.130***
(0.030) (0.030) (0.026) (0.027) (0.032) (0.030) (0.032) (0.032) (0.030) (0.032)
Log total household expenditure in lag (2000) -0.015** -0.015** -0.020*** -0.020*** -0.007 -0.015** -0.007 -0.007 -0.015** -0.007
(0.006) (0.006) (0.005) (0.005) (0.007) (0.006) (0.007) (0.007) (0.006) (0.007)
Migrant Households in 2007 -0.053** -0.007 -0.049** -0.050 -0.033 -0.011 -0.048 -0.039 0.005 -0.066
(0.024) (0.093) (0.023) (0.111) (0.036) (0.113) (0.111) (0.041) (0.161) (0.162)
Household has international migrants, and they are all female --2007 -0.024 0.026 0.06
(0.048) (0.116) (0.115)
Migrant Households * Share of female migrants, 2007 -0.013 0.027 0.102
(0.053) (0.179) (0.181)
Constant 1.311*** 1.308*** 1.150*** 1.307*** 1.147*** 1.149*** 1.306*** 1.146***
(0.058) (0.058) (0.064) (0.058) (0.064) (0.064) (0.058) (0.064)
Other controls Yes Yes Yes Yes
Observations 6126 6126 6126 6126 5343 6126 5343 5343 6126 5343
R-squared 0.208 0.207 . 0.202 0.207 0.201 0.202 0.206 0.201
Sargan's test for overidentification (p-value) 0.479 0.618 0.323 0.390 0.230
100
Table 6: Impact of Migration, by Gender, on Children Schooling -- Cross section 2007
Dependent variable: % of children 6-18 in the household that are female= Hh with ONLY female
in school migrants female= share of female migrants
(Sample mean = 0.82) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Urban status indicator 0.019 0.044*** 0.018 0.044*** 0.044*** 0.018 0.044*** 0.044*** 0.018 0.044***
(0.012) (0.016) (0.012) (0.016) (0.016) (0.012) (0.016) (0.016) (0.012) (0.016)
Household size 0.027*** 0.026*** 0.027*** 0.026*** 0.026*** 0.026*** 0.026*** 0.026*** 0.026*** 0.026***
(0.004) (0.005) (0.004) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)
Age of household head 0.001 0.001* 0.001 0.001** 0.001* 0.001 0.001** 0.001* 0.001 0.001*
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Household head is male 0.032* 0.023 0.031* 0.021 0.023 0.031* 0.021 0.023 0.032* 0.022
(0.017) (0.018) (0.018) (0.018) (0.018) (0.018) (0.019) (0.018) (0.018) (0.019)
Islamic household head indicator -0.043*** -0.043** -0.042** -0.041** -0.043** -0.041*** -0.041** -0.043** -0.042*** -0.042**
(0.016) (0.017) (0.016) (0.018) (0.017) (0.016) (0.017) (0.017) (0.016) (0.017)
Household head's highest educational level is primary 0.105*** 0.107*** 0.103*** 0.103*** 0.107*** 0.102*** 0.103*** 0.107*** 0.104*** 0.105***
(0.024) (0.025) (0.026) (0.026) (0.025) (0.025) (0.026) (0.025) (0.025) (0.026)
Household head's highest educational level is higher than primary 0.179*** 0.180*** 0.177*** 0.176*** 0.180*** 0.176*** 0.176*** 0.180*** 0.178*** 0.178***
(0.026) (0.027) (0.027) (0.028) (0.027) (0.026) (0.027) (0.027) (0.026) (0.027)
Number of household males over age 15 -0.080*** -0.080*** -0.080*** -0.080*** -0.080*** -0.080*** -0.080*** -0.080*** -0.080*** -0.080***
(0.009) (0.010) (0.009) (0.010) (0.010) (0.009) (0.010) (0.010) (0.009) (0.010)
Number of household children under age 5 -0.050*** -0.047*** -0.050*** -0.045*** -0.047*** -0.049*** -0.045*** -0.047*** -0.050*** -0.046***
(0.012) (0.013) (0.012) (0.013) (0.013) (0.012) (0.013) (0.013) (0.012) (0.013)
Proportion of household members that are female -0.027 -0.043 -0.028 -0.047 -0.043 -0.029 -0.047 -0.043 -0.028 -0.045
(0.035) (0.038) (0.036) (0.039) (0.038) (0.035) (0.039) (0.038) (0.035) (0.039)
Log total household expenditure in lag (2000) 0.035*** 0.036*** 0.034*** 0.036*** 0.036*** 0.034*** 0.035*** 0.036*** 0.035*** 0.036***
(0.008) (0.009) (0.008) (0.009) (0.009) (0.008) (0.009) (0.009) (0.008) (0.009)
Migrant Households in 2007 0.058** 0.048 0.027 -0.014 0.049 0.031 -0.01 0.056 0.069 0.016
(0.028) (0.030) (0.123) (0.130) (0.039) (0.138) (0.143) (0.042) (0.205) (0.217)
Household has international migrants, and they are all female --2007 -0.003 -0.034 -0.011
(0.056) (0.146) (0.152)
Migrant Households * Share of female migrants, 2007 -0.014 -0.042 0.002
(0.059) (0.239) (0.255)
Constant 0.411*** 0.365*** 0.416*** 0.374*** 0.366*** 0.418*** 0.375*** 0.366*** 0.413*** 0.370***
(0.073) (0.082) (0.075) (0.084) (0.082) (0.074) (0.082) (0.082) (0.074) (0.082)
Observations 3521 3036 3521 3036 3036 3521 3036 3036 3521 3036
R-squared 0.074 0.071 0.073 0.069 0.071 0.073 0.069 0.071 0.074 0.070
101
Interaction terms with female migrants indicate that households with more female
migrants show almost no difference in explaining the fraction of all children enrolled in school, as
shown in columns 5-10. This finding holds for the indicator for households with only female
migrants, and the female share among migrants. Even though the sign is oftentimes negative, the
point estimates are small and insignificant.
Second, as presented in Table 7, the dependent variable is the fraction of children 6-18
years old in the household that are working in the last 12 months of the 2007 survey. The sample
average indicates that 17% of children work, probably in family business or informal jobs.
Children in urban households with Islamic and educated head are less likely to work in the labor
market.
The OLS regression for this outcome variable suggests that children in migrant-sending
households are significantly less likely to work, whether or not controlling for community
variables. As shown in column 1, the share of children in migrant households who work is 7
percentage points less than that in non-migrant households. This finding becomes more nuanced
when we use the instrumental variable approach. The coefficients on the “migrant households”
variable, reported under columns 3 and 4, are not distinguishable from zero.
Columns 5-10 of Table 7 show that the coefficient of the gender interaction term is
robustly negative. We find that children work less when it comes to households with more
female migrants. This finding holds for the indicator for households with only female migrants, as
well as the female share among migrants. To interpret column 6, for example, households with
only female migrants reduce the share of children working by 32 percentage points more than the
impact of households with at least some male migrants. The coefficients of “Migrant
Households” variable, reported in the IV regressions (Columns 9 and 10), are positive but
indistinguishable from zero. It is inferred that migration has no statistically significant effect on
child labor supply in families with only male migrants. However, households with only female
migrants reduce the share of children working by 26.8 – 43.8 = 17 percentage points, as indicated
by the coefficients of the interaction terms in column 10. This effect is relatively large and
important given the sample average that 17% of children work on average in the full sample of
households.
Discussion of Robustness
Various exercises for robustness checks generally support the cross-sectional results
presented above. First, to address worries about household and community characteristics omitted
from the base regression, we checked additional specifications controlling for these variables.
The results are robust to controlling for land and house ownership as proxies for asset holdings,
indicator for having a bank in the village, village home ownership rate (except the household in
102
Table 7: Impact of Migration, by Gender, on Child Labor Supply -- Cross section 2007
Dependent variable: % of children 6-18 in the household female= Hh with ONLY female
that work in the last month migrants female= share of female migrants
(Sample mean = 0.17) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Urban status indicator -0.037*** -0.012 -0.033*** -0.012 -0.013 -0.035*** -0.013 -0.013 -0.035*** -0.015
(0.012) (0.016) (0.012) (0.016) (0.016) (0.012) (0.016) (0.016) (0.012) (0.016)
Household size -0.027*** -0.026*** -0.027*** -0.025*** -0.026*** -0.028*** -0.027*** -0.027*** -0.029*** -0.027***
(0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)
Age of household head -0.001* -0.001** -0.001** -0.001** -0.001** -0.001** -0.001** -0.001** -0.001* -0.001**
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Household head is male -0.034* -0.027 -0.026 -0.019 -0.024 -0.025 -0.018 -0.024 -0.024 -0.017
(0.018) (0.019) (0.018) (0.019) (0.019) (0.018) (0.019) (0.019) (0.019) (0.020)
Islamic household head indicator -0.090*** -0.088*** -0.097*** -0.095*** -0.087*** -0.092*** -0.090*** -0.087*** -0.093*** -0.091***
(0.020) (0.022) (0.020) (0.022) (0.022) (0.020) (0.022) (0.022) (0.020) (0.022)
Household head's highest educational level is primary -0.100*** -0.102*** -0.086*** -0.091*** -0.101*** -0.091*** -0.095*** -0.102*** -0.093*** -0.096***
(0.024) (0.025) (0.026) (0.026) (0.025) (0.025) (0.025) (0.025) (0.025) (0.025)
Household head's highest educational level is higher than
primary -0.163*** -0.172*** -0.151*** -0.160*** -0.172*** -0.158*** -0.167*** -0.173*** -0.160*** -0.170***
(0.026) (0.027) (0.027) (0.028) (0.027) (0.026) (0.027) (0.027) (0.026) (0.027)
(0.009) (0.010) (0.009) (0.010) (0.010) (0.009) (0.010) (0.010) (0.009) (0.010)
Number of household children under age 5 0.021* 0.015 0.016 0.011 0.015 0.019* 0.014 0.016 0.020* 0.014
(0.011) (0.012) (0.012) (0.013) (0.012) (0.012) (0.012) (0.012) (0.012) (0.012)
Proportion of household members that are female 0.065* 0.069* 0.076** 0.082** 0.068* 0.068* 0.073* 0.069* 0.070** 0.074*
(0.035) (0.039) (0.036) (0.040) (0.039) (0.035) (0.039) (0.039) (0.035) (0.039)
Log total household expenditure in lag (2000) -0.001 0.007 0.001 0.009 0.007 0 0.008 0.007 0 0.008
(0.008) (0.009) (0.009) (0.009) (0.009) (0.008) (0.009) (0.009) (0.008) (0.009)
Migrant Households in 2007 -0.071*** -0.086*** 0.136 0.117 0.002 0.179 0.146 0.027 0.3 0.268
(0.025) (0.025) (0.133) (0.140) (0.044) (0.152) (0.156) (0.050) (0.215) (0.227)
Household has international migrants, and they are all female --
2007 -0.174*** -0.320** -0.286*
(0.046) (0.150) (0.155)
Migrant Households * Share of female migrants, 2007 -0.192*** -0.477** -0.438*
(0.054) (0.241) (0.256)
Constant 0.447*** 0.348*** 0.413*** 0.319*** 0.353*** 0.432*** 0.344*** 0.353*** 0.430*** 0.346***
(0.074) (0.084) (0.078) (0.086) (0.083) (0.076) (0.084) (0.083) (0.076) (0.084)
Observations 3521 3036 3521 3036 3036 3521 3036 3036 3521 3036
R-squared 0.055 0.056 0.039 0.040 0.060 0.053 0.055 0.060 0.049 0.051
Robust standard errors in parentheses. For households with 6-18 children. Instruments for Migrant Households are Village networks in 1993 and 1997.
103
question), infrastructure such as road, market access, access to clean water, and the presence of
slums in a village. Taking into account all fixed effects at the district level, by including district
dummies, gives qualitative similar results even though the standard errors tend to be larger due to
smaller sources of variation, unsurprisingly.1
Second, in an attempt to improve precision, additional analysis was performed in a

restricted sample of the major migrant-sending provinces. The small capture of migrant
households in the IFLS, 5% or less in the full sample, is a sample size concern and is likely to
lead to imprecise estimates. Restricting the analysis to major migration provinces can bring the
migration rate up to roughly 8% in 2007, but at the cost of a much lower number of observations.
In the end, most analysis in the restricted sample turned out not to improve precision. The results
for the restricted sample are available upon request, but only those for the full sample are
reported. In addition to robust standard errors, bootstrapping standard errors gave similar results.
Standard errors clustered at the village level tend to be slightly larger, but also gave qualitative
similar results.
Third, the same analysis for 2000 suggests broadly similar results, except for those about
child labor supply. For 2000, only 1993 network could be used as an instrument for migration,
and an over-identification test was not possible. Assuming the 2000 IV specification is valid, the
only difference in the 2000 results is that households with female migrants in 2000 are not likely
to lead to a reduction in child labor supply.
Finally, in addition to the cross-sectional analysis, panel analysis using household fixed effects
usually gives very high standard errors and thus imprecise, insignificant point estimates. As one
exception, assuming that there are no unobserved time-varying determinants of the outcome
variables, panel analysis does show that migration reduces child labor supply.
Discussion of the Results
Consistent with income theory and with findings in other countries, we find that in
Indonesia, migration has a negative impact on the average labor supply of remaining household
members. This impact manifests through the intensive margin that remaining members work
fewer hours, rather than through the extensive margin that the household head withdraws from
the labor force. When disaggregated by region, Table 8 reports the results of the IV estimations
separately for urban and rural areas. In urban areas, we find a negative effect of migration on
total labor supply, regardless of the gender of the migrant. However, the main findings about the
gender influence in the whole sample reflect closely what happens in rural areas since more
migrants come from rural areas. As indicated in column 2, male migrants tend to reduce
remaining household members' labor supply in rural areas, but female migrants do not.
Moreover, the negative impact of female migration on children’s work behavior is only observed
in rural areas.
The negative impact of remittances on the labor supply of remaining household members
is not necessarily a concern unless work incentives are also distorted. The part of the fall in labor
supply due to an income effect, via increased leisure, represents a private welfare gain.
Alternatively, household members might substitute wage labor with more time in parenting and
1
The only additional control that seems to affect some of the results is the fraction of households in the
village with electricity. This is a concern only to the extent that one argues electricity coverage affects
household labor supply and children’s school and work behavior, and is correlated with historical
migration. Even so, it is not clear if this is a good control variable since the relationship between access to
electricity and hours worked, for example, can be reversed causality.
104
Table 8: Impact of Migration in Urban and Rural Areas -- Cross section 2007 (IV estimates)
#hours worked last % of children 6-18 in % of children 6-18
Dependent variable: week by all members school work
Urban Rural Urban Rural Urban Rural
(1) (2) (3) (4) (5) (6)
Migrant Households in 2007 -56.127* -35.933 -0.005 -0.102 0.083 0.453
(30.909) (23.645) (0.234) (0.272) (0.240) (0.284)
Migrant Households * Share of female
migrants, 2007 0.000 30.852 0.000 0.155 0.000 -0.611*
(10.456) (29.000) (0.043) (0.318) (0.044) (0.319)
Constant -14.483 -16.841 0.461*** 0.653*** 0.411*** 0.038
(16.199) (37.680) (0.140) (0.113) (0.156) (0.119)
Observations 2990 3136 1672 1849 1672 1849

R-squared 0.272 0.298 0.062 0.074 0.031 0.066
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Other household controls include demographic, log of lagged expenditure
Instruments for Migrant Households are Village networks in 1993 and 1997.
home production, or increased capital and improved labor productivity. On the other hand,
remittances seen as conditional on low household income can discourage work incentives of non-
migrating members. It is, unfortunately, difficult to empirically separate out the distortionary effect
on labor supply. But we know that a long-term impact on welfare may be limited if remittance
recipients continue to depend on external transfers and do not use remittance money for productive
investment that can bring returns in the future.
The difference in impacts on labor supply, as discussed earlier, seem to be driven by the
migrant’s gender, and his or her influence on household decisions, rather than who is left to lead the
household. Considering gender of the household head, we do not find any evidence that the impact
of migration varies significantly whether the head is male or female. Thus, it is not the case that
because women leave, the men manage the household and use remittance money differently. One
could expect that the physical absence of and the remittances sent by migrants play a complex role in
influencing their say in household decision making. For that reason, male and female migrants may
be expected to have differential impacts on the work-leisure decision of remaining household
members.
Concerning children’s schooling, we find that migration has no statistically significant effects on
school enrollment in Indonesia. This finding is not likely explained by supply factors. Inadequate
supply of schools is not a serious concern in Indonesia at the primary level, and to a lesser degree, at
the secondary level. Our analysis shows that even in urban areas, with better school supply, there is
no evidence of impact of migration on school enrollment. Estimated impacts on children’s school
enrollment are reported in columns 3 and 4 of Table 8. Then, is it possible that enrollment is not
responsive to migration and remittances, but school attendance may be? Column 4 of Table 9 shows
no statistically significant impact of migration on children’s hours in school per week. In this
context, the negative effect of an absent parent on childcare at home and the “signaling effect” about
105
Table 9: Impact of Migration on Child Outcomes by Age -- 2007 (IV estimates)
Hours in school last week (if
Dependent variable: Enrolled in school enrolled) Main activity is idleness
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Panel A: Children 6-10

Migrant Households in 2007 0.147 0.152 0.503** -6.443 -7.854 -13.612 -0.087 -0.112 -0.44
(0.123) (0.132) (0.250) (4.929) (5.328) (8.811) (0.125) (0.133) (0.276)
Household has international migrants, and
they are all female --2007 -0.18 4.848 0.331**
(0.146) (5.634) (0.154)
migrants, 2007 -0.531 12.192 0.659**
(0.325) (10.506) (0.332)
Constant -0.064 -0.052 -0.055 4.035 3.808 3.755 0.846*** 0.825*** 0.832***
(0.110) (0.110) (0.112) (4.083) (4.074) (4.111) (0.106) (0.106) (0.107)
Observations 1682 1682 1682 1499 1499 1499 1699 1699 1699
R-squared 0.175 0.181 0.136 0.065 0.06 0.031 0.109 0.115 0.077
Panel B: Children 11-14

Migrant Households in 2007 0.204 0.182 0.289 -1.576 0.148 11.427 -0.09 -0.07 -0.212
(0.160) (0.165) (0.248) (9.368) (9.366) (12.532) (0.145) (0.151) (0.227)
they are all female --2007 -0.231 -11.578 0.218
(0.170) (9.299) (0.163)
migrants, 2007 -0.371 -21.872 0.382
(0.301) (14.502) (0.279)
Constant 1.034*** 1.085*** 1.077*** 18.267*** 20.318*** 18.517*** 0.032 -0.016 -0.001
(0.113) (0.107) (0.108) (6.272) (5.617) (5.677) (0.100) (0.096) (0.096)
Observations 1692 1692 1692 1554 1554 1554 1702 1702 1702
R-squared 0.081 0.091 0.081 0.007 0.003 0.005 0.047 0.048 0.035
Panel C: Children 15-18

Migrant Households in 2007 0.25 0.285 0.465 -7.041 -5.512 -4.864 -0.053 -0.113 -0.189
(0.181) (0.212) (0.339) (6.852) (7.355) (10.938) (0.130) (0.149) (0.242)
they are all female --2007 -0.33 -1.849 0.282
(0.235) (8.064) (0.186)
migrants, 2007 -0.516 -2.314 0.31
(0.390) (13.261) (0.308)
Constant 1.703*** 1.754*** 1.785*** 31.175*** 31.330*** 31.480*** 0.035 -0.003 -0.016
(0.221) (0.219) (0.221) (8.967) (8.982) (9.179) (0.174) (0.172) (0.173)
Observations 1740 1740 1740 1118 1118 1118 1797 1797 1797
R-squared 0.191 0.194 0.193 0.02 0.023 0.024 0.031 0.037 0.035
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Instruments for Migrant Households are Village networks in 1993 and 1997.
Other controls: child and household demographics, log of lagged expenditure, and school availability and village controls
106
the returns to education may play a role in offseting the income effects from remittances. Half of the
migrants from Indonesia are primary school graduates. And to the extent that the process and
prospect of migration leads families to revise their perceived returns to education, migration might
not necessarily increase school enrollment in the end in a country where the primary enrollment rate
is already more than 90 percent.
Does the lack of impact on education outcomes for the average household hide differential
effects among richer and poorer families? Migration and remittances can be expected to influence
poor households more due to the extent of credit constraints or varying preferences for education.
However, our analysis (not shown) does not find heterogeneous effects for different levels of
household expenditure.
Since the factors shaping education decisions for children in primary, secondary, and tertiary
school age can vary widely, additional analysis is conducted separately for individuals in these age
groups. The first three columns of Table 9 present these results of impacts on enrollment. Even
though the direction of the impact of migration on school enrollment appears positive for all age
groups (column 1), such a conclusion is not evident since the empirical estimates cannot be
distinguished from zero. Subject to this caveat about imprecise estimation, the size of the impact
tends to be larger among older children aged 11 to 18 and smaller among children aged 6 to 10,
whose initial enrollment rate is already high. The positive sign of the estimated impacts reflect what
happens among households with male migrants rather than among those with female migrants, as
shown in columns 2 and 3 of Table 9. The type of families sending male migrants and the type
sending female migrants may be different for unobserved
reasons, or alternatively, male and female migrants themselves have different impacts. Since women
tend to be more involved than men in child care and monitoring children’s activities, when a mother
migrates for work overseas, this action can have worse consequences for children’s schooling
behavior than the father’s migration. In fact, columns 7-9 of Table 9 suggest that migration may
increase children’s idle time among families with a migrating female, particularly for young children
aged 6-10.
These results are very similar to findings about female migration and child outcomes in rural
El Salvador. Acosta (2011) finds that male migration has null to slightly positive effect on children’s
school enrollment while female migration appears to have the opposite effect. At the same time,
female migration tends to reduce child labor, in contrast to male migration from El Salvador. The
results are also consistent with the findings of Guzman et al. (2007) and Pfeiffer and Taylor (2007)
that female migration, as opposed to male migration, is associated with lower household expenditure
on education
2. CONCLUSION
Using the large IFLS data, this paper is the first attempt to quantitatively assess the impacts
of migration, differentiated by gender, from Indonesia on labor supply and child outcomes in sending
households. We apply the instrumental variable method using historical migration networks as
instruments for migration and remittance receipts.
Overall, we find that the impacts of migration are likely to vary depending on the gender of
the migrants. Migration and remittances reduce the labor supply of remaining members in sending
households. However, this result reflects what happens in cases of families with male migrants only.
Families with female migrants may be different for unobserved reasons, or female
107
migrants may prefer a different use of their remittances rather than increased leisure for adults. One
possible use is to pull children out of the labor force. Our analysis indeed shows that international
migration reduces child labor supply in households with female migrants.
Migration does not seem to significantly affect on school enrollment or attendance of

children of both genders and across age groups in Indonesia. This result is consistent with the
negative effect of an absent parent on childcare at home and the possible “signaling effect” about the
returns to education that may offset the income effects from remittances. The negative effect of an
absent parent is likely to be more severe in the case of female migrants since women tend to be more
involved than men in child care and monitoring children’s activities. Results show that migration
increases children’s idle time among families with a migrating female, particularly for young children
aged 6-10, and probably reverts any positive impact that migration may have on school enrollment.
More quality data on migration and policy evaluations would be needed to better understand
the exact mechanisms of how migration and remittances affect child outcomes as well as other
development outcomes. Given the different gender roles and preferences in the household, further
quantitative analysis is also required to obtain more empirical evidence to understand and unpack the
differential impacts by gender. Assessing the net impact on welfare and the long-term consequences
of migration and remittances would require looking more comprehensively at other economic and
social outcomes likely affected by this process.
108
REFERENCES
Acosta, P. (2011). “Female Migration and Child Occupation in Rural El Salvador.” Forthcoming in
Population Research and Policy Review.
Acosta, P. (2006). “Labor Supply, School Attendance, and Remittances from International
Migration: The Case of El Salvador.” World Bank Policy Research Working Paper 3903.
Acosta, P., P. Fajnzylber and J. Humberto Lopez (2008). “Remittances and Household Behavior:
Evidence for Latin America,” in P. Fajnzylber and J. Humberto Lopez (eds) Remittances and
Development: Lessons from Latin America. Washington, DC: World Bank.
Adams, R. (2010) “Evaluating the Economic Impact of International Remittances on Developing

Countries Using Household Surveys: A Literature Review.” Forthcoming in Journal of
Development Studies.
Adams, Jr., R. (1998). “Remittances, Investment and Rural Asset Accumulation in Pakistan,”
Economic Development and Cultural Change 47:1 (October): 155-173.
Adams, R. and A. Cuecuecha (2010). “The Economic Impact of International Migration and
Remittances on Poverty and Household Consumption and Investment in Indonesia.” World Bank,
Washington DC.
Beaudouin, P. (2005). “Economic Impact of Migration on a Rural Area in Bangladesh.” Mimeo,

Centre d’Economie de la Sorbonne, Universite Paris 1.
Buchori, C. and M. Amalia (2006). “Fact Sheet: Migration, Remittance and Female Migrant
Workers.” World Bank, Washington DC.
Cabegin, E. (2006). “The Effect of Filipino Overseas Migration on the Non-Migrant Spouse’s
Market Participation and Labor Supply Behavior.” Institute for Study of Labor (IZA) Discussion
Paper 2240. Bonn, Germany.
Guzman, J., A. Morrison and M. Sjoblom (2008). “The Impact of Remittances and Gender on
Household Expenditure Patterns: Evidence from Ghana” in A. Morrison, M. Schiff, and M. Sjoblom
(eds) The International Migration of Women. The World Bank, Washington, DC.
Halliday, T. (2008). “Migration, Risk and the Intra-Household Allocation of Labor in El Salvador.”
The University of Hawaii.
Hanson, G. (2008). “International Migration and Development.” Working Paper #42, Commission
on Growth and Development, World Bank, Washington DC.
Mansuri, G. (2006a). “Migration, School Attainment and Child Labor: Evidence from Rural
Pakistan.” World Bank Policy Research Working Paper 3945.
Mansuri, G. (2006b). “Migration, Sex Bias, and Child Growth in Rural Pakistan.” World Bank Policy
Research Working Paper 3946.
109
Mansuri, G. (2007). “Temporary Migration and Rural Development,” in C. Ozden and M.
Schiff (eds) International Migration Policy and Economic Development: Studies Across the
Globe. Washington, DC: World Bank.
McKenzie, D. and M. Sasin (2007). “Migration, Remittances, Poverty, and Human Capital:
Conceptual and Empirical Challenges.” World Bank Policy Research Working Paper 4272.
McKenzie, D. and H. Rapoport (2006). “Can Migration Reduce Educational Attainment? Evidence
from Mexico.” World Bank Policy Research Working Paper 3952, Washington DC.
National Commission for the Placement and Protection of Indonesian Migration Workers
(BNP2TKI), 2009.
Pfeiffer, L. and J.E. Taylor (2008). “Gender and the Impacts of International Migration: Evidence
from Rural Mexico” in A. Morrison, M. Schiff, and M. Sjoblom (eds) The International Migration of
Women. The World Bank, Washington, DC.
World Bank (2008a). The International Migration of Women, edited by A. Morrison, M. Schiff, and
M. Sjoblom. Washington, DC.
World Bank (2008b). “The Malaysia-Indonesia Remittance Corridor.” World Bank Working Paper
No. 149. Washington, DC.
Yang, D. (2008). "International Migration, Remittances and Household Investment: Evidence from
Philippine Migrants' Exchange Rate Shocks." Economic Journal, Royal Economic Society, vol.
118(528), pages 591-630, 04.
Yang, D. and H. Choi (2007). “Are Remittances Insurance? Evidence from Rainfall Shocks in the
Philippines.” The World Bank Economic Review (May).
110
Chapter 5: The Effects of Immigration on the Thai Wage Structure
DILAKA LATHAPIPAT1
The World Bank
ABSTRACT: This paper examines the impact of low-skilled immigration on the industry
structure and the wages of native and existing foreign workers in Thailand. In contrast to Thai
workers with comparable education, we find no evidence of the Hecksher-Olin style absorption
of foreign labor in any immigrant-intensive industry. This is primarily due to the temporary
nature of foreign workers in Thailand. Like previous studies, we also find that the negative
effects of immigration on low-skilled Thai wages are small. By comparison, immigration has a
much more negative impact on the wages of existing foreign workers. Younger Thai workers
with less than a high school education also suffer more from immigration than their older
counterparts. Finally, we find that immigration raises the productivity of high-skilled Thai
workers with high school and college education.
1. INTRODUCTION
During the five-year period from 2002 to 2007, the total number of immigrants in Thailand
roughly doubled. As of 2007, foreign workers represented about 5 percent of the country’s workforce
of 36 million (see Table 2 in Martin 2007). These immigrant workers are largely low-skilled with less
than a high school education, and are employed mainly in the agricultural sector, fishing industry,
labor intensive manufacturing, construction, and some service industries. Most immigrant workers in
Thailand come from the Greater Mekong Sub-region (with approximately 75 per cent from
Myanmar, and 12 per cent each from Cambodia and the Lao People Democratic Republic). Around
three quarters of these immigrant workers are unregistered or undocumented.
In this paper we examine the effects of low-skilled immigration on the industry structure and
wage outcomes of Thai and existing foreign workers in Thailand. Because of limitations in our data
on immigrants, we employ a geographical approach in our analysis of the effects of immigration on
industries across different provinces in Thailand. Our focus is primarily on the effects of
immigration on foreign and Thai workers with less than a high school education.
One possibility suggested by the Hecksher-Olin international trade model is that

geographical differences in the relative supplies of a type of workers can be absorbed by a shifting
industry structure, that is, through the expansion or contraction of industries that are intensive in the
use of that particular type of labor input, and with little or no change in relative wages. The
framework that we employ to test this hypothesis in Thailand is borrowed from Card and Lewis
(2005). As will be explained in the following section, we make a small modification to this
decomposition method so that we can make statements about the part of labor absorbed into
unemployment.
Since the idealised Hecksher-Olin conditions are unlikely to be satisfied in a real world
situation like Thailand, we would expect that some portion of the excess supplies will be absorbed
through increases in the employment intensity of the abundant skill group. This adjustment would
1
This paper was written when the author was with the Thailand Development Research Institute.
111
likely entail changes in relative wages. Our contribution to the analysis of changes in the wage
structure due to immigration is to adapt an econometric model proposed by Ottaviano and Peri (2008)
that was originally used to study the effects on wages arising from immigrant labor supply shocks
over time in the United States. Due to the data constraint mentioned, we have to modify the model so
that it can be used on our cross-province analysis in Thailand. One further simplification that we
make to the model is to assume that local industries adjust their physical capital to restore the long-
run equilibrium growth path. In other words, the capital-labor ratio is unaffected by the labor supply
shocks from immigration.
2. ABSORPTION OF IMMIGRANTS AND CHANGES IN THE INDUSTRY STRUCTURE
Data Processing for the Study on Immigrant Labor Absorption
This study uses data from the third quarter of the 2007 Thai Labor Force Survey (LFS)
collected by the National Statistical Office (NSO). The advantage of using these data is that they
provide the most recent coverage of the (registered) immigrant labor force in Thailand. The surveys
do not usually cover foreign workforce, and the previous one that did collect information on
immigrants was the 2001 survey. The main disadvantage of the 2007 Labor Force Survey, however,
is the significant under-coverage of foreign immigrants who are mostly low-skilled.2 With this caveat
in mind, our sample contains men and women between 16 and 65 years of age, who have completed
formal schooling at the time of survey. We divide workers into four broad skill groups according to
the number of years of formal education completed. The four schooling groups are: lower primary
(schooling  3 years), upper primary (schooling between 4 to 6 years), high school (schooling
between 7 to 12 years), and college (schooling  12 years).3
We eliminate from our sample people who reported positive working hours and zero wages,
as well as unpaid family workers. To construct our measures of labor supply and unemployment, we
primarily employ the variables “MORE_HR” and “TOTAL_HR” from the LFS, which record the
number of additional hours per week that the person would like to work, and the usual number of
hours worked per week (in all jobs). The sum of these two variables gives us the (weekly) labor
supplied by each person. Each worker therefore has a measure of hours employed, as well as hours
unemployed. Using these measures of labor supply and unemployment rather than the conventional
metrics allows us to better capture the extent of involuntary underemployment in the labor market.4 A
person who was unemployed, but reported to be seeking work during the 30 days prior to the
interview is counted as fully unemployed, and his/her desired working hours is the total labor supply
of this person. Following normal convention, we discard persons who recorded zero hours of work
and were not seeking work during the 30-day period before the interview. That is, they are regarded
as not being in the labor force.
The resulting sample contains 52,416 individuals. It should be mentioned that while 864 of
these individuals reported that they were unemployed and were seeking work, they did not report
their desired number of working hours. These missing labor supply data are imputed using regression
2
By some estimates, the undercount could be as high as 80 per cent. This arises mainly from the presence of unauthorised
workers from the Greater Mekong Sub-region.
3
The variables used to construct the variable “years of schooling” are “GRADE_A” and “GRADE_B”. The
exact definitions are available upon request.
4
The unemployment rate calculated using our broader measure of labour supply is 2.87 per cent, while the
conventional headline measure would have indicated an unemployment rate of 1.81 per cent
112
analysis.5 The supply of foreign labor accounted for 0.91 per cent of the total weekly hour supply in
our sample. Note that the LFS sampling weights are used in the calculations of all average and
aggregate statistics and variables throughout this study. Our definitions of different industries in
Thailand follow those given in the LFS data dictionary. In particular, we identify fourteen types of
industry by combining “agriculture, hunting and forestry” and “fishing” into “agribusiness”, and
“other community, social and personal service activities”, “private households with employed
persons” and “extra-territorial organisations and bodies” into “other industries”. The remaining
industries are defined as in the data dictionary. The summary statistics and the industry distribution of
weekly hours supplied by nativity-schooling groups are shown in Tables A1 and A2 in the Appendix.
Theoretical Framework
In order to evaluate the responses of employers in different industries and locations to

differences in the composition of skills across local labor markets in Thailand, we utilise a simple
decomposition framework proposed by Card and Lewis (2005) that is an extension of a
decomposition method proposed by Lewis (2003). The Hecksher-Olin international trade model
suggests that an expansion in the relative supply of a particular skill group can be absorbed by a
shifting industry structure. Industries that are intensive in the use of that particular type of labor input
would expand, and there would be little or no change in relative wages. Before we describe the Card-
Lewis decomposition in detail, it should be noted that this method decomposes the variation in the
overall fraction of a particular skill group employed in a given city from the national average into
three distinct components. These components are the “between industry component” (B), the “within
industry component” (W), and the interaction component (I), which is an ambiguous term that cannot
be assigned to either of the between or within components of the excess supply of labor. In this paper
we extend the Card-Lewis model to decompose the variation in the overall fraction of a particular
skill group supplied in a given city from the national average instead. This modification necessarily
entails a fourth component, which is the “unemployment component” (U).
The decomposition starts with the following identity that expresses the fraction of workers in
i
skill group i, defined over educational qualifications, supplied in province p – denoted by s ( p) - as
a weighted sum of that particular skill intensity in each industry j plus the fraction unemployed
sui ( p ) :
Li ( p) 1 u i ( p) N j ( p) N ij ( p) i
(2.1) s i ( p)    j
L( p ) L( p ) j
N i
( p ) 
L( p )
  j L( p) N ( p)  su ( p)
j
  j  j ( p) ij ( p)  sui ( p)
i
where L(p) is the total supply of labor in province p, L ( p) is the total supply of skill group i in
province p, N ij ( p ) is the total employment of skill group i in industry j in province p, N j ( p ) is the
total employment in industry j in province p,  j ( p )  N j ( p ) / L ( p ) is the share of labor supply
employed in industry j in province p, and  ij ( p )  N ij ( p ) / N j ( p ) is the share of skill group i
i
employed in industry j in province p. The deviation of s ( p) from the national average fraction of
skill group i supply can be decomposed into the four components - B, W, I, and U - as follows,
5
The imputation was carried out for each of the four schooling groups using regressions of weekly hour supply
on a quartic in age, gender, marital status, a quadratic in years of schooling, and a set of interactions between
these variables as well as the region of residence dummies.
113
(2.2) si ( p)  si  B( p)  W ( p)  I ( p)  U ( p) ,
where the between industry component B( p)    j ( p)   j  , the within industry term
j
i
j
W ( p)   j  j  ij ( p)   ij  , the interaction term I ( p)   j  j ( p )   j   ij ( p)   ij  , and

the unemployment component U ( p )  sui ( p )  sui .
Under the Hecksher-Olin conditions, the cross-province variation in the share of skill group i
supply can be absorbed by the expansion or contraction of industries that employ this particular skill
group more intensively, that is, via B(p). To formally test this hypothesis, we compute the four
components and estimate the following set of regression equations:
(2.3) C ( p)  C  C  s i ( p)  si    C , C   B,W , I ,U 
 B , W ,  I , and U necessarily sum to one, since

Note that the regression coefficients,
equation (2.2) holds as an identity. Under the maintained hypothesis,  B would be close to unity. In
reality, however, the idealised Hecksher-Olin conditions are unlikely to be satisfied as factors of
production are not perfectly mobile and some goods are local in nature. The excess supply of a
particular skill type will cause their relative wages to fall and/or their unemployment to rise.
Industries would then be expected to adjust the intensity of factor use towards the cheaper inputs.
This adjustment would work through the term W(p).
Results
Our focus in this section is on the two least-skilled labor groups – lower primary and upper
primary. About 85 per cent of the immigrants in our sample are in these two groups (~ 67 per cent in
the lower primary and ~18 per cent in the upper primary group). he first four columns of Table 1
present OLS estimates of equations (2.3) for these two least-skilled labor groups. For lower primary
labor, only 9.8 per cent of excess supply is absorbed by the between industry component. The within
industry term, on the other hand, accounts for almost 70 per cent of the excess supply absorption.
There is thus limited evidence of the HO-style absorption of this type of labor through changing
industry structure. In other words, industries intensive in the use of lower primary labor are on
average not substantially larger in provinces where this type of labor was abundant. In contrast, a
much larger share of the absorption takes place through more intensive use of lower primary labor.
Similarly, the bottom half of Table 1 indicates that the majority of the absorption of excess upper
primary labor occurs through the within industry term. However, the between industry coefficient is
much larger than that of the lower primary group. The estimates of  B and W for the upper
primary group are 0.31 and 0.71 respectively.
The contributions of selected industries to the absorption of excess supplies of lower and
upper primary labor are presented in the last eight columns of Table 1. These industries are selected
on the basis of the size of employment of these two labor categories. In particular agribusiness,
manufacturing, construction, and wholesale industries6 are the largest employers of these workers in
the economy. Consider the agribusiness industry. An investigation of the between and within industry
6
The “agribusiness” sector is as defined in Section 2.1, while the “wholesale” sector comprises wholesale and retail trade,
repair of motor vehicles, motorcycles and personal and household goods.
114
Table 1: Absorption of Excess Lower Primary and Upper Primary Workers across Thai Provinces
Agribusiness Manufacturing Construction Wholesale
Interactio Unemplo Betwee Betwee Withi

Between Within n y- Between Within n Within n Within Between n
Industr Industr Industr Industr Indus

Industry Industry ment Industry y y Industry y y Industry try
0.696** 0.117**
Excess Fraction 0.098*** * 0.134** 0.072** 0.115*** 0.112* -0.015* 0.311*** 0.013 * -0.004 0.027
of Lower (0.02
Primary (0.021) (0.090) (0.065) (0.030) (0.029) (0.057) (0.009) (0.046) (0.011) (0.020) (0.003) 6)
Labor
Observations 75 75 75 75 75 75 75 75 75 75 75 75
R-Squared 0.302 0.730 0.119 0.382 0.271 0.188 0.040 0.472 0.067 0.659 0.045 0.049
*** p<0.01, ** p<0.05, *

p<0.1
Note: Robust standard errors in parentheses. All regressions are weighted by the provincial size of the particular skill group
supply.
115
Table 1 (Continued): Absorption of Excess Lower Primary and Upper Primary Workers across Thai Provinces
Agribusiness Manufacturing Construction Wholesale
Unempl Betwee Betwee Betwee

Between Within Interaction oy- n Within Between Within n Within n Within
Industr Industr Industr Industr Industr Industr Industr Industr

Industry y ment y y Industry y y y y y
0.714** 0.063** 0.592** - 0.279** 0.223** 0.104** 0.141**

Excess Fraction 0.309*** * -0.086 * * 0.122** 0.274*** * * * -0.030 *
of Upper
Primary (0.030) (0.066) (0.086) (0.014) (0.068) (0.058) (0.095) (0.043) (0.034) (0.016) (0.030) (0.020)
Labor
Observations 76 76 76 76 76 76 76 76 76 76 76 76
R-Squared 0.519 0.771 0.055 0.235 0.537 0.275 0.143 0.387 0.352 0.453 0.041 0.519
*** p<0.01, ** p<0.05, * p<0.1
Note: Robust standard errors in parentheses. All regressions are weighted by the provincial size of the particular skill group supply.
116
components reveals that agribusiness accounts for the entirety of the between industry components of
both skill groups. For upper primary labor, the only other industry where a higher proportion of the
absorption is associated with an increase in the size of the sector is the construction industry.1
As a final exercise, we split the lower and upper primary labor groups into Thai and foreign
portions. Provinces with no observations on immigrant workers are discarded. The above analysis is
then repeated on these more disaggregated schooling groups and the regression results are presented
in Table 2. The most striking result in this table is the substantially lower between industry
absorption of immigrant labor force relative to its Thai counterpart. For immigrant labor, only 1.8 per
cent of the excess supply fraction is associated with the between industry component, while the
estimate for similarly educated domestic workforce is 21.6 per cent. This is evidence that employers
regard immigrants as temporary workers, and as a result sectors that employ these workers more
intensively are only slightly larger, in terms of employment, in provinces where there are abundant
supplies of this type of labor. A closer examination into the contribution of selected industries to the
absorption of lower primary immigrant labor reveals that sector-specific estimates of the  B ’s are all
close to zero, and none of the estimates are statistically significant at conventional levels. Comparing
the contribution of the agribusiness industry in the absorption of native and immigrant supplies for
this schooling group reveals that while the majority of immigrant excess supplies are absorbed
through the within industry component, the exact opposite phenomenon is observed for the Thai
counterparts. A similar, but stronger conclusion can be reached with the native and immigrant upper
primary skill groups from the regression results presented in bottom half of Table 2
It is also interesting to note from Table 2 that, unlike for their Thai counterparts, the excess
supplies of immigrant workers in both the lower and upper primary schooling groups are essentially
uncorrelated with unemployment. This fact should not come as a surprise, since the observations of
immigrants in our dataset are all registered workers, and hence are employed by definition. As for
Thai workers in these schooling groups, around 6 per cent of the excess supplies are absorbed into
unemployment.
3. THE EFFECTS OF IMMIGRATION ON THE WAGE STRUCTURE
The findings from the previous section suggest a very small role of shifting industry structure
in absorbing the excess supplies of low-skilled immigrant workers across Thai provinces. Most of the
absorption occurs through increases in the intensity of factor employment within industries. In
contrast to the idealised HO-style absorption, the results indicate that the rates of unemployment
and/or the wage structure must adjust to accommodate changes in immigrant supplies.
As discussed earlier, our data on immigrant workers are restricted to those who are registered
and employed. This severely limits our ability to analyse the part of excess immigrant labor
absorption through unemployment. However, we can still analyse the effects on the wage structure
from shifts in supplies of immigrants in different skill groups. Furthermore, we have earlier shown
that immigrants and natives are imperfect substitutes in production because immigrants tend to have
only temporary status in Thailand. This suggests that the imperfect substitutability between
1
To complete the picture, OLS regressions of the absorption of excess high school and college labour are also estimated.
Again, there is limited evidence of the HO-style absorption. For high school labour we estimated B=0.431 (SE. 0.038),
W=0.466 (SE. 0.052), and U=0.007 (SE. 0.011). For college labour; B=0.323 (SE.0.037), W=0.641 (SE.0.064), and
U=0.024 (SE. 0.004). An interesting observation is that substantially lower fractions of excess labour supplies in these
relatively high-skilled groups were associated with higher unemployment.
117
Table 2: Absorption of Excess Lower and Upper Primary Thai and Immigrant Workers across Thai Provinces
Agribusiness Manufacture Construction Wholesale
Betwee Interactio Unemploy Betwee Betwee Betwee

n Within n - n Within Between Within n Within n Within
Industr Industr Industr Industr Industr Industr Industr Industr Industr

y y ment y y Industry y y y y y
0.740** 0.307** 0.151**

Excess Fraction 0.018** * 0.241*** 0.001 -0.001 0.182** 0.035 * -0.005 0.052* -0.003 *
of Lower
Primary (0.008) (0.042) (0.048) (0.002) (0.017) (0.086) (0.028) (0.070) (0.004) (0.026) (0.005) (0.026)
Immigrant
Labor
Observations 35 35 35 35 35 35 35 35 35 35 35 35
R-Squared 0.312 0.554 0.116 0.031 0.001 0.453 0.195 0.176 0.181 0.170 0.053 0.674
0.216** 0.578** 0.220** - 0.218** 0.039** 0.105**

Excess Fraction * * 0.140** 0.066* * 0.056 0.024*** * * * -0.003 0.016
of Lower
Primary (0.039) (0.119) (0.068) (0.033) (0.052) (0.073) (0.006) (0.048) (0.010) (0.022) (0.003) (0.024)
Thai Labor
Observations 46 46 46 46 46 46 46 46 46 46 46 46
118
Agribusiness Manufacture Construction Wholesale
Betwee Interactio Unemploy Betwee Betwee Betwee

n Within n - n Within Between Within n Within n Within
Industr Industr Industr Industr Industr Industr Industr Industr Industr

y y ment y y Industry y y y y y
R-Squared 0.545 0.769 0.262 0.229 0.484 0.051 0.23 0.608 0.39 0.676 0.043 0.022
0.004** 0.914** 0.142** 0.534** 0.024** 0.005** 0.174**

Excess Fraction * * 0.091*** -0.009 0.000 * -0.001 * 0.000 * * *
of Upper
Primary (0.001) (0.010) (0.003) (0.009) (0.001) (0.002) (0.001) (0.011) (0.000) (0.005) (0.001) (0.005)
Immigrant
Labor
*** p<0.01, ** p<0.05, * p<0.1
Note: Robust standard errors in parentheses. All regressions are weighted by the provincial size of the particular skill group supply.
119
immigrant and native workers within the same narrowly defined skill group should be taken into
account in our analysis of the effects of immigration on the wage structure.
Theoretical Framework
The theoretical framework proposed by Ottaviano and Peri (2008) (referred to as OP (2008)
from here on) allows a thorough examination of the effects of immigration on the wages of natives
and existing immigrant workers in similar or in different skill groups through a network of cross
elasticity parameters. The skill groups are defined over their highest educational attainments and
years of labor market experience. The key difference between our approach and the “national
approach” of OP (2008) is that we evaluate variations in employment and wages of different
narrowly defined skill groups across the Thai provinces, instead of over time. The geographical
approach taken here is employed because we lack sufficient time series data on immigrant labor
force. Furthermore, we make an implicit assumption that regions adjust their physical capital to
accommodate variations in labor quantity in the event of supply shocks.
We begin by defining a general production function of the form F ( K p , N p ) , where Kp and

Np are the physical capital and the Constant Elasticity of Substitution (CES) aggregate of different
types of labor in province p respectively. Following OP (2008), the labor aggregate Np is defined as:
 HL
 HL 1  HL 1  1
  HL
(3.1) N p   Lp N Lp   Hp N Hp HL 
 HL
 
where NLp and NHp are aggregate measures of labor with low (L) and high (H) education observed in
province p respectively. The  ’s used throughout are relative productivity levels specific to the
particular skill groups - indexed by subscripts - within the same CES nest. The parameter  HL is the
elasticity of substitution between the two broad schooling groups. These two groups, L and H, are in
turn CES aggregates of detailed schooling groups of lower primary (LP), upper primary (UP), high
school (HS), and college (CO) labor as follows:
 LL  HH
 LL 1  LL 1  1  HH 1  HH 1  1
  LL  LL
 LL   HH  HH
 HH
(3.2) N Lp   LPp N LPp  UPp NUPp  and N Hp   HSp N HSp   COp N COp 
   
where the parameters  bb ’s are the elasticity of substitution parameters between education subgroups
within a broad schooling group b, where b {L, H } .
A detailed education group k {LP,UP, HS , CO} further nests labor groups with different
experience levels. In the spirit of Card and Lemieux (2001), this specification allows us to explore the
possibility that similarly educated workers in different experience groups are imperfect substitutes in
production. Specifically:
 EXP
 EXP 1 
 3  EXP 1
(3.3) N kp    j 1 kj N kjp EXP 
 
where  EXP is the elasticity of substitution between workers with different experience levels within
the same detailed education subgroup, and the subscript j indexes the experience group. In this paper,
120
we separate workers into three experience levels. That is, workers with 0-14 years, 15-29 years, and
30 or more years of experience are allocated to groups j=1,2, and 3 respectively.1 Note also that in
order to conserve degrees of freedom, we follow OP (2008), OP (2006), Borjas (2007)2, and Borjas
(2003) and assume that the experience-education specific relative efficiency parameters,  kj ’s, are
constant across provinces.
Finally, the N kjp ’s are CES aggregates of supplies of domestic ( Dkjp ) and foreign ( Fkjp )
workers within the same k, j, education-experience cell:
 IMMI
 IMMI 1  IMMI 1 
  IMMI  IMMI
 IMMI 1
(3.4) N kjp   Dkj Dkjp   Fkj Fkjp 
 
where  IMMI is the elasticity of substitution between domestic and immigrant workers.
Equating the value of marginal product of labor of a generic domestic worker in the b, k, j
skill group to the wage rate, it is straight forward to show that:
(3.5) wDbkjp   p  N p / Nbp  Nbp / N kp  N kp / N kjp  N kjp / Dkjp 
where  p  q p FN ( K p , N p ) , q p is the price level of aggregate output in province p, and FN () is the
marginal product of the composite labor Np. From equations (3.1) to (3.4), solve the various
derivatives, then take logarithm, and rearrange the right hand side to get:
1  1 1 
(3.6) ln( wDbkjp )  ln  p  ln( N p )  ln bp     ln( Nbp )  ln  kp 
 HL   HL  bb 
 1 1   1 1 
   ln( N kp )  ln  kj     ln( N kjp ) 
  bb  EXP    EXP  IMMI 
1
ln  Dkjp  ln( Dkjp )
 IMMI
The log wage equation for a generic immigrant worker in the same b, k, j skill group is
derived in the same manner.
Description of Data and Variable Construction
The data used in this part of the study are the same as those used in Section 2 to study the
absorption of immigrant labor. For this study on wages, however, we drop the 864 unemployed
workers from the sample.
We begin by describing how the hourly wage rate for each individual is calculated. The total
monthly wage earning is obtained by summing approximate monthly wage earning, “APPROX”,
with annual “BONUS” divided by twelve, and average overtime payment received per month, “OT”.
1
Details for the construction of this, as well as other variables discussed here are explained in Section 3.2.
2
See chapter “The Evolution of the Mexican-Born Workforce in the United States” by Borjas, G. and L. Katz.
121
The total monthly earning is then divided by four to obtain the weekly wage rate, which is then
further divided by the usual number of hours worked per week to obtain the hourly wage rate.3
The weekly hours supplied is constructed by multiplying the usual number of hours worked
per week for each individual by the sampling weight. The total weekly wage bill for each (weighted)
individual is then calculated by multiplying the weekly wage rate by the hours supplied variable. The
next step is to aggregate the hours supplied of all individuals in each province, as well as their wage
bills into 4 schooling × 3 experience × 2 nativity groups. This aggregation gives the total weekly hour
supplies of foreign ( Fkjp ) and native ( Dkjp ) workers in each k, j education-experience cell in
province p. The average wage rate for each cell is calculated by dividing the aggregate wage bill by
the corresponding aggregate weekly hour supply to obtain wFbkjp and wDbkjp for foreign and Thai
labor groups respectively.
Note that the four detailed schooling groups and the three experience groups are as defined
earlier. We assume that workers with less than a high school education entered the labor force at the
age of 15, and those with a high school education entered at the age of 17. For the remaining workers
with post-secondary education, their (potential) experience variable is calculated from Age –
Schooling – 5.4
Estimation of Elasticity of Substitution Parameters
Elasticity of Substitution between Immigrants and Natives
Following OP (2008), we start with the estimation of the elasticity of substitution between
workers in the most disaggregated level of the CES structure, and then work progressively up through
larger labor aggregates. Specifically, we begin with the estimation of the elasticity of substitution
between domestic and immigrant workers  IMMI within the same k, j education-experience cell. To
estimate  IMMI , we take a difference between the native log wage equation (3.6) and its foreign
counterpart to obtain:
w    Fkj  1 F 
(3.7) ln  Fbkjp   ln    ln  kjp 
w  
 Dbkjp    Dkj   IMMI  Dkjp 
where wFbkjp / wDbkjp measures the relative (mean) hourly wage of immigrants and natives in detailed
education group k (within broad group b) and experience group j in province p. The relative weekly
hour supply between immigrants and Thais is given by Fkjp / Dkjp , while  Fkj /  Dkj is the relative
foreign-native productivity assumed to be constant over provinces. Equation (3.7) leads to the
following regression equation that is used to estimate  IMMI :
3
Note that it is important to use hourly wages to accurately gauge the prices of labour in different skill-nativity groups. To
see this, consider workers with lower primary education. For Thai workers in this skill group, the average hours supplied
was around 45 hours per week, while it was almost 54 hours for foreign workers. Thai and foreign workers with upper
primary education supplied 47 and 51 hours respectively. Using monthly wage earnings, say, would severely bias our
parameter estimates in the next section.
4
It should also be mentioned that before we aggregated the hour supplies and the wage bills into education-experience-
nativity groups; we had dropped immigrant workers with a high school education in experience groups j=2 or 3 (those with
more than 14 years of experience), since there are only ten of these observations (from more-than-50,000 sample). The only
one immigrant observation in our sample with a college education was also eliminated prior to the aggregation into skill
groups.
122
w  1 F 
(3.8) ln  Fbkjp   I kj  ln  kjp   ubkjp
w  IMMI  Dkjp 
 Dbkjp 
where the relative foreign-native productivity terms are captured by the education×experience fixed
effects I kj . As in OP (2008), we treat the residual ubkjp as potentially correlated within education-
experience groups, and hence the regression estimate of 1/  IMMI is reported with clustered robust
standard error in the first column of Table 3
Table 3: Regression Estimates of the Elasticity of Substitution Parameters
 IMMI  EXP  LL  HH  HL
(1) (2) (3) (4) (5)
Inverse Elasticity -0.016 -0.037 -0.006 -0.061 -0.101**

Estimate (0.021) (0.028) (0.037) (0.073) (0.055)
p-value (one tail) 0.232 0.106 0.435 0.204 0.034
Elasticity of Substitution ˆ IMMI  60.85 ˆ EXP  26.77 ˆ LL  162.51 ˆ HH  16.50 ˆ HL  9.90
Fixed Effects:
Education×Experience Yes Yes No No No
Education×Province No Yes No No No
Province No Yes Yes Yes Yes
Observations 111 880 75 76 75
R-Squared 0.117 0.959 0.053 0.440 0.588
*** p<0.01, ** p<0.05, * p<0.1

Note: Robust standard errors in parentheses.
The first two columns report (Education×Experience) clustered robust standard errors, while the rest report heteroskedasticity
robust standard errors. Due to insufficient degrees of freedom, the regressions in columns (3) to (5) use five region dummies to absorb
the province fixed effects.
Unfortunately, we do not have observations on immigrant workers with more than a high
school education. Furthermore, while we do observe foreign workers with lower primary and upper
primary education in all three experience groups in the overall sample, we observe them only in the
experience group j=1 (0-14 years) for the high school group. Therefore, we only have seven (out of
twelve) education×experience fixed effects in our regression. In addition, we do not observe
immigrants in many of the provinces in the sample. For some provinces with immigrant workers, we
123
do not even observe them in all of the remaining seven narrowly defined skill groups. That is why,
out of the available 76 provinces and 7 skill groups, the number of observations in our regression is
only 111. With this limited dataset, our estimate of the inverse elasticity 1/  IMMI is -0.016 (
ˆ IMMI  60.85 ), and it is statistically insignificant with a p-value of 0.232 (one tail test).
Next we construct the composite labor input N kjp for the k, j education-experience group in
province p. From the fixed effect estimates Iˆkj , we follow OP (2008) and compute the assumed
geographically invariant5 efficiency terms  Fkj and  Dkj by first normalising them to sum to one as
follows:
exp( Iˆkj ) 1
(3.9) ˆFkj  , and ˆDkj 
1  exp( Iˆkj ) 1  exp( Iˆkj )
The estimates from (3.9), together with ˆ IMMI from regression (3.8) are used to construct the
composite labor input Nˆ kjp as per equation (3.4). In the next section the composite labor input is
used in estimating the elasticity of substitution  EXP between workers within the same detailed
education group, but having different experience levels.
Elasticity of Substitution between Experience Groups
A marginal pricing condition similar to that given in equation (3.5), but at one level of aggregation
higher up implies that:
1  1 1 
(3.10) ln( wkjp )  ln  p  ln( N p )  ln bp     ln( Nbp )  ln  kp 
 HL   HL  bb 
 1 1  1
   ln( N kp )  ln  kj  ln( N kjp )
  bb  EXP   EXP
where wkjp  wFbkjp  Fkjp / N kjp   wDbkjp  Dkjp / N kjp  is the (weighted) average wage for a k, j
education-experience cell in province p. The inverse elasticity parameter 1/  EXP can be estimated
from an empirical version of equation (3.10) as follows:6
1
(3.11) ln( wkjp )  I p  I kp  I kj  ln( Nˆ kjp )  ukjp
 EXP
where the province fixed effects Ip absorb the term ln  p  1 /  HL ln( N p ) , the schooling×province
fixed effects Ikp control for the term ln  bp  1 /  HL  1 /  bb  ln( N bp )  ln  kp 
5
Note that due to the geographically invariant assumption, the estimated ˆDkj ’s and ˆ IMMI are used to rescale
the native labour supplies in the corresponding k, j education-experience cells (up to high school level in
experience group j=1) for all provinces with missing immigrant supplies using equation (3.4).
6
In contrast to our cross-province approach, OP (2008) estimate a “national approach” version of equation (3.13) by 2SLS,
using the (logarithm of) hours worked by immigrants in the k, j cell at time t, ln( Fkjt ) , as instruments for ln( Nˆ kjt ) . While
the instrumental variable method is the preferred estimation approach that addresses the attenuation bias problem, the
method is not viable for our particular application due to our insufficient data on foreign workers discussed earlier.
124
1 /  bb  1 /  EXP  ln( N kp ) , and the schooling×experience fixed effects Ikj capture the geographically
invariant ln  kj term. The random error u kjp is again treated as potentially correlated within
education-experience groups. Note that the composite labor input Nˆ kjp constructed in the previous
section is used in the regression (3.11). Our estimate of the inverse elasticity of substitution between
experience groups, 1/  EXP , reported in the second column of Table 3 is -0.037 ( ˆ EXP  26.77 ).
Clustered robust standard error is again reported. The estimate is not statistically significant at
conventional levels and has a p-value of 0.106.7
Following OP (2008), the estimates of ˆkj ’s are obtained from the twelve
education×experience fixed effects Iˆkj (4 schooling×3 experience) from regression (3.11) using the
following normalisation:
exp( Iˆkj )
(3.12) ˆkj 
 j 1 exp( Iˆkj )
3
 EXP , together with the ˆkj ’s are then used to construct a further
The estimated value of
level of CES composite labor input Nˆ kp using equation (3.3) for each schooling group
k {LP,UP, HS , CO} .
Elasticity of Substitution between Schooling Groups within the same Broad Education Category
Consider first the two lowest schooling groups; the lower primary (LP) and the upper primary (UP)
groups within the same broad education category L. With one further level of CES aggregation, the
now-familiar marginal pricing condition implies that:
1  1 1 
(3.13) ln( wkp )  ln  p  ln( N p )  ln  Lp     ln( N Lp )  ln  kp 
 HL   HL  LL 
1
ln( N kp ), for k   LP,UP
 LL
where wkp   j
( N kjp / N kp ) wkjp is the supply-weighted average (over all experience groups) wage
for schooling group k, and wkjp is as in the previous section. Applying the method of (Katz and
Murphy 1992) – KM (1992) from here on – to this particular CES nesting structure, we take a
difference between expression (3.13) for the upper primary and that for the lower primary schooling
groups to obtain:
7
Another version of regression (3.11) is also estimated, where foreign and native workers are assumed to be perfect
substitutes in production. However, this estimate is not reported in Table 3.1. Specifically, the composite labour supply
N kjp is simply constructed as the sum of immigrant and native hours in the corresponding k, j education-experience cell for
each province p. The resulting estimate of 1/  EXP is -0.038 ( ˆ EXP  26.34 ) with a p-value of 0.103. Note that the
similarity of the estimates obtained under the perfect and imperfect substitution assumptions is due to the large, but finite
estimate of  IMMI .
125
w   UPp  1 N 
(3.14) ln  UPp   ln    ln  UPp 
w 
 LPp    LPp   LL  N LPp 
A regression version of equation (3.14) is used to estimate the inverse elasticity parameter
1/  LL :
w  1  Nˆ UPp 
(3.15a) ln  UPp 
 LpI  ln    uLp
w  LL  Nˆ LPp 
 LPp  
where the province fixed effects ILp control for the geographical variation in ln UPp /  LPp  for this
broad education category L, and the composite labor inputs Nˆ UPp and Nˆ LPp are obtained from the
previous section. An identical derivation to the one given above for high school and college educated
groups yields the following equation that is used to estimate the inverse elasticity parameter 1/  HH :
w  1  Nˆ 
(3.15b) ln  COp   I Hp  ln  COp   uHp
w  HH  Nˆ HSp 
 HSp  
Note that we do not have enough degrees of freedom to estimate regressions (3.15a) and
(3.15b) if the fixed effects for all provinces are included. Instead, we group the provinces into five
regions, namely; Bangkok, Rest of Central, North, Northeast, and South.
Columns 3 and 4 of Table 3 report our estimates of the inverse elasticities 1/  LL and
1/  HH together with their robust standard errors. The estimate of 1/  LL is -0.006 (p-value of
0.435). This implies a very high degree of substitutability ( ˆ LL =162.51) between lower primary and
upper primary workers in production. On the other hand, our estimate of 1/  HH obtained using
equation (3.15b) is around -0.061 (p-value of 0.204), implying a much lower elasticity of substitution
between college and high school workers ( ˆ HH =16.50). Once again, both of our estimates are not
statistically significant at conventional levels.
As before, the estimated region fixed effects for each broad education category Iˆbp are used
to compute the standardised relative productivity terms ˆlp ’s and ˆmp ’s for the relevant (l,m)-pair of
detailed schooling groups in the broad education category b.8 Specifically:
exp( Iˆbp ) 1
(3.16) ˆlp  , and ˆmp 
1  exp( Iˆbp ) 1  exp( Iˆbp )
The estimates obtained from (3.16), together with the ˆ bb estimated for each broad
schooling category b {L, H } are then used to construct the composite labor inputs Nˆ Lp and Nˆ Hp
as per (3.2).
Elasticity of Substitution between Broad Education Categories
8
Recall that at this level of CES nesting structure, the relative productivity terms are allowed to vary by region.
126
Aggregating one level still further yields the following log wage equation for each broad education
category b:
1 1
(3.17) ln( wbp )  ln  p  ln( N p )  ln bp  ln( N bp ), for b   L, H 
 HL  HL
   
where wbp  wlp Nlp / Nbp  wmp N mp / Nbp , for a relevant (l,m)-pair in the broad education
group b, and wlp and wmp are obtained from the preceding section. Apply the KM (1992) procedure
once again to obtain the following relative wage expression:
w    Hp  1 N 
(3.18) ln  Hp   ln    ln  Hp 
w 
 Lp    Lp   HL  N Lp 
Equation (3.18) serves as a basis for the following regression that is used to estimate the
parameter 1/  HL :
w  1  Nˆ Hp 
(3.19) ln  Hp   I p  ln    up
w  HL  Nˆ Lp 
 Lp  
where the composite labor inputs Nˆ Lp and Nˆ Hp for each province are computed in the previous
section. In order to conserve degrees of freedom, we again use the region fixed effects instead of the

province fixed effects to control for variations in ln  Hp /  Lp . 
The estimate of 1/  HL obtained using regression (3.19), and reported in the last column of
Table 3 is -0.101 ( ˆ HL  9.90 ) with a robust standard error of 0.055. It is thus quite significantly
different from zero (p-value of 0.034).
The relative magnitudes of all the elasticity of substitution estimates presented in Table 3 are
very plausible, even though only one of the five inverse elasticity estimates is statistically significant.
From the first two columns, we can see that immigrants and Thais within the same k, j education-
experience group are closer substitutes in production than similarly educated workers with different
experience levels. Between workers with lower primary and upper primary schooling, their
substitutability in production is extremely large ( ˆ LL =162.51). By contrast, the elasticity of
substitution between high school- and college-educated workers is smaller by a factor of ten. In
regards to the comparative magnitudes of the two elasticity estimates, ˆ LL and ˆ HH , we would
expect the substitutability between workers in the two broad education categories, H and L, to be
relatively much lower. This is indeed the case, as is shown in the last column of Table 3, where ˆ HL
=9.90.
Simulating the Effects of Immigration on the Wage Structure
Equipped with the estimates of  IMMI ,  EXP ,  LL ,  HH , and  HL from Section 3.3, we
now proceed to the final step of investigating the effects of immigration on the Thai wage structure
using simulation.
Recall that the wage model considered in this paper is a long-run model in which the capital-
labor ratio is unaffected by local labor supply shocks. Denote the change in the supply of foreign
127
labor due to immigration in a k, j schooling-experience group by Fbkjp / Fbkjp . Consistent with the
demand for native labor equation (3.6), the following expression is used to derive the total change in
the wage rate of a generic Thai worker in the k, j skill group in response to immigrant supply shocks
across all skill groups:
wDbkjp 1 3 wFcqip Fcqip Fcqip  1 1  3 w
Fbqip Fbqip Fbqip
(3.20) 
ˆ HL
    
wDbkjp cB qE i 1 wp N p Fcqip  ˆ HL ˆ bb  qb i 1 wbp Nbp Fbqip
 1 1  3 wFbkip Fbkip Fbkip
  
 ˆ bb ˆ EXP  i 1 wkp N kp Fbkip
 1 1  wFbkjp Fbkjp Fbkjp
  
 ˆ EXP ˆ IMMI  wkjp N kjp Fbkjp
 
where B   L, H  , E   LP , UP , HS , CO , and w  wL N Lp / N p  wH N Hp / N p . The rest of  
the (weighted) average wages from various levels of the CES nesting structure are calculated as
shown in Section 3.3. The change in the wage rate for a corresponding immigrant worker is simulated
as follows:
wFbkjp wDbkjp 1 Fbkjp
(3.21)  
wFbkjp wDbkjp ˆ IMMI Fbkjp
Take note that expressions (3.20) and (3.21) are equivalent to expressions (25) and (26) in
Appendix A of OP (2008), without the adjustment in the capital-labor ratio in response to
immigration.
Using equations (3.20) and (3.22), as well as the estimated substitution elasticities, the
simulated long-run effects on wages of Thai and immigrant workers in all of the k, j education-
experience groups are reported in columns 5 and 6 of Tables 3.2 and 3.3. By some estimates, over the
five-year period from 2002 to 2007, the number of foreign workers increased rapidly from nearly
0.97 million to 1.8 million (see Table 2 in Martin (2007)). Hence, we have chosen our simulated
increases in foreign workers Fbkjp / Fbkjp to be 100 percent across all k, j skill groups. For ease of
exposition, columns 7 and 8 of Tables 3.2 and 3.3 also present aggregated effects on the average
wages of the schooling-nativity groups. As an example, the change in the college-educated native
wage rate is calculated as follows:
3
w D ,CO , j DD ,CO , j  wD ,CO , j / wD ,CO , j 

i 1
(3.22) 3
w
i 1
D ,CO , j DD ,CO , j
In words, the change in the average wage of college educated Thai workers is the weighted
average of the wage changes of similarly educated Thais in all experience groups, where the weights
are equal to the experience group-specific wage bills as fractions of the total wage bill paid to native
college workers.
The simulation results presented in Tables 1 and 2 require some explanation. In Table 4 we
have aggregated the hour supplies of all skill types over all 44 provinces that recorded positive hours
of immigrant supplies in any skill group in 2007. The aggregate hour supplies of foreign and Thai
workers are reported in columns 3 and 4 respectively. The corresponding skill-specific average
hourly wage rates (prior to immigrant inflows) are also reported in columns 1 and 2. From the last
four columns, we can see that the effects from our simulated balanced immigration inflows are very
128
Table 4: Simulated Long Run Effects of Immigration on the Thai Wage Structure using all Provinces with Positive Immigrant Hours
Experien Hourly Hourly Hours Hours Foreign Thai Foreign Thai

Schooling ce Wage Wage Supply Supply Wage Wage Wage Wage
Foreign Change Change
(Years) (Baht) Thai (Baht) Foreign Thai Change % % Change % %
(1) (2) (3) (4) (5) (6) (7) (8)
Lower
Primary 0-14 13.97 17.86 2,653,409 3,452,363 -2.07 -0.43 -1.85 0.15
15-29 19.10 14.11 1,663,214 6,283,221 -1.56 0.08
30 plus 22.69 26.65 126,322 6,943,878 -1.16 0.48
Upper
Primary 0-14 12.11 19.96 733,507 32,100,000 -1.81 -0.17 -1.80 -0.14
15-29 21.75 27.15 262,812 85,200,000 -1.78 -0.14
30 plus 22.16 26.89 189,990 53,400,000 -1.78 -0.14
High School 0-14 19.02 24.49 865,974 94,600,000 -1.62 0.02 -1.62 0.03
15-29 40.52 51,600,000 0.04
30 plus 59.12 12,100,000 0.04
College 0-14 60.58 73,700,000 0.06 0.06
15-29 78.62 39,000,000 0.06
30 plus 180.58 9,462,483 0.06
Note: The percentage of Immigrant to total hours supplied prior to simulated shocks in all provinces with positive immigrant hours is 1.37 per cent.
The results provided are the long-run simulated effects on wages from a 100 per cent increase in the supply of immigrant workers in each of the k, j skill groups.
129
uneven across native and foreign skill groups over the long run. In particular, while we observe
negative impacts on the wages of foreign workers across all skill groups - ranging from -1.16 to -2.07
per cent, the effects are more ambiguous for Thai workers. For Thais with lower primary education,
the average wage rate for the least experienced group decline by 0.43 per cent, while those of the two
older groups increase by 0.08 and 0.48 per cent respectively. With the higher educated Thai workers,
the effects on their average wages are unambiguously positive, albeit very small, ranging from 0.02
to 0.06 per cent.
In Table 5, we repeat the same simulation of balanced immigrant inflows on a small subset of
provinces. Specifically, we selected five provinces from our sample with the highest proportion of
foreign workforce. In our view, the results obtained in this exercise are more realistic with regards to
wage responses in local labor markets, since the impacts on native and foreign wages in immigrant-
intensive labor markets would be more substantial than in other locations. If we aggregate too many
provinces with low immigrant intensity, the effects on wages would be more diffused due to the
small share of immigrant wage bill to total wage bill. This reasoning is intuitive and can be readily
inferred from equations (3.20) and (3.21). The percentage of foreign labor to total labor supplied in
these selected provinces was 9.69 per cent, compared to only 1.37 per cent for the previous exercise.
As expected, column 7 in Table 5 shows that a doubling of immigrant workforce across all
skill groups leads to slightly larger declines in foreign wages (-1.94 per cent for lower primary, and -
2.45 per cent for upper primary immigrants). However, column 8 shows that the impact on Thai
wages is greater, with average wages for high school- and college-educated workers increasing more
than half a per cent each. At a more disaggregated level, column 6 of Table 3.3 shows that younger
Thai workers suffer disproportionately more than older workers in terms of wage losses. For workers
with upper primary schooling, the wage decline for the youngest group is at 0.85 per cent, while the
fall is slightly less at 0.76 per cent for the oldest group of native workers. For the lower primary
Thais, the decline in the average hourly wage rate for the youngest group is at 0.49 per cent, while the
wage rate for the oldest group of workers actually rises by 0.62 per cent. Column 5 of Table 3.3 also
shows larger declines in wages for younger immigrant workers. The disproportionate declines are
primarily caused by the fact that 57 per cent of labor supplies of immigrants with less than a high
school education are in the youngest experience group, while the supplies of the oldest group of
workers account for only 5.6 per cent.
The results of Tables 4 and 5 show that immigration has a larger negative effect on the
wages of existing foreign workers than the wages of local workers because of their imperfect
substitutability in production. Furthermore, results suggest that inflows of low-skilled foreign
workforce raise the productivity of high-skilled native workers with high school and college
education and result in their getting higher wages. However, the size of these wage impacts is quite
small
.
4. CONCLUSION
In this paper, we examine the impact of low-skilled immigration on the industry structure and
the wages of native and existing foreign workers in Thailand. To study the relationships between skill
supplies, industry structure, and unemployment, we modify the Card-Lewis model to decompose
excess fractions of different types of skill supplies across Thai provinces into four components,
namely, the between and within industry components, an ambiguous interaction term, and an
unemployment component. Our focus is on workers with less than a high school education, since
most of the immigrants fall into this category. On the one hand, the Hecksher-Olin international trade
model suggests that an expansion in the relative supply of a particular skill group can be absorbed by
a shifting industry structure. However, we find limited evidence of this adjustment mechanism in the
130
Table 5: Simulated Long Run Effects of Immigration on the Thai Wage Structure using only the Top Five Immigrant-Abundant
Provinces
Experien Hourly Hourly Hours Hours Foreign Thai Foreign Thai

Schooling ce Wage Wage Supply Supply Wage Wage Wage Wage
Foreign Change Change
(Years) (Baht) Thai (Baht) Foreign Thai Change % % Change % %
(1) (2) (3) (4) (5) (6) (7) (8)
Lower
Primary 0-14 21.29 19.03 1,299,899 483,251 -2.13 -0.49 -1.94 -0.03
15-29 21.76 19.01 841,593 707,258 -1.77 -0.12
30 plus 22.36 17.16 104,333 479,045 -1.02 0.62
Upper
Primary 0-14 13.64 23.09 282,792 2,759,004 -2.49 -0.85 -2.45 -0.79
15-29 21.04 26.16 194,973 6,051,362 -2.42 -0.78
30 plus 24.21 27.06 49,587 2,945,617 -2.40 -0.76
High School 0-14 18.64 26.98 56,100 5,853,519 -1.09 0.55 -1.09 0.56
15-29 43.21 2,293,615 0.56
30 plus 96.76 486,764 0.56
College 0-14 55.37 2,717,392 0.57 0.57
15-29 99.97 1,319,722 0.57
30 plus 159.53 286,738 0.57
Note: The percentage of Immigrant to total hours supplied prior to simulated shocks in the top 5 provinces with positive immigrant hours is 9.69 per cent
The results provided are the long-run simulated effects on wages from a 100 per cent increase in the supply of immigrant workers in each of the k, j skill groups
131
absorption of low-skilled labor by Thai industries. Most of the absorption occurs through the within
industry term. This is especially the case for low-skilled immigrant workers, where there is strong
evidence that employers regard immigrants as temporary workers.
A closer investigation into the contribution of key industries reveals no evidence of larger
low-skilled immigrant-intensive sectors in immigrant-abundant provinces. Especially interesting is
the agribusiness sector; while the absorption of excess fractions of foreign supplies occurs entirely
through the within industry component, the exact opposite phenomenon is observed for similarly
skilled natives. There is thus evidence of the HO-style absorption of low-skilled Thai workers in the
agribusiness sector.
To examine the effects of low-skilled immigration on wages, we employ a structural model

of labor demand from Ottaviano and Peri (2008). The model is slightly modified to accommodate our
cross section dataset (geographical approach). Results show that immigration adversely affects the
wages of existing foreign workers much more than those of similarly low-skilled Thai workers.
Younger workers are also found to suffer greater wage losses than older workers. Furthermore,
inflows of low-skilled immigrants are found to raise the productivity of high-skilled Thai workers
with high school and college education. However, as has been found in previous studies,1 the effects
of immigration on the wages of all these groups are quite small. This is due to the diffusion effect of
wage changes arising from the relatively low proportions of immigrant labor supplies in most
provinces. We expect the impact of immigration on the wages of Thai and foreign workers to be
larger in immigrant-abundant provinces.
1
For example; using a Computable General Equilibrium (CGE) model, (Sussangkarn 1996) estimates that the removal of
immigrants, which account for around 2 percent of the 1995 labour force, would increase the wages of Thai workers with
less than a primary education by 3.5 percent, while the wages of more educated natives would fall. This model is based on
the assumption that foreign workers are substitutes for Thais with a primary and lower education.
132
REFERENCES
Borjas, G. 2003. The Labor Demand Curve is Downward Sloping: Reexamining the Impact
of Immigration on the Labor Market. Quarterly Journal of Economics 118(4): 1335-1374.
Borjas, George. 2007. Mexican Immigration to the United States. Cambridge Ma.: National Bureau
of Economic Research Conference Report.
Card, D. and T. Lemieux. 2001. Can Falling Supply Explain the Rising Return to College for
Younger Men? A Cohort-Based Analysis. Quarterly Journal of Economics 116(2): 705-746.
Card, D. and E. G. Lewis. 2005. The Diffusion of Mexican Immigrants during the 1990s:
Explanations and Impacts. NBER Working Paper Series, No. 11552.
Katz, L. and K. Murphy. 1992. Changes in Relative Wages 1963-1987: Supply and Demand Factors.
Quarterly Journal of Economics 107(1): 35-78.
Lewis, E. G. 2003. Local, Open Economies Within the U.S.: How Do Industries Respond to
Immigration?. SSRN eLibrary.
Martin, Philip. 2007. The Economic Contribution of Migrant Workers to Thailand: Towards Policy
Development. Bangkok: International Labor Organization.
Ottaviano, G. I. and G. Peri. 2006. Rethinking the Effect of Immigration on Wages. NBER Working
Paper Series, No. 12496, Cambridge Ma.
Ottaviano, G. I. and G. Peri. 2008. Immigration and National Wages: Clarifying the Theory and the
Empirics. NBER Working Paper Series. No. 14188.
Sussangkarn, C. 1996. Macroeconomic Impacts of Migrant Workers: Analyses with a CGE Model.
TDRI Quarterly Review. 11(3): 3-11.
133
Data Appendix
Table A1: Summary Statistics
Observations Mean Standard

Deviation
Average weekly hours - Thai men 27,165 47.04 12.52
Average weekly hours - Thai women 23,914 46.19 12.11
Average weekly hours - Thai employed workers

Lower Primary 1,709 45.00 13.84
Upper Primary 18,305 47.68 13.33
High School 14,697 48.71 12.03
College 16,368 42.93 10.00
Average weekly hours - Foreign men 269 53.60 11.96

Average weekly hours - Foreign women 204 52.99 12.21
Average weekly hours - Foreign employed workers

Lower Primary 317 53.54 12.24
Upper Primary 131 51.20 13.79
High School 25 55.34 7.03
Average age - Thai workers (Employed and

Unemployed):
Lower Primary 1,742 42.17 11.74
Upper Primary 18,524 39.12 10.70
High School 15,027 30.60 9.73
College 16,650 35.02 9.87
Average age - Foreign workers (All foreign workers

were employed):
Lower Primary 317 28.84 7.31
Upper Primary 131 31.42 11.13
High School 25 23.96 4.10
Distribution of Actual Weekly Hours Supplied: Percent
Thai
Lower Primary 3.08%
Upper Primary 38.52%
High School 33.26%
College 24.25%
Foreign
Lower Primary 0.61%
Upper Primary 0.16%
High School 0.12%
Total 100.00%
134
Table A2: Industry Distribution of Weekly Hours Supplied by Nativity-Schooling Groups
Lower Lower Upper Upper High
Total Hours Primary Primary Primary Primary School High
Supply Foreign Thai Foreign Thai Foreign School Thai College
Agribusiness 100% 1.11% 10.60% 0.15% 67.40% 0.04% 18.54% 2.16%
Mining and Quarrying 100% 0.00% 3.48% 0.00% 51.42% 0.00% 31.81% 13.29%
Manufacturing 100% 0.84% 1.51% 0.14% 36.52% 0.02% 45.53% 15.43%
Electricity, Gas and Water Supply 100% 0.00% 0.00% 0.00% 21.90% 0.00% 37.36% 40.73%
Construction 100% 0.57% 4.86% 0.13% 64.22% 0.00% 23.64% 6.58%
Wholesale and Reatil Trade 100% 0.53% 2.11% 0.38% 31.39% 0.44% 41.48% 23.66%
Hotels and Restaurants 100% 0.71% 2.53% 0.30% 40.69% 0.94% 40.90% 13.92%
Transport, Storage and Communication 100% 0.26% 2.15% 0.09% 29.36% 0.26% 33.89% 33.99%
Financial Intermediation 100% 0.00% 0.04% 0.00% 5.71% 0.00% 14.42% 79.83%
Property 100% 0.02% 2.44% 0.05% 34.82% 0.32% 28.72% 33.64%
Public Administration and Defence 100% 0.00% 0.17% 0.00% 13.59% 0.00% 29.68% 56.56%
Education 100% 0.09% 0.67% 0.00% 8.32% 0.00% 8.49% 82.43%
Health and Social Work 100% 0.31% 0.85% 0.01% 10.64% 0.02% 30.92% 57.25%
Other Services 100% 0.87% 4.60% 0.60% 55.34% 0.14% 29.68% 8.77%
135
Chapter 6: The Impact of Foreign Labor on Labor Productivity and
Wages in Malaysian Manufacturing, 2000-2006
THAM SIEW YEAN and LIEW CHEI SIANG

Institute of Malaysian and International Studies
Faculty of Economics and Management
Universiti Kebangsaan Malaysia
43600 Bangi, Selangor, Malaysia
E-mail: tham.siewyean@gmail.com; csliew@yahoo.com
ABSTRACT: The increasing presence of foreign workers in Malaysia and its impact on the
Malaysian economy is a widely debated issue. This study seeks to add empirical evidence to the
debate by examining the impact of foreign labor on labor productivity and wages in Malaysian
manufacturing, using establishment level data from 2000-2006. The findings show that the use of
foreign workers has a negative impact on labor productivity and total wages and salaries for all
workers (both native and foreign). However, the negative impact of foreign workers on labor
productivity and total wages and salaries is small and not very robust across different
occupational groups. Furthermore, the negative impact on labor productivity is smaller than the
negative impact on total wages and salaries for all workers. This suggests that the use of foreign
workers has helped improve the competitiveness and profitability of manufacturing firms in
Malaysia by reducing unit labor costs.
1. INTRODUCTION
The increasing presence of foreign workers in Malaysia and its impact on the Malaysian
economy is a widely debated issue. For example, the most recent Central Bank Annual Report
(2010) found that the ready availability of cheap, low skilled foreign labor in Malaysia has
hindered the adoption of automation in the manufacturing sector (Central Bank 2010:22). This
finding suggests that the use of foreign labor has reduced both the productivity and
competitiveness of manufacturing in Malaysia.
This paper seeks to add empirical evidence to the debate regarding foreign labor in
Malaysia by examining the impact of such labor on labor productivity and wages in Malaysian
manufacturing during the period, 2000-2006. This is a new and needed study, because in the past
the impact of foreign labor on manufacturing in Malaysia has not been systematically analyzed.
One of the earliest studies on the economic impact of foreign labor in East Asia as a
whole was done by Athukorala and Manning (1999), who compared structural change and
international migration in East Asia. They found that relatively few unskilled foreign workers
were employed in the key industrial sectors that were targeted for industrial upgrading by East
Asian governments before the Asian Financial Crisis in 1997. Subsequently, Evelyn and Chan
136
(2007: 10) and Evelyn (2009: 20) found that the presence of unskilled foreign workers in
Malaysia did not have a significant effect on skill upgrading or wage inequality in the
manufacturing sector. More recently, Tham and Liew (2004) examined the impact of foreign
labor on average labor productivity of the manufacturing sector in Malaysia, using industry level
data from 1988 - 1996. In this present study, the same issue is examined using establishment level
data from Malaysia for the period 2000-2006.
This paper is divided into five sections. Section 2 presents an overview of foreign
workers in the Malaysian economy, focusing on their role in the manufacturing sector. Section 3
presents the econometric model and the data used in the analysis. Regression results are presented
and discussed in Section 4. The final section, Section 5, summarizes findings and presents policy
conclusions.
2. OVERVIEW OF FOREIGN LABOR IN THE MALAYSIAN ECONOMY
The total number of foreign workers in Malaysia increased from 409,660 in 1999 to 1.9
million in 2009 (Table 1). This figure excludes the large number of illegal foreign workers in
Malaysia, which has been estimated at about one third of the total number of legally recruited
foreign workers (Azizah 2010: 5).
Most foreign workers in Malaysia work in the manufacturing sector. According to Table
1, the share of total foreign workers employed in manufacturing has always been higher than that
of any other sector of the economy. In 2009 about 35 percent of all foreign workers were
employed in manufacturing.
Table 2 summarizes the available data on manufacturing establishments (both Malaysian

and non-Malaysian firms) in Malaysia. Between 2001 and 2006 the total number of
establishments that responded to the annual survey of manufacturing firms fell from 13,934 to
10,326. The reasons for this decline are unclear.
The number of manufacturing firms included in this study is based on a sub-set of the
firms shown in Table 2. The number of firms included here had to be reduced for three reasons.
First, we had to exclude some establishments because of a lack of data on the Producer Price
Index (PPI). Out of 192 industries in the 5 digit codes, data on Producer Price Index (PPI) is only
available for 146 industries. Second, establishments that are classified into a combination of two
to five industry codes are also excluded because of the unavailability of PPI for these industries.
Third, the availability of data on the concentration ratio for industries is used to filter out the final
number of establishments in the analysis. Overall, we are left with 85% of the total number of
manufacturing establishments shown in Table 2.
Table 3 shows the percentage of full time foreign workers employed by Malaysian and
non-Malaysian firms over the period 2000 to 2006. During this time period the share of foreign
workers employed in Malaysian firms increased from 15.0 to 26.7 percent, and the share of
foreign workers in non-Malaysian firms increased from 15.2 to 20.5 percent. According to the
data, the sub-sectors that use more foreign workers are the older industries, including: industries
17 (textiles), 18 (wearing apparel), 19 (footwear, tanning and dressing of leather) and 20 (wood
products).
137
Table 1: Number of Foreign Workers in Malaysia by Sector, 1999-2009
Year Maid Manufacturing Plantation Construction Services Agriculture Total

94,192 155,277 74,501 49,080 36,610 409,660
1999 n.a.
(23.0%) (37.9%) (18.2%) (12.0%) (8.9%) (100%)
177,546 307,167 200,474 68,226 53,683 807,096
2000 n.a.
(22.0%) (38.1%) (24.8%) (8.5%) (6.7%) (100%)
194,710 312,528 222,886 63,342 56,363 849,829
2001 n.a.
(22.9%) (36.8%) (26.2%) (7.5%) (6.6%) (100%)
232,282 323,299 298,325 149,342 64,281 1,057,156
2002 n.a.
(21.8%) (30.3%) (27.9%) (14.0%) (6.0%) (100%)
263,465 385,478 350,351 252,516 85,170 1,239,862
2003 n.a.
(19.7%) (28.8%) (26.2%) (18.9%) (6.4%) (100%)
285,441 475,942 384,473 231,184 93,050 1,470,090
2004 n.a.
(19.4%) (32.4%) (26.2%) (15.7%) (6.3%) (100%)
320,171 581,379 472,246 281,780 159,662 1,815,238
2005 n.a.
(17.6%) (32.0%) (26.0%) (15.5%) (8.8%) (100%)
310,662 646,412 354,124 267,809 166,829 123,373 1,869,209
2006
(16.6%) (34.6%) (18.9%) (14.3%) (8.9%) (6.6%) (100%)
314,295 733,372 337,503 293,509 200,428 165,698 2,044,805
2007
(15.4%) (35.9%) (16.5%) (14.4%) (9.8%) (8.1%) (100%)
293,359 728,867 333,900 306,873 212,630 186,967 2,062,596
2008
(14.2%) (35.3%) (16.2%) (14.9%) (10.3%) (9.1%) (100%)
251,355 663,667 318,250 299,575 203,639 181,660 1,918,146
2009
(13.1%) (34.6%) (16.6%) (15.6%) (10.6%) (9.5%) (100%)
Note: n.a. – not available Source: Ministry of Home Affairs
Table 2: Distribution of Selected Manufacturing Establishments in Malaysia
Total Establishments Selected Establishments

Non- Non-Malaysian
Year Malaysian Malaysian firm Overall
Malaysian Overall firm
firm
firm Total % Total % Total %
2000 18759 1696 20455 16253 86.6 1268 74.8 17521 85.7
2001 12471 1463 13934 10847 87.0 1127 77.0 11974 85.9
2002 12002 1480 13482 10354 86.3 1149 77.6 11503 85.3
2003 12212 1460 13672 10427 85.4 1076 73.7 11503 84.1
2004 11048 1403 12451 9341 84.5 990 70.6 10331 83.0
2005 26655 1602 28257 23152 86.9 1206 75.3 24358 86.2
2006 9193 1133 10326 7797 84.8 825 72.8 8622 83.5
Overall 102340 10237 112577 88171 86.2 7641 74.6 95812 85.1
Source: Department of Statistics (DOS)
138
Table 3: Percentage of Full Time Foreign Workers in All Occupation Groups in Manufacturing
2000 2001 2002 2003
Non-Malaysian Non-Malaysian Non-Malaysian Non-Malaysian
MSIC Malaysian firm Malaysian firm Malaysian firm Malaysian firm
firm firm firm firm
No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW
15 11450 10.4 937 9.4 11720 12.1 705 7.0 12787 12.3 522 8.6 15523 14.6 738 8.8
16 145 2.0 37 7.2 140 2.0 - - 166 2.2 - - 134 2.0 48 7.8
17 3651 18.4 6552 34.7 3843 20.1 5438 31.4 4325 22.1 6902 33.0 4783 25.1 5746 36.7
18 11117 17.2 3580 20.7 11239 25.1 4745 21.5 12679 27.7 9828 31.2 14081 30.1 6469 32.7
19 498 7.8 659 40.3 483 10.6 477 39.7 537 10.4 264 26.2 700 15.3 287 31.7
20 35468 32.8 8952 57.1 33976 35.0 7809 55.6 34681 37.6 9334 64.6 37732 40.8 505 21.5
21 2854 10.6 853 22.1 2902 12.6 399 8.4 3621 13.2 251 6.8 3943 15.1 325 11.0
22 247 2.8 - - 265 3.6 12 3.4 369 4.3 11 2.8 339 9.9 19 4.0
23 100 1.5 58 5.9 73 2.2 90 4.5 122 3.3 86 4.6 193 5.0 38 4.8
24 1441 5.1 486 4.5 1628 6.4 480 3.4 1529 6.0 439 2.7 2087 7.4 471 3.1
25 19562 17.3 5209 13.0 21575 20.3 5508 13.0 23014 21.0 6319 12.1 27722 25.5 7807 14.1
26 5167 11.3 894 9.2 5072 11.8 1629 13.9 5589 13.3 1793 15.0 6461 15.5 1521 13.9
27 2204 8.2 1287 13.8 2209 9.2 1575 17.4 2638 10.3 1759 17.2 3106 12.0 1680 15.3
28 6113 10.3 1889 12.3 6281 14.3 1391 10.3 7584 14.9 2040 13.6 8423 16.6 2309 13.6
29 915 7.5 213 5.1 1237 9.4 510 11.0 1589 11.6 295 11.6 2404 16.9 270 20.9
30 1728 15.1 - - 1981 21.8 - - 1079 19.2 7949 14.7 - - - -
31 2405 7.5 4469 13.2 2617 11.9 4028 12.3 4143 13.4 3096 11.0 3627 14.1 4636 13.2
32 10003 10.9 23494 11.1 6475 11.7 19121 9.8 7232 9.8 18683 9.1 9418 11.3 21584 10.1
33 1020 20.6 1625 10.9 187 17.8 1308 7.2 - - 912 6.2 - - 1018 6.4
34 1140 4.0 - - 2364 6.8 - - 2488 5.9 - - 3014 7.3 - -
35 480 10.9 - - 292 7.7 - - 568 13.9 - - 539 12.5 - -
36 16307 21.3 5183 27.6 16493 25.7 5327 29.0 23551 32.4 6019 32.7 27117 35.7 4123 34.3
Overall 134015 15.0 66377 15.2 133050 17.8 60553 14.0 150291 18.5 76502 15.1 171347 21.2 59592 13.6
139
2004 2005 2006
Non-Malaysian Non-Malaysian
MSIC Malaysian firm Malaysian firm Malaysian firm Non-Malaysian firm
firm firm
No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW No. FW % FW
15 17641 16.4 1052 12.1 21838 17.7 862 11.4 22190 17.3 505 10.1
16 89 1.3 77 9.8 46 1.0 79 7.8 102 3.1 55 18.1
17 4903 27.8 5808 38.7 5590 29.6 5922 41.3 6562 37.5 6095 42.3
18 12632 31.8 16571 52.4 11991 28.7 7013 43.6 15599 33.4 - -
19 940 16.1 245 26.0 1054 16.1 137 25.8 1033 19.0 - -
20 42138 44.6 799 23.3 52638 48.1 10393 69.5 52158 47.7 11158 67.6
21 4564 16.3 761 21.8 5436 17.4 639 20.6 4931 17.0 964 20.0
22 336 10.2 18 3.5 589 4.8 - - 718 4.7 26 8.1
23 97 2.7 34 4.4 - - 28 3.5 19 0.5 30 3.4
24 2476 8.4 513 3.7 3900 10.1 1099 8.0 4996 11.9 1097 8.1
25 22636 24.4 10314 17.3 40742 30.2 12954 20.2 48550 34.4 16987 25.3
26 6547 15.9 2163 21.8 8785 17.8 1603 13.5 10376 20.8 1291 12.0
27 4455 15.8 604 16.5 5062 15.6 1473 15.1 7869 21.4 1250 29.4
28 9661 17.6 3030 18.8 11995 19.3 3632 21.4 15633 24.1 5800 31.1
29 2934 17.0 921 10.3 2976 23.4 1806 11.4 3665 25.3 1107 12.8
30 - - - - 5855 36.8 - - - - 17760 25.0
31 3840 14.7 4883 14.0 4629 16.0 5861 20.2 4915 17.2 5032 14.6
32 8170 10.7 27335 13.1 16667 14.2 25502 14.8 14505 16.7 27076 13.2
33 488 31.5 644 5.5 153 5.9 826 8.0 378 11.8 2952 12.4
34 3292 7.7 782 13.7 3769 8.6 - - 4990 10.4 1224 20.0
35 567 12.8 26 8.6 842 14.2 30 9.1 776 12.7 - -
36 29614 38.2 6092 34.7 38468 42.2 7213 38.3 40852 43.2 6126 45.0
Overall 178017 22.3 82672 18.2 243025 24.7 87072 20.6 260817 26.7 106535 20.5
Note: - Some firms are excluded from analysis after the screening process.Source: DOS
140
3. Model Specification and Estimation Method
To examine the factors affecting firm productivity, the Cobb-Douglas production
function is specified as:
uijt
Yijt  Aijt ( Lijt )1 ( K ijt ) 2 e (1)
where Yijt represents output (value added); Lijt , K ijt , and Aijt denote the number of workers,
capital stock, and total factor productivity (TFP); u ijt is a random disturbance term. The
subscripts i, j and t refer to the ith firm in the jth industry at time t.
Dividing Eqn.(1) by Lijt ,
Yijt  ( K ijt ) 2  uijt

= Aijt  1
e
Lijt  ( Lijt ) 1 
11
 K ijt  uijt
= Aijt   ( K ijt ) 1 2 1 e (2)
 Lijt 
 
Take natural logarithm (ln) in Eqn.(2)
 Yijt   K ijt 
ln  = ln Aijt   1 ln    2 ln K ijt  u ijt (3)
L  L 
 ijt   ijt 
Yijt K ijt
where LPijt  is the value added per worker or labor productivity, is the capital-labor
Lijt Lijt
ratio, and TFPijt  ln Aijt . The capital stock ( K ijt ) is included on an additional variable in order
to relax the constant return to scale assumption (Kohpaiboon 2006). The simple correlation
coefficient between capital stock and capital-labor ratio is in excess of 0.8, suggesting that
multicollinearity might be a problem1. However, this variable is retained in the regression
because multicollinearity does not seem to affect the overall estimation results, as presented in
Section 4.
Technology spillover from foreign direct investment (FDI) is expected to take place when
the presence of a foreign firm generates productivity or efficiency benefits for the host country’s
local firms (Blomstrom and Kokko 1998). Multi-national corporations (MNCs) normally possess
proprietary assets – including intangible assets related to production, management and marketing
-- that non-MNCs lack (Blomstrom et.al 2000; Caves 1996; Dunning 1993). The possession and
transfer of such assets by MNCs enable their foreign affiliates to compete with local firms in
overseas markets even when local firms have better knowledge of those markets. This suggests
that foreign-owned firms may have higher technology levels, and thus higher productivity, than
1
As a rule of thumb, when the simple correlation coefficient between two regressors is high, say, in excess
of 0.8, then multicollinearity is a serious problem. However, this rule is a sufficient but not a necessary
condition for the existence of multicollinearity. Moreover, multicollinearity may not pose a serious problem
when the coefficient of determination, R2 is high and most of the regression coefficients are individually
significant (Gujarati 2009: 338, 347).
141
locally owned firms (Takii 2004). The technology spillover from FDI is not automatic but rather
conditioned on the nature of the trade policy regime across industries. Therefore, the level of
technology represented by TFPijt is influenced by the level of foreign presence and the nature of
the trade policy regime in the host country. However, in the Malaysian context, a proxy for trade
policy is not available. The foreign presence is proxied by a dummy variable to capture
ownership. Hence, we obtain
 K ijt 
ln LPijt =  0  1 ln    2 ln K ijt   3 DTijt   X ijt  u ijt (4)
 Lijt 
 
where DTijt = 1 if foreign establishments, 0 otherwise and X ijt is a set of explanatory variables
containing firm-specific and industry-specific factors.
Thirty (30) additional explanatory variables are used.

i) Labor quality. This is usually proxied by the ratio of supervisory and management
workers to total employment. However, in this study, we use the share of foreign workers
in six (6) occupational grouping to capture its impact on labor productivity.
ii) Dummy variable for date of establishment - This dummy variable is used to examine the
labor productivity differential between the firms established before and after the 1997
Asian financial crisis.
iii) Market concentration- Market concentration is included as an explanatory variable

because two industries with the same technical efficiency may show a different value
added per worker because of different domestic market concentrations (Kohpaiboon
2006). In addition as argued by Hall (1988), the impact of any possible exogenous factors
on industry productivity would be conditioned by the degree of market concentration.
This is proxied by the sum of market share of the four largest firms (CR4) in the
individual industry.
iv) Dummy variables for industry at two-digit level - A total of 21 dummy variables are used
to examine the labor productivity differential between industries.
v) Dummy variable for firms’ competitiveness - This dummy variable is used to examine
the labor productivity differential between the firms that reported losses and firms that
exhibited profits. The latter are considered to be competitive while the former are not.
Based on the above discussion, the balanced panel data regression model is specified as
follows:
 K ijt 
ln LPijt =  0  1 ln    2 ln K ijt   3 DTijt   4 QM ijt   5 QTijt

L
 ijt 
  6 QC ijt   7 QE ijt   8 QDijt   9 QPijt   10 CR jt
  11 DYijt   12 DC ijt   DI ijt   i   j  u ijt (5)
where LPijt = real value added per worker
142
K ijt Lijt = capital-labor ratio
K ijt = real value of fixed assets
DTi = dummy variable: 1 if foreign establishments, 0 otherwise
QM ijt = share of foreign workers in managers, professionals &
executives
QTijt = share of foreign workers in technicians & associate professional
QC ijt = share of foreign workers in clerical & related occupations
QEijt = share of foreign workers in elementary occupations
QD ijt = share of foreign workers in plant & machine operators and
assemblers (directly employed)
QPijt = share of foreign workers in plant & machine operators and
assemblers (employed through labor contracts)
CR jt = concentration ratio
DYijt = dummy variable: 1 if firm established after 1997, 0 otherwise
DC ijt = dummy variable: 1 if firm encountered losses, 0 otherwise
DI ijt = a set of two-digit industry code dummies with manufacture of
food and beverages (industry 15) is taken as reference group
i = unobserved firm specific effect
j = unobserved industry specific effect
u ijt = remainder stochastic disturbance
Regression analysis with these independent variables is done on the log of unit labor cost
(ULC) and the log of real total wages & salaries for all workers (both native and foreign workers,
WG). In addition to using the share of foreign workers in six occupational groupings, further
regression analysis is carried out by using the overall share of foreign workers (FL) as one of the
explanatory variables in each of the above models (see Equations 2 in Tables 6, 7 and 8).
Unit labor cost is defined as the ratio of real total wages & salaries for all workers to real
value added. It is a proxy for total factor productivity, which is not available and not readily
computed using the existing data set. The Producer Price Index (2000=100) for local production
is used to deflate the nominal value of value-added, and value of fixed assets while the Consumer
Price Index (2000=100) is used to deflate the nominal value of total wages & salaries for all
workers. The share of foreign workers in each occupational group is expressed in terms of
percentage.
Based on Malaysia Standard Industrial Classification (MSIC), the Malaysian

manufacturing sector consists of 192 industries at five-digit level (or 22 industries at two-digit
level). The establishment level data employed for this study are obtained from the Department of
Statistics, Malaysia (DOS) from its annual survey (2000 to 2004 and 2006) and census (2005) of
the manufacturing sector. After taking into account the availability of Producer Price Index and
concentration ratio at the industry level, the panel data used in this study consists of 2,341 firms
(matched by firm code) from the year 2000 to 2006, with a total of 16,387 observations. Panel
data regression is used in this study as the total numbers of firms have substantially changed over
the time period of the study (2000 to 2006).
143
The period fixed-effects specification with time dummies is used to estimate the model as
the productivity, the unit labor cost and the total wages & salaries for all workers function may
shift over time because of factors such as technological changes and other external effects. The
results of hypothesis testing show that the period fixed-effects are found to be statistically
significant.
The models are estimated by Panel Generalized Least Squares method with period
weights or period heteroskedasticity (allows for a different residual variance for each period)
using EViews, version 6. As far as the estimation method is concerned, the estimators are robust
as the computation of robust estimators is necessary when the models are estimated by standard
OLS without cross-section or period effects (EViews User’s Guide II 506).
One of the key issues when estimating the impact of foreign workers on the performance
of firms (productivity, unit labor cost, total wages & salaries for all workers) is the endogeneity of
the share of foreign workers in each firm. It is possible that the share of foreign workers is not an
exogenous variable due to selectivity based on past performance or characteristics of the firms
(Pholphirul et al. 2010). In order to overcome the endogeneity problem, we perform our
regression analysis by using the first lag of the share of foreign workers as an instrumental
variable (IV). Our choice of this IV variable is based on the expectation that the share of foreign
workers for the previous year in the same firm can affect the share of foreign workers in the
current year through past performance. But we would not expect that the share of foreign workers
for the previous year to affect the firms’ performance in the current year beyond its effects
through the share in the current year. Generalized Method of Moments (GMM) is used to
estimate the specified models with the inclusion of a first-order autoregressive error term, AR(1)
into the models.
3. ESTIMATION RESULTS
Descriptive statistics of variables
Table 4 summarizes the descriptive statistics of the variables used in the regression
analysis. The negative values for some of the variables in natural logarithm (ln) form indicate that
the actual values are less than one. In some of the occupational groups the percentage of foreign
workers is small, less than 5 percent. Aggregating the number of foreign workers over all
occupational groups, the share of foreign workers in the total sample is positively distributed with
a mean of 15.6% and a standard deviation of 21.8%.
Estimation results for overall manufacturing sector
The results of the first-stage instrumental variables regression are shown in Table 5. In
this table various combinations of the instrumental variables are shown in conjunction with all the
exogenous variables included in equations (1) and (2). As expected, the instrumental variables –
the first lags of the share of foreign workers in each occupational group – are positive and highly
significant in each equation. This suggests that these instrumental variables are not subject to
weak instrument concerns. In Table 5 the F-statistics of the excluded instruments are all quite
large, suggesting instrument relevance.
Table 6 shows the second-stage IV estimation results for equation 1 (with the share of
foreign workers in each occupational group) and equation 2 (with overall share of foreign
workers). Results suggest that the capital-labor ratio and value of fixed assets are significantly
144
Table 4: Summary Statistics for Variables
Variable Mean Median Maximum Minimum Std. Dev. Skewness

real value added per worker, lnLP 10.35 10.35 14.91 3.65 0.98 -0.22
log of unit labor cost, lnULC -0.85 -0.79 4.95 -9.91 0.66 -0.18
log of real total wages & salaries for all workers (both native
13.90 14.18 19.54 -0.08 2.04 -0.69
and foreign), lnWG
log of capital-labor ratio, lnK/L 10.29 10.65 15.67 -5.76 2.06 -2.51
log of real value of fixed assets, lnK 14.69 15.37 23.14 -0.34 3.15 -1.50
dummy variable for type of establishment, DT 0.14 - 1.00 0.00 - -
share of foreign workers in managers, professionals &
4.24 0.00 100.00 0.00 13.01 4.33
executives, QM
share of foreign workers in technicians & associate
1.81 0.00 100.00 0.00 8.80 7.07
professional, QT
share of foreign workers in clerical & related occupations, QC 0.48 0.00 100.00 0.00 4.74 15.09
share of foreign workers in elementary occupations, QE 2.28 0.00 100.00 0.00 11.02 5.99
share of foreign workers in plant & machine operators and
18.22 0.00 100.00 0.00 27.44 1.42
assemblers (directly employed), QD
share of foreign workers in plant & machine operators and
6.55 0.00 100.00 0.00 23.16 3.49
assemblers (employed through labor contracts), QP
concerntration ratio, CR 0.42 0.39 7.85 0.07 0.25 6.59
dummy variable for date of establishment, DY 0.06 - 1.00 0.00 - -
dummy variable for competitiveness, DC 0.18 - 1.00 0.00 - -
overall share of foreign workers, FL 15.56 2.32 100.00 0.00 21.77 1.49
Source: DOS
145
Table 5: First-stage IV Estimates for the Share of Foreign Workers
Equation 1 Equation 2
share of share of
foreign foreign
share of share of share of
share of workers in workers in
foreign foreign foreign
foreign plant & plant &
workers in workers in workers in overall share
Independent Variables workers in machine machine
managers, technicians & clerical & of foreign
elementary operators and operators and
professionals associate related workers, FL
occupations, assemblers assemblers
& executives, professional, occupations,
QE (directly (employed
QM QT QC
employed), through labor
QD contracts), QP
Instruments
first lag of the share of foreign
workers in managers, professionals 0.754*** - - - - - -
& executives, QM t 1
workers in technicians & associate - 0.777*** - - - - -
professional, QTt 1
workers in clerical & related - - 0.787*** - - - -
occupations, QCt 1
workers in elementary occupations, - - - 0.633*** - - -
QEt 1
workers in plant & machine
operators and assemblers (directly - - - - 0.829*** - -
employed), QDt 1
workers in plant & machine - - - - - 0.624*** -
operators and assemblers (employed
146
Equation 1 Equation 2
share of share of
foreign foreign
share of share of share of
share of workers in workers in
foreign foreign foreign
foreign plant & plant &
workers in workers in workers in overall share
Independent Variables workers in machine machine
managers, technicians & clerical & of foreign
elementary operators and operators and
professionals associate related workers, FL
occupations, assemblers assemblers
& executives, professional, occupations,
QE (directly (employed
QM QT QC
employed), through labor
QD contracts), QP
through labor contracts), QPt 1
first lag of the overall share of
- - - - - - 0.896***
foreign workers, FLt 1
Included exogenous variables
log of capital-labor ratio, ln(K/L) -0.203*** -0.178*** -0.012 -0.443*** -1.069*** -1.123*** -0.809***
log of real value of fixed assets, lnK 0.191*** 0.171*** 0.019 0.442*** 1.133*** 1.037*** 0.820***
dummy variable for type of
2.515*** -0.206 0.020 -0.604** -1.036** -1.408*** -0.711***
establishment, DT
concerntration ratio, CR 0.287 -0.409 -0.278* -1.160*** -1.114 -1.325 -0.887*
dummy variable for date of
-0.024 0.221 0.038 0.014 -0.249 0.082 0.152
establishment, DY
dummy variable for
0.292* 0.115 0.024 -0.433** 0.221 -0.993** -0.031
competitiveness, DC
**
Constant -0.633 -0.101 0.035 0.163 -1.796 -0.390 -1.287**
Dummy variables for industry, DI YES YES YES YES YES YES YES
Observations 14046 14046 14046 14046 14046 14046 14046
R-squared 0.664 0.588 0.566 0.417 0.697 0.399 0.809
F-statistics excluded instruments 987.68 713.64 653.27 357.72 1149.68 332.10 2120.90
P-value 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Note: * significant at 10%, ** significant at 5%, *** significant at 1%
Source: Regression results
147
Table 6: IV Estimation Results of the Labor Productivity Model
Independent Variables Eqn. 1 Eqn. 2

0.137 0.169
log of capital-labor ratio, ln(K/L)
[0.035]*** [0.027]***
0.105 0.066
log of real value of fixed assets, lnK
[0.027]*** [0.019]***
-0.151 0.127
dummy variable for type of establishment, DT
[0.142] [0.069]*
share of foreign workers in managers, professionals & 0.008
-
executives, QM [0.005]*
share of foreign workers in technicians & associate professional, -0.009
-
QT [0.011]
-0.028
share of foreign workers in clerical & related occupations, QC -
[0.015]*
-0.002
share of foreign workers in elementary occupations, QE -
[0.003]
share of foreign workers in plant & machine operators and -0.005
-
assemblers (directly employed), QD [0.003]*
-
assemblers (employed through labor contracts), QP [0.001]
0.627 -0.365
concerntration ratio, CR
[1.427] [0.487]
0.067 -0.339
dummy variable for date of establishment, DY
[0.376] [0.448]
-0.169 -0.199
dummy variable for competitiveness, DC
[0.036]*** [0.019]***
0.806 0.799
AR(1)
[0.010]*** [0.003]***
-0.0057
overall share of foreign workers, FL -
[0.0009]***
7.199 7.908
Constant
[0.816]*** [0.304]***
Dummy variables for industry, DI YES YES
Observations 14046 14046
R-squared 0.8526 0.8776
Note: Standard error in brackets [ ] * significant at 10%, ** significant at 5%, *** significant at 1%
Equation (1) is estimated by using the first lag of the share of foreign workers in each occupational group as the
instrumental variables
Equation (2) is estimated by using the first lag of the share of foreign workers as the instrumental variable
148
related to labor productivity. Ceteris paribus, a one percent increase in capital-labor ratio and value of
fixed assets, on average, will raise labor productivity by 0.14% and 0.11% (Eqn.1); 0.17% and 0.07%
(Eqn.2) respectively.
Analysis by occupational groups indicate that a one percentage point increase in the share of
foreign workers in each occupational grouping (except for the managers, professionals & executives
group) will lead to a decrease in average labor productivity between 0.2% and 2.8% 71. The impact of
the share of foreign workers is only significant in the directly employed plant & machine operators
group and the clerical & related occupations group while the impact of foreign workers in the
managers, professionals and executives group is positive and significant.
The estimated coefficient for the overall share of foreign workers (FL) in equation 2 indicates
a significant negative relationship with labor productivity. Holding other factors constant, a one
percentage point increase in the overall share of foreign workers, the average labor productivity will
fall by 0.57%. In order to calibrate the total effect of the large increase in share of foreign workers,
when the share of foreign workers increases by 10 percentage point (for example, from 10% to 20%),
then average labor productivity will decrease from RM 50 thousand to RM 47.15 thousand, which is
equivalent to 5.7%.
The results of the estimation on all the unit labor cost functions in Table 7 show that the
capital-labor ratio and the value of fixed assets are significantly related to unit labor costs. Ceteris
paribus, a one percent increase in value of fixed assets on average will increase unit labor cost by
0.12% (Eqn.1) and 0.06% (Eqn.2). As expected, unit labor cost is negatively related to capital-labor
ratio, where a one percent increase in capital-labor ratio on average will decrease the unit labor cost
by 0.27% (Eqn.1) and 0.21% (Eqn.2).
The differences in unit labor cost between the two types of establishments and the firms’
competitiveness are also statistically significant. The median of the unit labor cost of foreign
establishments is 26.7% (Eqn.1) and 18.0% (Eqn.2) lower as compared to local establishments. On
the other hand, the median of the unit labor cost of establishments that exhibited losses is 30.5%
(Eqn.1) and 16.9% (Eqn.2) higher than establishments that reported profits.
A one percentage point increase in the share of foreign workers in plant & machine operators
and assemblers (directly employed) group will lead to a marginal increase (0.4%) of average unit
labor cost. Besides this occupational group, the impact on unit labor cost is also found to be
significant in the case of technicians & associate professional group. The estimated coefficient for the
overall share of foreign workers in Eqn.2 indicates a significant negative relationship with unit labor
cost, but the magnitude is rather small (0.19%).
The results of the estimation on all the total wages and salaries for all workers (both native
and foreign) functions show that the capital-labor ratio and the value of fixed assets are significantly
related to total wages and salaries for all workers (Table 8). Ceteris paribus, a one percent increase in
value of fixed assets on average will increase the total wages and salaries for all workers by 1.24%
(Eqn.1) and 1.14% (Eqn.2). On the other hand, the total wages and salaries for all workers are
negatively related to capital-labor ratio, where a one percent increase in capital-labor ratio on average
will decrease the total wages and salaries for all workers by 1.18% (Eqn.1) and 1.06% (Eqn.2).
71
In the log-linear model, ln Y = 1 + 2 X, a one-unit increase in X leads to (approximately) a 100 x 2 %
change in Y (Hill et al. 2008: 87)
149
Table 7: IV Estimation Results of the Unit Labor Cost Model

-0.272 -0.213
[0.025]*** [0.020]***
0.118 0.057
[0.017]*** [0.012]***
-0.311 -0.199
[0.108]*** [0.047]***
share of foreign workers in managers, professionals & 0.009
-
executives, QM [0.007]
share of foreign workers in technicians & associate professional, -0.052
-
QT [0.024]**
0.007
[0.009]
-0.018
[0.012]
share of foreign workers in plant & machine operators and 0.004
-
assemblers (directly employed), QD [0.001]***
-
assemblers (employed through labor contracts), QP [0.006]
1.364 -0.580
[0.256]*** [0.215]***
-0.069 -0.717
[0.081] [0.367]*
0.266 0.156
[0.034]*** [0.017]***
0.717 0.754
AR(1)
[0.006]*** [0.005]***
-0.0019
[0.0007]***
-0.345 0.869
Constant
[0.149]** [0.140]***
R-squared 0.5563 0.7091
150
Table 8: IV Estimation Results of the Total Wages and Salaries for All Workers (both Native
and Foreign) Model
-1.177 -1.056
[0.029]*** [0.018]***
1.236 1.139
[0.021]*** [0.011]***
-0.203 0.045
[0.468] [0.085]
share of foreign workers in managers, professionals & -0.006
-
executives, QM [0.042]
share of foreign workers in technicians & associate professional, 0.105
-
QT [0.147]
-0.206
[0.216]
0.001
[0.006]
-
assemblers (directly employed), QD [0.002]*
-
assemblers (employed through labor contracts), QP [0.006]**
0.535 -0.375
[0.203]*** [0.162]**
-0.907 -0.437
[0.216]*** [0.291]
0.106 0.036
[0.055]* [0.013]***
0.744 0.792
AR(1)
[0.005]*** [0.005]***
-0.0075
[0.0006]***
7.775 8.321
Constant
[0.135]*** [0.091]***
R-squared 0.9706 0.9905
151
Analysis by occupational groups indicate that a one percentage point increase in the share of
foreign workers directly employed in plant & machine operators and assemblers group will cause a
0.3% decline in the average total wages and salaries for all workers. The impact is relatively larger
(1.4%) for the foreign workers employed through labor contracts in the same occupational group.
Holding other factors constant, a one percentage point increase in the overall share of foreign
workers, the average total wages and salaries for all workers will decrease by 0.75%.
Summary of key findings on impact of foreign workers on labor productivity, unit labor costs and
total wages and salaries for all workers
Using the results shown in equation 2, Table 6 – Table 8, we can summarize the impact of
using foreign workers in Malaysia as follows. The use of foreign workers has a small, negative
impact on both labor productivity (Table 6) and on total wages and salaries for all workers (both
native and foreign) (Table 8). But the negative impact on labor productivity (-0.57%) is smaller than
the negative impact on total wages and salaries for all workers (-0.75%). Consequently, the use of
foreign workers can help to enhance the competitiveness of manufacturing firms in the country by
reducing the unit labor cost which is a proxy for competitiveness.
In terms of occupational groupings, most of the results are small and not statistically robust
across different occupational groups. The only occupational group that is statistically significant for
all three estimated equations is the share of workers in plant and machine operators that are directly
employed. For this occupational group, the use of foreign workers reduced labor productivity by -
0.5%, which is more than the fall in total wages and salaries for all workers (-0.3%). For this
occupational group, unit labor costs have increased (0.4%), thereby reducing the competitiveness of
firms.
Discussion of results
In Tables 9 to Table 14, we have separated the analysis of firms’ performance by year, type
of establishment (local or foreign) as well as the employment of foreign workers, namely those firms
which employ foreign workers and those which do not.
Table 9 shows that on an overall basis, establishments with foreign workers are more capital
intensive than establishments without foreign workers. However, when we compare local with
foreign establishments, local establishments with foreign workers have a higher capital labor ratio
than local establishments without foreign workers.
Overall, establishments with foreign workers have more productive workers (Table 10),
higher real total wages and salaries (for both native and foreign workers) per worker (Table 11) and
lower unit labor costs (Table 12). However, when we disaggregate by type of establishment, we find
that these trends are true for local establishments only.
Overall, we can also see that establishments with foreign workers have a higher median of
real profit/loss than establishments without foreign workers (Table 13). This holds true for local
establishments. However, foreign establishments without foreign workers have higher median real
profit/loss, except for the years 2000, 2002 and 2005.
152
On the basis of the above, it would appear that local establishments have used foreign labor to
enhance their competitiveness (as proxied by the unit labor costs), and profits compared to local
establishments without foreign workers.The pattern for foreign establishments is not so Table 9:
Median of Real Capital-Labor Ratio (RM thousand), 2000-2006
Local establishments Foreign establishments All establishments
Year
F NF Overall F NF Overall F NF Overall
2000 54.6 26.0 40.4 67.5 85.6 69.9 58.2 28.5 44.3
2001 53.7 28.1 40.6 64.7 82.8 66.6 56.6 30.6 43.6
2002 51.5 27.7 40.7 63.9 84.8 67.3 54.1 29.9 43.8
2003 50.7 24.1 38.0 58.9 63.3 60.5 52.0 25.8 41.7
2004 49.5 22.7 37.1 58.3 79.1 61.9 51.3 24.5 41.3
2005 48.6 23.0 37.4 57.9 48.6 56.1 50.8 24.4 40.1
2006 46.4 20.2 35.2 51.6 44.1 50.8 47.2 22.0 37.8
Overall 50.5 24.5 38.3 61.2 68.0 62.2 52.3 26.7 42.1
Note: F – employ foreign labor, NF – did not employ foreign labor

Source: DOS
Table 10: Median of Real Labor Productivity (RM thousand), 2000-2006

Year
2000 35.1 21.4 27.9 51.5 67.2 53.1 38.7 22.6 31.0
2001 34.0 24.4 29.1 47.4 54.8 48.3 36.9 25.4 31.8
2002 34.2 24.6 29.5 50.9 71.8 54.0 37.3 26.1 32.7
2003 33.8 24.5 29.2 50.3 70.7 53.1 36.6 25.5 31.4
2004 35.0 23.9 28.9 46.9 76.6 51.4 37.3 25.5 31.4
2005 32.2 22.4 27.6 48.0 56.0 48.2 34.8 23.4 30.2
2006 31.3 22.8 27.2 44.1 56.1 46.1 33.9 24.0 29.1
Overall 33.5 23.4 28.4 48.5 64.2 51.0 36.4 24.6 31.1

Source: DOS
Table 11: Median of Real Total Wages and Salaries (for both Native and Foreign Workers) per
Worker (RM), 2000-2006
Year
2000 14627 10741 12691 18814 23147 19197 15589 11225 13663
2001 14167 11767 13088 18826 19483 18836 15055 12137 13990
2002 14273 12013 13257 18626 23981 19330 15376 12386 14108
2003 14923 11984 13477 19073 27117 19693 15611 12239 14212
2004 14885 12059 13680 19232 23689 20182 15824 12572 14528
2005 14573 11922 13400 20352 21997 20614 15583 12333 14240
2006 14383 11976 13560 20234 23302 20418 15229 12420 14248
Overall 14550 11804 13327 19309 23186 19739 15457 12149 14149

Source: DOS
153
Table 12: Median of Unit Labor Cost, 2000-2006
Year
2000 0.42 0.46 0.44 0.38 0.33 0.36 0.41 0.46 0.43
2001 0.43 0.47 0.45 0.38 0.35 0.38 0.42 0.46 0.44
2002 0.42 0.48 0.45 0.38 0.31 0.37 0.41 0.47 0.43
2003 0.44 0.48 0.46 0.39 0.30 0.38 0.43 0.48 0.45
2004 0.46 0.51 0.48 0.41 0.30 0.39 0.45 0.49 0.46
2005 0.46 0.52 0.49 0.44 0.40 0.43 0.46 0.51 0.48
2006 0.49 0.51 0.49 0.44 0.38 0.43 0.47 0.50 0.49
Overall 0.45 0.49 0.47 0.40 0.35 0.39 0.44 0.48 0.45

Source: DOS
Table 13: Median of Real Profit and Loss (RM thousand), 2000-2006
Year
2000 885.3 44.7 212.1 4070.9 3558.3 3995.8 1328.2 55.0 376.2
2001 695.8 43.1 168.5 2522.5 2793.9 2610.4 1091.7 53.2 298.0
2002 890.3 56.2 211.4 4177.4 3462.9 3911.1 1292.4 65.1 373.1
2003 923.8 60.9 213.4 3887.1 4165.9 3902.4 1255.4 70.1 363.5
2004 943.0 63.8 235.5 4101.5 4794.1 4333.6 1286.1 72.5 392.5
2005 921.3 40.5 204.5 3618.9 3261.5 3379.4 1191.8 46.5 358.3
2006 894.5 95.7 291.7 3263.5 3512.8 3346.9 1226.6 103.6 447.1
Overall 885.2 56.9 221.1 3626.2 3576.1 3615.1 1236.7 65.4 372.7

Source: DOS
Table 14: Median of Investment and Change in the Real Profit and Loss (RM),
2000-2006
Establishments employ Establishments did not
All establishments
foreign labor employ foreign labor
Year Change in Change in Change in
Investment real profit Investment real profit Investment real profit
and loss and loss and loss
2000 n.a. n.a. n.a. n.a. n.a. n.a.
2001 5306.47 -7341.51 -34.43 3260.15 0.00 1887.55
2002 -24752.44 30842.52 -4200.68 4595.73 -5894.64 7490.90
2003 -327668.90 9321.56 -11334.04 2106.72 -62839.96 3224.88
2004 -214481.40 34444.60 -7777.40 3311.56 -40686.37 6698.01
2005 -417131.45 -45144.55 -18313.07 -5046.10 -124997.80 -11218.79
2006 4858.14 52647.79 4979.69 30006.51 4979.69 37342.44
r 0.191*** 0.043*** 0.176***
Note: r is the simple correlation coefficient beween investment and change in real profit and loss
n.a. – not available ; *** significant at 1%
Source: Department of Statistics (DOS)
154
clear as establishments without foreign workers are more competitive due to lower unit labor costs,
but their profits/losses are higher than the establishments with foreign workers for only four out of the
seven years in our data set.
Table 14 (previous page) shows that for all establishments, investment as measured by the
change in real fixed assets, fell progressively from 2001 to 2005, while the change in profits
flucutated over time. This holds true for establishments both with and without foreign workers. There
is, however, a positive relationship between investment and the change in profit/loss as shown by the
small, but positive, simple correlation coefficient in Table 14.
Overall, the trend over time shows the following. First, the number of firms with foreign
workers has increased over time (from 1267 in 2000 to 1410 in 2006), while the number of firms
without foreign workers has declined (from 1074 in 2000 to 931 in 2006). Second, the median of the
real capital labor ratio for establishments with foreign workers has fallen. Similarly, the median value
of real labor productivity has decreased from 2000 to 2006. There are also annual fluctuations in real
total wages and salaries (for both native and foreign workers) per worker over time, but the median
value of this variable has largely fallen for establishments with foreign workers.
Establishments without foreign workers also show some similar patterns: namely the median
of their capital labor ratio has fallen from 2001 to 2006 but the median of their real labor productivity
and total wages and salaries (for both native and foreign workers) per worker has increased. Their
unit labor costs have increased progressively over time and the median of real profit and loss has
increased. In general, it would appear that both firms with and without foreign workers have tended
to substitute labor for capital and this has affected their respective real labor productivity over time.
MC8-125
4. CONCLUSION
This paper has examined the impact of foreign labor on labor productivity and wages and
salaries in Malaysian manufacturing, using establishment level data from 2000-2006. It finds that the
use of foreign workers has a negative impact on labor productivity and total wages and salaries for all
workers (both native and foreign). However, the negative impact of foreign workers on labor
productivity and total wages and salaries is small and not very robust across different occupational
workers. Moreover, the negative impact on labor productivity is smaller than the negative impact on
total wages and salaries for all workers. This suggests that the use of foreign workers has helped
improve the competitiveness and profitability of manufacturing firms in Malaysia by reducing unit
labor costs. However, there is some evidence that unit labor costs have been rising over time, while
conversely, capital labor ratios and labor productivity have been falling.
The first policy implication from these findings concerns the relationship between labor
productivity and total wages and salaries in Malaysia. The findings show that firms use foreign
workers to increase their competitiveness by lowering unit labor costs. Firms are able to do this
because the use of foreign workers decreases salaries and wages for all workers (both native and
foreign) by more than the decrease in labor productivity. It would therefore seem that changes in
labor productivity are not being accurately reflected in the wage and compensation package for all
workers, and that steps need to be taken to address this imbalance in order to make more efficient use
of foreign labor in Malaysia.
Second, the findings of this study show that increases in the capital-labor ratio and the value
of fixed assets have a positive impact on real labor productivity in manufacturing. These results
suggest that increasing automation (which will increase K/L and the value of fixed assets) will help to
155
increase labor productivity in Malaysian manufacturing. On the basis of its recent biennial survey of
manufacturing in Malaysia, the Central Bank Annual Report (2010) found that up to 60% of the jobs
that are currently being done by low-skilled foreign workers in Malaysia can to some extent be
replaced by automation (Central Bank 2010: 22). If the government is anxious to reduce the
country’s dependence on foreign workers, it is important for it to encourage the process of automation
in manufacturing. This will require new training and educational programs for native workers in all
segments of the manufacturing sector.
Third, the findings of this study show that foreign establishments have higher labor
productivity than local establishments. For this reason, it is important for the government to continue
attracting foreign direct investment in Malaysian manufacturing in order to improve labor
productivity in this sector. It is also important to continue efforts to improve the quality of labor in
Malaysian manufacturing in order to improve the absorptive capacity of labor. This will enhance the
positive technology spillovers from foreign establishments that have been found in other studies (see
for example Noor Aini and Radziah (2009) and Tham et al (2011)).
156
REFERENCES
Athukorala, P.C. and Manning, C. 1999. Structural Change and International Migration in East Asia:
Adjusting to Labor Scarcity. Melbourne: Oxford University Press.
Azizah Kassim. 2010. Transnational Inflow and Housing Pattern among Urban Migrants. Paper
presented at the 7th International Malaysian Studies Conference (MSC 7), 16-18 March 2010,
Universiti Sains Malaysia, Penang.
Blomstrom, M. and Kokko, A. 1998. Multinational Corporations and Spillovers. Journal of Economic
Surveys, Vol.12, No.2: 247-277.
Blomstrom, M., Kokko, A. and Zejan, M. 2000. Foreign Direct Investment: Firm and Host Country
Strategies. London: Macmillan Press.
Caves, R.E. 1996. Multinational Enterprise and Economic Analysis (second ed.). Cambridge:
Central Bank. 2010. Annual Report 2009. Kuala Lumpur: Central Bank.
Dunning, J.H. 1993. Multinational Enterprise and the Global Economy. Wokingham: Addison-
Wesley.
Evelyn, S. Devadason. 2009. Socio-economic Effects of Foreign Workers on the Host Economy: The
Case of the Malaysian Manufacturing Sector. Faculty of Economics and Administration (FEA)
Working Paper Series No. 2009-26. University of Malaya. http://www.fep.um.edu.my Accessed 20
February 2010.
Evelyn, S. Devadason and Chan, W.M. 2007. Inflow of Foreign Labor into Malaysia:
Conceptualizing Economic and Legal Issues: Faculty of Economics and Administration (FEA)
Working Paper Series No. 2007-17. University of Malaya.
http://www.fep.um.edu.my Accessed 20 February 2010.
Gujarati, D.N. 2009. Basic Econometrics. (fifth ed.). Boston: McGraw-Hill.
Hall, R.E. 1988. The Relation Between Price and Marginal Cost in US Industry. Journal of Political
Economy, Vol.96, No.5: 921-947.
Hill, R.C., Griffiths, W.E. and Lim, G.C. 2008. Principles of Econometrics. (third ed.). New Jersey:
John Wiley & Sons.
Kohpaiboon, Archanun. 2006. Foreign Direct Investment and Technology Spillover: A Cross-
industry Analysis of Thai Manufacturing. World Development, Vol. 34, No. 3: 54-556.
Noor Aini Khalifah and Radziah Adam. 2009. Productivity Spillovers from FDI in Malaysian
Manufacturing: Evidence from Micro-panel Data. Asian Economic Journal, Vol. 23 No. 2: 143-167.
Pholphirul, P., Rukumnuaykit, P. and Kamlai, J. 2010. Do Immigrants Improve Thailand’s

Competitiveness? Paper presented at the World Bank and IPS Conference on Cross-Border Labor
Mobility and Development in the East Asia and Pacific Region, 1-2 June, Singapore.
157
Quantitative Micro Software. 2007. EViews 6 User’s Guide II.
Takii, Sadayuki. 2004. Productivity Differentials between Local and Foreign Plants in Indonesian
Manufacturing, 1995. World Development, Vol. 32, No. 11: 1957-1969.
Tham, S.Y. and Liew, C.S. 2004. Foreign Labor in Malaysian Manufacturing: Enhancing Malaysian
Competitiveness? Chapter 12 in Abdul Rahman Embong (ed.), Globalisation, Culture and
Inequalities: in Honor of the Late Ishak Shari. Bangi: Penerbit UKM.
Tham Siew Yean, Liew Chei Siang and Marziah Mokhtar. 2011. Foreign Direct Investment and
Spillovers in Malaysia. Chapter 3 in Chalongphob Sussangkarn, Yung Chul Park, Sung Jin Kang
(eds.), Foreign Direct Investments in Asia. London: Routledge.
158
Chapter 7: Indonesia’s Regulatory, Institutional and Governance
Structure for International Labor Migration
ARI KUNCORO
University of Indonesia
ARIE DAMAYANTI
University of Indonesia
IFA ISFANDIARNI
Institute of Economics and Social Research
ABSTRACT This paper examines Indonesia’s regulatory and institutional structure for governing
international labor migration and the roles of various public and private agencies in facilitating such
migration. The paper finds that the lucrative profits enjoyed by the migration industry have led to the
creation of many new migrant sending firms. This has increased recruitment costs among migrant
sending firms and has also affected the distribution of values (rents) in the industry. Some of the
increase in recruitment costs can also be attributed to the introduction of Law 39/2004 which
formalizes the role of local recruiters (brokers) in the migrant recruitment process. While this law
also provides some protection to migrant workers, the net benefits to workers may not be great since
sending firms can always shift cost burdens onto workers through salary deductions. Recent efforts
to create a new independent agency in Indonesia to monitor international labor migration have yet to
bear fruit. A clearer division of labor between government agencies is needed, but due to the huge
rents involved in the migration industry, this may need political intervention from the highest levels of
government.
1. OVERVIEW OF INTERNATIONAL LABOR MIGRANT IN INDONESIA
In Indonesia, as in any other sending country, migratory pressures and domestic conditions
relating to employment and unemployment are inextricably linked. Since the 1999 Asian economic
crisis Indonesia has faced high levels of unemployment, especially in the formal sector. Although the
data in Figure 1 suggest that the unemployment rate has moderated to 8 percent in recent years, this
does not take into account the equally pressing problems of underemployment and hidden
unemployment. With an average annual GDP growth rate of 5.3 percent per year since 2000,
Indonesia faces a very difficult time of generating enough jobs to meet the needs of both the
chronically underemployed and the nearly 2.8 million people who enter the labor market every year
(Table 1). For this reason, the role of international labor migration is absolutely critical in Indonesia.
It is difficult to measure with any precision the exact number of Indonesians working abroad.
In 2008 the Indonesian Ministry of Manpower estimated that about 750,000 Indonesians went to
work abroad (Table 2). However, this number includes only those Indonesians who went to work
abroad with legal and registered contracts. The number of Indonesian migrants working abroad on an
159
Figure 1
Unemployment Rate
12.0
10.0
8.0
Unemployment Rate
6.0 Unemploy…
4.0
2.0
0.0
1995199619971998199920002001200220032004200520062007200820092010
Year
Table 1: New Entrants into the Indonesian Labor Market
Education Level 2000 2008
Entrants % of Total Entrants % of Total
Labor Force Labor Force
No schooling 178272 0.11 26953 0.01
Elementary Sch. 1793969 1.11 902171 0.48
Junior HS 810204 0.50 703196 0.37
Vocational Junior HS 54262 0.03 17404 0.01
Senior HS 683310 0.42 629635 0.34
Vocational Senior HS 54262 0.19 378773 0.20
1-2 year college 46810 0.03 42176 0.02
Junior College 49605 0.03 66485 0.04
University 110271 0.07 116097 0.06
Total Entrants 4029657 2.49 2882890 1.53
Total Labor Force 162000000 188000000
Source: calculated from Labor Statistics
160
Table 2: Skilled and Unskilled Migrants in Indonesia, 2007-2009
Year Skilled Unskilled Total
2007 196,191 (28.2%) 500,555(71.8%) 696,746
2008 266,749 (35.6%) 482,076(64.4%) 748,825
2009 103,918 (16.4%) 528,254(83.6%) 632,172
Source: Ministry of Manpower
illegal and unofficial basis is unknown, but probably quite high. For example, it has been estimated
that about 1.8 million of the Indonesians working in Malaysia are there illegally,
Although the educational attainment of the Indonesian labor force has been improving over
time, most Indonesian migrant workers are still unskilled. According to Table 2, 84 percent of all
legal Indonesian migrant workers were unskilled in 2009 and working in such jobs as housekeeping,
transportation and agriculture. This explains why in terms of gender most legal Indonesian migrant
workers are female. Table 3 shows that in 2008 74 percent of all Indonesian migrant workers were
female, working primarily as housemaids and care givers.
Table 3: Indonesian Migrant Workers by Gender

Gender 1994 %1994 2000 %2000 2008 %2008
Male 42,833 24.4 137,949 31.7 200,188 26.4
Female 132,354 75.6 297,273 68.3 548,637 73.6
Total 175,187 100.0 435,222 100.0 748,825 100.0
In recent years there has been shift in the primary destination of Indonesian migrant workers.
According to Table 4 (next page), between 1994 and 2008 the share of total legal migrants going to
Asia Pacific increased from 40 to 54 percent, while the share going to the Middle East fell from 56 to
46 percent. Malaysia, which shares a common border with Indonesia, is now the most popular
destination country for Indonesian migrant workers, accounting for about 35 percent of all legal
migrant workers.
2. STRUCTURE OF THE MIGRANT INDUSTRY
The structure of the Indonesian migrant industry has evolved from oligopolistic in the late
1980s and the early 1990s to near monopolistic-competition in 2004. This is reflected in the number
of migrant recruitment firms, which increased sharply from about 100 in the 1990s to around 500 in
2009. Although the number of recruitment firms is relatively large, the industry is fairly concentrated
with about 100 firms controlling 70 percent of the business. All recruitment firms are privately owned
and are required to become a member of one of 5 recognized associations.1 These five associations
serve as lobbyists for the industry’s interests, and also act as barriers to entry, particularly
discouraging the entry of small firms with little capital.
1
It is not clear if there are any significant links between the ownership of these firms and government officials regulating
this industry. But active membership of these firms in at least one of the existing 5 associations and active lobbying by
associations against government policies detrimental to the business, suggest that they are not or at least remotely linked to
government officials. The biggest and the most active association is APJATI which has about 200 members.
161
Table 4: Indonesian Migrant Workers by Destination
Destination 1994 1994 (%) 2000 2000 (%) 2008 2008
(%)
ASIA PACIFIC (AP) 70,733 40.4 304,186 69.9 391,784 53.8
MALAYSIA 41,712 23.8 191,700 44.0 257,710 35.4
SINGAPORE 15,678 8.9 25,707 5.9 21,867 3.0
BRUNEI. D 1,846 1.1 4,370 1.0 4,967 0.7
HONGKONG 3,306 1.9 21,709 5.0 30,207 4.1
TAIWAN 3,423 2.0 50,508 11.6 62,433 8.6
SOUTH KOREA 3,294 1.9 6,689 1.5 13,546 1.9
JAPAN 333 0.0
OTHERS IN Asia-Pacific 1,474 0.8 3,503 0.8 721 0.1
MIDDLE EAST 98,710 56.3 129,168 29.7 334,440 45.9
SAUDI ARABIA 96,533 55.1 114,067 26.2 234,643 32.2
UAE 1,948 1.1 9,558 2.2 38,200 5.2
QUWAIT 76 0.0 3,771 0.9 29,224 4.0
BAHRAIN 1 0.0 169 0.0 2,325 0.3
QATAR 19 0.0 949 0.2 8,716 1.2
OMAN/ TUNISE 8,314 1.1
JORDAN 6 0.0 11,165 1.5
OTHERS IN ME 133 .1 648 0.1 1,853 0.3
AMERICA 4,036 2.3 1,509 0.3 0.2
USA 3,950 2.3 1,302 0.3 0.2
OTHERS 86 0.0 207 0.0
EUROPE 1,708 1.0 359 0.1 1,293 0.2
TOTAL 175,187 100.0 435,222 100.0 696,746 100.0
162
In Indonesia the small, semi-formal firms with little capital are often called “kakilima” (street
vendor).2 Since these “kakilima” firms do not have facilities like decent offices, training centers and
holding facilities, they typically offer little training to prospective migrants. To cut costs these
“kakilima” firms also tend to take shortcuts and to falsify documents. For example, “kakilima” firms
usually offer training to migrants for only a day or two. However, according to the association, an
inexperienced domestic worker would need about 200 hours of training to work abroad, while an
experienced one would need 150 hours.
“Kakilima” firms exist because there is a market for their services. For poorer migrants the
key issue is choosing between becoming a legal or an illegal migrant. Being a legal migrant is
obviously the safest choice because legal migrants can defend their rights in the event of a labor
dispute. But the costs in terms of money and delays in obtaining legal status are often prohibitive,
and so many poor migrants opt to move illegally. As noted above, about 1.8 million of the 4.0
million Indonesian migrant workers in Malaysia are working there illegally.
Migrant Sending Firms (PPTKIS)
Attracted by the prospect of large profits, many new recruitment firms have opened in the last
25 years. As a result, profits have declined from between USD 1000 to USD 1250 per unit migrant
sent in the late 1980s, to between USD 100 to USD 300 per unit migrant sent in more recent years
(Table 5). Sending firms have attempted to make up for this decline in profits by sending more
workers abroad, but this is not easy to do. Also, profit per worker varies by country of destination.3
Sending migrants to the Middle East is more profitable than sending them elsewhere, at a rate of
about USD 200 to US300 per person sent abroad to the Middle East.4
Table 5: The change of the industry
1990s After the Law 39/2004
Market structure Oligopolistic More competition
Number of firms 100 500
Profit per migrant worker USD 1000-1250 USD 100-300
Source: interview with APJATI
The decline of profits comes not only from the rising number of firms but also from the
various government regulations which raise costs which cannot be directly passed on to the migrant.
Most of these cost increases can be attributed to the government requirement to establish recruitment
branches in rural cities. This however does not deter new firms from venturing into the industry. The
2
Ministry Ruling in 2009 based on the Law 39/2004 requires every private recruiter to have a training center after the start
of the firm operation. At present about 100 members own training facilities, while the rest rent from other member
possessing it.
3
Sending firms obtain orders from various sources but the internet has become the most important source of information. A
firm usually sets up website to promote itself. For the mid east countries however personal visit is still important to establish
business relationship.
4
Airfare dramatic increases after 2004 also contributed to the decline of profits per unit worker. At the same time job order
fee remains flat.
163
reason is the growing demand for Indonesian domestic helpers in the Middle East (Saudi Arabia),
East Asia (Hong Kong and Taiwan) and Southeast Asia (Malaysia and Singapore).
Law 39/20-04 increases the capital requirement for a recruitment sending firm from IDR 700
million to IDR 3 billion (USD310,000). The official explanation for this increase is to prevent
irresponsible small firms from entering the recruitment market. However, the most logical
explanation is that the continuous erosion of profits has prompted the association to lobby the
government for higher barriers to entry. The large increase in the number of migrant recruitment
firms in recent years signals just how lucrative this industry has become over time.5
In Indonesia the small, illegitimate firms usually send their workers to Malaysia.
Having no proper licenses, these firms tend to falsify travel documents just to get migrant workers
across the border into Malaysia. Once there, agents from the firms manufacture documents to make
the migrants appear “legal” and to send them onto requesting parties.
Although Indonesia has a human trafficking law that specifically prohibits the trafficking of
illegal and under age workers, the government can do little to stop the movement of these migrants
into Malaysia. From time to time the police set up roadblocks and conduct patrols in border areas to
try to stop illegal migrants, but these actions are seldom effective. In the field, the police try to find
workers being transported by legitimate firms. Since the new law contains stiff penalties for illegally
transporting workers, these workers are subject to police search.6 During a police search, all
legitimate documents become useless and migrants typically wind up paying bribes to the police to let
them go.7 This in turn effectively raises the recruitment firm’s cost, since the cost of these bribes must
in the end be borne by migrant workers in the form of wage deductions. In short, the new legislation
that is supposed to protect workers is used by corrupt officials to extract bribes, which eventually are
passed onto unsuspecting migrants by recruitment firms.
Indonesian law requires that firms set up rural branches in order to recruit migrants from rural
areas. This creates unfair competition for recruitment sending firms that are not located in Jakarta.
The problem stems from the lack of enforcement from the government. Under the new law only the
headquarters of the firm has a permit to send workers abroad. But in practice, as shown in the case of
East Java, rural branches can also send people abroad through regional airports. This puts local
sending firms in East Java and other provinces at a disadvantage relative to those in Jakarta.
In order to send workers abroad, each sending firm (PPTKIS) needs to acquire business
licenses. These licenses are required not only for headquarters, but also for all rural branch offices
and recruitment units.8
5
Arnold (2007) has advocated to lower barrier to entry to put firms in the industry to more competition. This however
should be done with caution. Without entry barriers new entrants tend to be small and less willing to spend on training
facility and tend to shorten the mandatory training period to cut costs. This unhealthy competition would undermine other
firms that follow the book, which may induce them to follow similar suit. Besides sacrificing the quality of training,
unhealthy competition could also be potentially detrimental to migrant at the bottom of the pyramid through the other way.
If the profits are pushed too thin near firms’ reservation profits, then firms could always shift this burden to workers. Now
migrant workers no longer have to pay IDR 1 million but in exchange workers will have to forgo their salaries in the first
few months.
6
The law of Trafficking of 2008 carries the maximum penalty of IDR 5 billion for transporting illegal workers as well for
recruitment of underage workers.
7
For an illustration, for a train trip transporting 40 workers from Malang East Java, to the capital city of Jakarta, a firm must
set aside IDR 1 million. About IDR 500,000 will be paid to station officials, while the rest is paid to street patrol/police in
the short trip from the train station to the company headquarter.
8
The introduction of regional decentralization in 2001 brought other players into the industry namely regional governments.
Soon various local regulations emerge in an attempt to extract revenues from the international migrant industry. For sending
firms as well as for prospective migrants these serve only to lengthen the bureaucratic chain and to drive the costs up
164
Brokers or Sponsors
The main players in the migration industry are not limited to firms and migrants. The industry
is perhaps one of the best examples of asymmetric information. On the one side, there is a modern
industry with a good knowledge of labor demand abroad. On the other side, there are many hopeful
migrants from the countryside with limited education who are eager to escape poverty. Between these
two sides are the brokers or sponsors who attempt to reconcile demand with supply. For the
recruitment firms it is almost impossible to work without brokers – the search costs for identifying
recruits would be prohibitive. Brokers usually operate in the proximity of their home village, but it is
also not unusual for professional full time brokers to also operate within the kabupaten (district)
where their villages are located. Each broker maintains an army of informants who are paid according
to the number of potential recruits that they can identify.
Before 2004, most brokers would never work for just one recruitment firm. During these
years the number of firms was limited while the number of brokers was large and so the bargaining
position of the latter was limited. In effect, firms could pit one broker against another to lower
commission costs. Before 2004 the pool of potential migrants was also quite large. In fact, the
supply of migrants was so great before 2004 that it was possible to charge each potential migrant
between IDR 500,000 to IDR 1 million to migrate. This was a bleak period for migrants since the
costs to pursue international migration were high. But the lure of high wages abroad and poverty at
home was enough for them to press on. This was the heyday for recruitment firms as they enjoyed
large profits which were once as high as USD 1000 per unit migrant.
The introduction of Law 39/2004 is very pivotal to the migrant recruitment industry. Not
only does this law regulate the entry of new firms into the industry, but it also changes the
distribution of rents in the industry. Previously, almost all rents were captured by sending firms and
very little was left for government officials and brokers. Law 39/ 2004 shifted the balance toward
brokers and government/security officials in the form of legal and not so legal payments.
Law 30/2004 changed the position of informal brokers by requiring each recruiting firm to
establish rural branches if it wanted to recruit rural workers. With this law each firm established local
branches and typically appointed its most trusted broker as branch head. The task of the branch office
is to make contractual arrangements with a number of local brokers in order to mobilize rural recruits.
Instead of brokers, those working for the firm are now called field representatives (petugas lapangan)
or PL for short. Every PL is paid in advance IDR 3 million (USD 310) for each worker that he
recruits.9
Migrants Workers
without any real benefits added to the process. Although the practice differs from one district (kabupaten/kota)
to another, basically there are three documents issued by this third tier government; the letter of notification
from kabupaten government to sending firms that the respective migrants have participated in the district
briefing program, placement agreement and passport recommendation.
9
Typically, an experienced PL could recruit 5 to 6 persons a month. Instead of being paid a commission fee, income for a
PL is IDR 3 million minus costs he or she incur in the search process, including signing bonus he pays to potential workers
which costs between IDR 1 million to IDR 1.5 million. After deducting from all costs incurred during the process, for a
single recruit a PL can earn between IDR 1 million to IDR 1.5 million per person recruited or equivalent to a monthly
income between IDR 5 million to IDR 7.5 million, a handsome pay by a village standard.
165
Migrants and the Industry Structure
At the forefront of the migrant recruitment business are the migrant workers themselves.
Once a potential recruit agrees to be sent abroad by a firm, he/she will be escorted to a temporary
holding facility where he/she will undergo a basic training course including household chores, basic
manners and foreign language instruction for 20 days. To lure a potential recruit into working for
them, a firm will typically offer a signing bonus ranging between IDR 300,000 to IDR 500,000.
Recruitment firms compete among themselves to attract suitable candidates. In practice, a

prospective migrant can wait up until 3 months before being sent abroad because a job order has not
arrived. Although the law requires each firm to recruit a person only if it has a definite job order, in
practice firms do not follow this procedure. The unpredictability of job orders and the time required
for training makes this provision impractical and thus it is almost never enforced by the government.
So for the industry, holding a stock of prospective migrant workers is still the common practice.
Some firms have done good job in predicting the timing of job orders so that they can minimize
costly idle waiting time.
By changing the recruitment process, Law 39/2004 in effect has accelerated the change of the
market structure from oligopolistic to a more competitive one. But one change in the law that also
impacts the recruitment industry is often overlooked, namely, raising the minimum age of migrants
from 18 to 21 years. This increase in the minimum age effectively reduces the pool of potential
migrants. As a result, firms now must compete to get good candidates from a shrinking labor pool.
Because of the new law, potential migrants now have more bargaining power vis-à-vis the
recruitment firms. Many researchers have argued that migrants are the weakest player in the
recruitment industry, especially first-time migrants. With limited knowledge about the recruitment
process and their legal rights, migrants are subject to fraud and extortion. The fact that migrants still
have to bear much of the costs of getting abroad tends to reinforce this argument. Dwindling profits
and rising costs now force recruitment firms to shift pre departure and document costs onto migrants.
Ironically, the law (Law 39/204) that enables migrants to have more choices and protection, also
introduces more competition and raises the costs of recruitment. Some firms have sent more workers
to work abroad in order to recover lost profits, but it appears that the cost of recruitment has also
increased with the number of workers sent.10
The pressure of competition and increasing costs have made it is almost impossible for
sending firms not to shift some burdens to migrant workers. This usually comes in the form of the
future wages deduction. For example, in Malaysia, a typical migrant has to forgo the first 8 months of
his/her salary to cover for transportation from the village, food and accommodation during waiting
periods, training costs, and permits.
Comparison of the Treatment of Migrant Workers among Receiving Countries
The potential earnings of international migrants vary considerably depending on the country
of destination. Fees and charges also vary. Migrants choose their country of destination based on
information received from returning family members, relatives and brokers. This explains why
migrants in one village tend to specialize in one particular country destination. The choice of country
will in turn determine the risks associated with salary, placement procedures, and working habits. In
general, the more open the destination country is the better treatment that the migrant worker is likely
to receive. Some details described below depict these differences between countries.
10
Increasing the size of workers sent also makes firms more vulnerable to extortions.
166
Salary and Its Deduction
The salary received by a migrant varies by country and the type of job, experience and
competency of the migrant (Table 6). For example, experienced domestic helpers in Hong Kong with
ability to speak Cantonese, Mandarin and English can earn up to US$461 per month, while an
inexperienced helper would earn only US$290 per month. According to Hong Kong law, tax and
insurance are borne by employer.
For migrants going to work in East Asia, the sending firm will typically retain 8 months of
the migrant’s salary to pay for pre-departure and placement costs (Table 6). In the more extreme case
of Taiwan, the salary will be deducted gradually for 18 months to repay all costs. Since a migrant
working contract is normally valid for 2 years, a domestic helper in Taiwan will receive full amount
of salary for only six months, starting from the 19th month of her working period.
Table 6: Estimated Migrant Salaries by Destination Country
Region/Country Monthly Salary Deduction
Middle East
Saudi Arabia USD190 – USD230 None
Jordan USD150 – USD175 None
Qatar USD165 – USD175 None
United Arab Emirate USD218 None
Bahrain USD154 None
Asia Pacific
Malaysia USD136 5 months salaries and tax paid

by employers
Singapore
Experienced helper USD267
Inexperienced helper USD242
Hong Kong 8 months
Experienced helper USD461
Inexperienced helper USD290
Taiwan USD416 18 months salary in a decreasing

scheme
Source: Field interview with migrant workers
167
By contrast, the pre departure and placement costs for domestic helpers in the Middle East
are not deduced from the migrant’s salary.11 Thus, a domestic helper will receive the full amount of
her salary from the first month. This monthly salary will amount to USD 165 in Qatar, USD 150 in
Jordan, USD 154 in Bahrain, USD 218 UAE and USD230 in Saudi Arabia. Since Saudi Arabia is the
highest paying country, this is the most favored destination in Middle East. The opposite situation
obtains in Malaysia, where a domestic helper can only expect to earn USD 136 per month. Yet since
Malaysia is culturally similar to Indonesia, many migrants prefer to work in Malaysia. However, in
Malaysia PPTKIS retains 5 months of the migrant’s salary to cover pre-departure and placement
costs.
On the basis of our findings in the field, migrant workers normally receive their monthly
salary in cash. After receiving payment, migrants can remit money by themselves to their bank
account in Indonesia. However, in the Middle East women are not allowed to go outside, and so in
these countries the workers would have to request their employer to send their money home to
Indonesia through the employer’s bank.
Legal versus Illegal
Since so many Indonesian migrants go to work abroad without official contracts, it is

important to consider the various types of legal and illegal migrants. There are three basic types of
legal-illegal cases in Indonesia. First, a migrant is considered as legal in Indonesia but treated as
illegal abroad. For example, at the beginning, the migrant is working for the first employer Mr. X, but
for some reason, he/she does not feel comfortable and decides to go and work for a second employer,
Mr. Y. However, the name of first employer Mr. X is still recorded on the migrant’s working
documents. According to the law, when working for the second employer Mr. Y, the migrant is
considered as illegal. The migrant could become legal if he/she reports to the foreign agency or
PPTKIS or Indonesian Embassy to issue a new document with the new employer’s name.
In the second case, the migrant worker is considered as illegal in Indonesia but treated as
legal abroad. This case usually needs some ‘collaboration’ between Indonesian officials and the
destination country’s agencies so that the migrant’s documents (e.g. passport and visa) are “doctored”
to appear legal for working purposes, when in fact these documents are still irregular. While these
documents are not necessarily bogus, a careful inspection would reveal their irregularities. However,
even when these irregularities are discovered, officials may choose to look the other way after being
paid bribes.
In the third case the migrant is considered as illegal in both Indonesia and the destination
country. This case occurs only if all the stakeholders – including the police and company officials in
the destination country -- “collaborate” in order to cover each other. Most of these illegal cases come
from Indonesian migrants working in Malaysia, because Malaysia has 198 entry gates from Pangkal
Pinang in Sumatera, and Pontianak and Nunukan in Kalimantan.
Regardless of the type of legality, there are some strong reasons underlying the choice of
Indonesians to become illegal or irregular migrants. In most cases, the basic motivating factor is
poverty: remaining in the village means that the prospective migrant will barely be able to afford
his/her basic needs. While some of these migrants can occasionally find work in plantations or
construction in Indonesia, they earn very little income. For this reason, many of these poor rural
villagers choose to go work abroad as soon as they can. Even with the salary deductions for pre-
11
This reflects how thin is the profit for sending people to Southeast and East Asia destinations
168
departure and replacement costs, rural migrants from Indonesia can earn much more working abroad
than they can at home.
Another reason for choosing to be an illegal migrant is the fear of failing the medical
examination given to all prospective migrants. By definition, poor rural migrants lack the nutrition
and sanitary conditions needed to pass the medical examination. Only those migrants who pass the
examination can hope to move on a legal and official basis.
With regard to risks, many prospective migrants are not well informed about the
consequences of working abroad illegally, especially with regards to protection. Since it takes both
time and money to get the necessary documents to migrate legally, many prospective migrants choose
to move illegally. However, once abroad, these illegal workers have little or no protection from the
abuse of unethical employees abroad.
Formal or High Skilled Sector: A Move toward Skill Upgrading?
The formal sector usually hires technicians at a factory or other industrial site, such as an
automotive or electronic plant. Korea is one of the countries that recruit technical workers from
Indonesia. The basic salary for a technician in Korea is around USD 441 per month, plus allowances
received. If all allowances are included, the total take home pay for a migrant could reach USD970.
For higher skilled technical workers in Korea, such as welders, the salary varies according to the level
of expertise. For instance, the salary for the lowest grade welder is USD2.4 per hour with a minimum
working time of 10 hours per day, while the salary for the highest grade welder is USD 8 per hour. In
other words, the monthly salary of a welder in Korea ranges between USD 600-2000. These figures
represent net income because health insurance is paid by the employer.
Migrants in the formal sector from Indonesia could use their savings from jobs like this to
start up businesses at home. In addition, migrants could use their experiences learned abroad in such
areas as entrepreneurship, discipline and hard word to improve the quality of human capital in rural
areas. As an illustration, Arjowilangun village from our field survey, known as a traditional sending
village in East Java, is famous for its good entrepreneurship and high tax revenue. Our field visits to
this village revealed a wide range of small businesses established by former migrants, including
convenience stores, workshops, furniture and garment stores, and herbal shops. Sending formal
workers abroad is currently being carried out as a project package in Indonesia covering both medium
skilled workers, like cooks, drivers and security people, and high skilled workers, such as welders and
plumbers. Recruitment firms do not have to train these people because they only recruit experienced
candidates who have specific skills. Recruiting mechanisms for these formal sector workers include
advertising in mass media, and on community and network sites. For very high skilled candidates
such as welders in the oil and mining industry, the sending firms will invite the prospective
employers to Indonesia to examine how the candidates perform their required job. After the recruiting
process, sending firms can start preparing documents required to work abroad. These documents are
basically the same with those for informal ones as discussed previously12. An interview with one of
the largest private recruitment firms specializing in sending formal workers abroad suggests that
employers in destination countries often complain about the attitudes of Indonesian workers.
Specifically, Indonesian migrant workers in the formal sectors are typically known for their lack of
long term commitment to their jobs. They often resign after their two-year work contract is over, and
are very reluctant to renew or to renegotiate. This is a problem because formal sector jobs have their
12
Basic documents for working abroad are as follows: Passport, Working Visa, Tax Registration Number (NPWP) for Free
Fiscal Tax, Migrant Worker ID (KTKLN). Basic ID documents such as Family Card, ID card, and Birth Certificate are still
needed to apply passport.
169
own career paths which need long term commitment to reach higher positions. Each career path needs
a specific skill so a company can allocate funds to train workers as human capital investment. For this
reason, if a migrant worker chooses not continue his/her job after a two year period, the employer will
regard this as a substantial loss. Since there are very few repeat orders for Indonesian formal workers,
to stay in business sending firms in Indonesia always have to find new markets for their skilled
workers. From our findings in the field, this really increases the costs of sending firms, since the
market penetration costs of a typical sending firm are about 4 times of its overhead cost or about USD
20,000.13
The above attitude is very unusual compared with skilled migrants from other countries. In
Indonesia the main motivation to go abroad is to gain prestige and additional work experience. For
Indonesian formal sector workers 2 years of international experience abroad is enough to secure
employment in the local Indonesian market. In many cases, Indonesian formal sector workers are
anxious to return home because the costs of living abroad are typically two to three times higher than
living in Indonesia. For this reason, the same standard of living or higher can be achieved in
Indonesia with a much lower salary. From the standpoint of sending firms, the biggest impediment
for sending skilled workers to work abroad is its low profits. Recruitment firms in Indonesia do not
earn any more profit in sending skilled workers vs. unskilled workers abroad. In both cases the
profits per worker amount to between USD 100 and USD 200. For this reason, sending firms can only
make money if they send a great number of skilled workers. The problem is in the supply side,
namely, it is difficult to find suitable skilled candidates in Indonesia who are willing to go work
abroad.
Comparison of Net Incomes among Workers: Illegal, Legal, Unemployed
As discussed earlier, salaries for migrant workers vary greatly depending on the country of
destination, and the type of job, experience and competency. At first glance, it would appear that an
illegal migrant working abroad could earn more than a legal migrant, since the former would not have
to pay deductions for pre departure and other costs. However, since legal migrants typically receive a
basket of benefits that illegal migrants do not receive, the net income of legal migrants is usually
higher.
Regardless of legality, migrant workers earn more than those in their home villages. For
example, many illegal migrants working in construction in Malaysia can earn RM30 per day or
RM900 per month (equivalent with USD271 per month). By contrast, a worker in an Indonesian
village can only work maybe one or two weeks per month (in the rice fields or sugar mills) and earn a
daily wage of IDR 15,000 or USD 1.5 (equivalent to USD21 per month)14. Since this is not enough to
meet their living costs, many villagers have to seek complementary jobs in trade and services in
neighboring towns or cities
Meanwhile, legal workers in Malaysia working as domestic helpers can earn monthly salaries
of RM450 or about USD136 (Table 6 in page 169).). At first glance, it appears that these salaries
(USD 136 per month) are lower than those earned by illegal workers (USD 271 per month).
However, this comparison is not accurate because illegal workers have to pay all of their own
expenses, including meals and accommodation. By contrast, domestic helpers in Malaysia receive
food and accommodation for free from their employers. Moreover, illegal workers also face the
constant risk of being caught and deported by the police at any time.
13
The overhead cost of a typical firm sending formal sector workers is estimated at USD 5,000 per month.
14
Wage less than USD2 per day is categorized as working poor.
170
Structure of Governance
With respect to the idea of worker protection, the attitude of the government has evolved
from indifference to embracing the need for increased protection. This is an important shift in
government attitude and one that has been primarily motivated by pressure from NGOs. The
emergence of a free press in Indonesia after the fall of Suharto has enabled NGOs to keep the issue of
the abuse of migrant workers in the headlines of newspapers and periodicals.15 This change in
government attitude started under the Megawati’s presidency with the issuance of Ministry Decree
104A/2002. This decree was later upgraded to become Law 39/2004 (Ananta and Arifin [2007]).
As a matter of principle, the Indonesian government seeks to pay back its ‘foreign exchange
heroes’ by providing them more services and more protection from the threat of fraud and extortion
abroad. The government believes that the migrant industry is not doing enough to protect the rights
of migrant workers, and so it must be better regulated. This is the spirit of Law 39/2004. At the top of
pyramid is the Ministry of Manpower which oversees regulation of the migrant industry.
For all its good intentions, Law 39/2004 contains many flaws. For example, it is often
alleged that this law is quite naive with regards to its requirement for migrant sending firms to
establish branches in destination countries. This requirement is impossible under the present laws of
most destination countries. At best, migrant sending firms in Indonesia can only make contractual
partnerships with counterpart firms in destination countries.
The law also serves as a basis for a creation of a new agency under the President called
BNP2TKI (National Agency for Placement and Protection of Indonesian Overseas Workers). The
task of this agency is to oversee the integration of services in the placement and protection of migrant
workers among the various government agencies (Ananta and Arifin [2007]). BNP2TKI is
independent of the Manpower Ministry, and is responsible directly to the President.
Officially, the Manpower Ministry is the supreme policy maker or regulator and BNP2TKI
acts as ‘operator.’ However, the law is so vague that there is no clear demarcation of authority
between the Manpower Agency on the one hand, and the BNP2TKI on the other. According to
Presidential Rule No. 81/2006, BNP2TKI has two basic functions. First, it is supposed to provide
placement for Indonesian workers under Government to Government scheme (G to G) and
Government of Indonesia to foreign agents. This tends to wrest away the business of sending skilled
and educated workers abroad from private firms. Second, BNP2TKI is supposed to provide services,
monitoring and coordination to all stages of international worker migration, including the checking of
documentation, sources of financing, departure and return, improving the quality of migrant workers,
and protecting the welfare of families left behind (Ananta and Arifin [2007]).
The creation of BNP2TKI therefore creates a direct challenge to the supreme position of the
Manpower Ministry. It is not just a power struggle, at stake here are huge rents generated by the
whole migrant industry. On the one hand, the Manpower Ministry continues to assume its authority
in implementing the monitoring of recruitment firms. In 2008, the ministry issued Ministry of
Manpower Regulation 22/2008 that basically emphasizes its authority, in cooperation with local
government, to implement the placement and protection of migrant workers system for non-
15
Interestingly secular NGOs are more vocal on this issue, while Islamic organizations are more subtle on this matter. One
plausible explanation is that since the majority of migrant workers go to Islamic countries especially Saudi Arabia and
Malaysia. Also, Islamic organizations with huge political base in the country side may be more pragmatic that they do not
want to alienate their constituency who see working abroad as the only way for their constituents to escape from poverty.
171
government scheme.16 On the other hand, BNP2TKI believed that this new ministry ordinance
restricted its authority to exercise its mandate to protect the rights of international migrant workers.
BNP2TKI brought its position to the Supreme Court and won.
For migrant sending firms the creation of BNP2TKI only adds to the confusion surrounding
the migrant industry. Although the creation of this agency creates a loophole which can be exploited
by sending firms, the uncertainty caused by the dualism of authority between the Manpower Ministry
and BNP2TK exceeds the potential benefits from any loophole. For example, if for some reason the
application for a recruitment license (SIP) is rejected by BNP2TKI, then sending firms can always go
to the Manpower Ministry for the same license. The costly redundancy takes place when one agency
only recognizes licenses issued by its own institution and not recognizing the other. So as a
precaution migrant sending firms usually keep documents, permits and licenses from both agencies.17
Distribution of Rents
The basis of the rent distribution calculation is different from one player to another. For
migrant workers, the earned wages during 2 years of placement is the basis for the income
calculation. So the calculation of overall workers incomes will be based on the ‘stock’ of workers at a
particular time. For sending firms, they receive profits as long as there is a continuous flow of
migrants seeking jobs abroad. This is also true for incomes or rents received by brokers, and
government agencies both at the national and local levels. The rent distribution calculation uses one
year as the time reference since in some countries workers will receive no salary for the first 6 to 8
months in order to pay back their pre departure costs.
According to the APJATI association there are 2000 Indonesian workers departing daily to
overseas locations, or 60,000 per month, or 720,000 a year.18 From this figure a conservative estimate
suggests that about 60 percent will work in the informal sector. From these migrants, and not from
those who are already settled for more than 6 months, much of profits, incomes, rents and informal
payments for other players other than migrants are generated.19
For one job order a migrant sending firm receives a fee between USD 800 to USD 1500,
depending on the country of destination. The lower figure is for nearby countries like Malaysia and
Singapore, while the higher figure is for Middle East countries. The sending cost (including airfare)
per migrant ranges between USD 700 to USD 1300 depending on distance. This already includes non
formal payment to smooth business.20 These figures suggest that the profit per one unit of departure is
between USD 100 to USD 200. Profits are slightly higher for the Middle East. Using a weight of 0.62
to 0.38 for Southeast Asia versus Middle East we arrive at the average daily profits of USD 168,000,
16
Under this ordinance, the local government is authorized to conduct the pre-departure brief and to issue the migrant worker identity card
(KTKLN). Later the ordinance was cancelled after judicial review in 2009 by the Indonesia Supreme Board.
17
One example is pre departure briefing (Pembekalan Akhir Pemberangkatan or PAP). The worker identity card (KTKLN) is just another
example of the dualism. To be safe sending firms may keep PAP and KTKLN from both agencies. The cost of KTKLN from the Manpower
Ministry is IDR 4900, while the BNP2TKI version costs IDR 36,000. The cost of PAP from The Ministry of Manpower is IDR 30,000,
while the cost of PAP from BNP2Tki is IDR 70,000. These duplications make the accurate monitoring the flow of migrant workers
becomes difficult if not impossible since these two databases are separate. For one thing, the dualism is certainly to increase costs of
recruitment by service duplications.
18
This estimate from APJATI seems to be correct, in 2006 the actual figures is 680,000 migrants (World Bank
[2008])
19
Overall there are about 6 million migrant workers abroad both legal and illegal. Most of them or 4 million
went to Malaysia of which 1.8 million are illegal.
20
Mostly this will involve cash but sometimes it can also include luxurious goods.
172
or USD 5 million per month, or USD 60 million per year for the industry.21 This profit will be shared
by about 500 sending firms.
For migrant workers, it is assumed that for an Asia destination the payback period is 6
months; therefore, in the first year they will receive salaries for only 6 months. There is no such
payback for Middle East so migrants will receive a full 12 months’ salary. Assuming that in one year
there are 432,000 informal sector migrants going abroad, of which 62 percent are going to Asia while
38 percent are going to the Middle East, with a weighted average salary of USD 197 in Asia and USD
206 in the Middle East, we arrive at a figure of USD 722 million a year of combined departing
migrant salaries.
So for the sending firms the share of yearly profit rate from the total departing migrant
salaries is 9 percent, which is a decent rate of profit. Without wage payback the profit rate would be
much lower. The payback for Asia destinations is about USD 528 million; without this self financing
by workers, sending firms could not make money. Profit rates would be much higher if one compares
a profit for a single departure with the value of recruiting fee from abroad. For the Asian destinations
the profit rate per unit would be USD 100 over USD 800 or 12.5 percent, while for Middle East is
USD 200 over USD 1500 or 13.3 percent. Again these figures are high but far from spectacular
profits.
The unit cost of sending worker abroad is about IDR 13 million which includes air fare,
recruitment fee (broker), license fees and other informal payments. All legal fees related to complete
documentation are presented in Table 7 (next page). The estimate of costs from licensing is around
IDR 1.7 million. For informal payment a typical firm which has the capacity to send 250 workers per
month must set aside a significant sum of IDR 30 million per month. This means the informal costs
are roughly at IDR 1.2 million per worker. Sending firms have to finance brokers about IDR 4 million
per worker. Adding up this together we arrive at figure of IDR 6.9 million per worker. This still
leaves a room of IDR 7.1 million for airfare and other expenses like placement fees that have to be
paid to agencies in destination countries. The breakdown of the sending costs is shown in Table 7.22
The costs associated with the administrative process is IDR 2.9 million (both legal (1.7
million) and informal (1.2 million)), which is about equal to 2 months wage deduction in Malaysia.
Although it is high by Indonesian standard, it is not excessively high. The percentage of non-formal
payment (9.2 percent) in Table 8 (next) appears in line with that of manufacturing industry, which has
a figure of 10.8 percent (Kuncoro [2004]).23 This confirms the earlier assertion that the migrant
industry does not generate super high profits.
Through various licenses, the government generates IDR 1.7 million or USD 185 for a single
departing worker. For the estimated 432,000 departing migrants per year, the total amount generated
would be about USD 80 million. The question is then what kind of services the migrants get in return
beyond the cost of running two agencies – Manpower Ministry and BNP2TKI? The answer is
probably very little. The industry appears to finance itself from the recruitment stage to the final
departure, with all expenses being paid by job order fee from abroad, wage deduction (for the Asia
destinations) and to smaller extent from profit reduction. The most recent requirement to own training
21
This figure only covers legal migrants.
22
Sending firms can lower costs if they can negotiate lower airfares and/or under the table or informal payments.
Negotiating airfare is easier since for one batch of departure they can send a large number of passengers, but under the table
cost or payments always come in piecemeal fashion and solicited by different agencies.
23
Bribe rate above 10.0 percent of the total cost is considered as excessively high in Indonesia (Kuncoro [2004]). For
example for wood industry the bribe rate is estimated in excess of 15 percent of production cost.
173
Table 7: Cost of Licensing
Type of documents Institution Cost in IDR
1. Passport Immigration 250,000
2. Heath Certificate Health Office 300,000
3. Birth Certificate Village, Sub-district 500,000
4. Pre Departure Briefing Association 50,000
BNP2TKI 70,000
5. Worker Identity Card Manpower Ministry 4,900
BNP2TKI 36,000
6. Competency certificate Manpower Ministry 50,000
7. Insurance Consortium of firms 400,000
8. DP3TKI Manpower Ministry 145,000
9. Total Costs Allowing for duplication 1,805,900
Ministry of Manpower 1,699.900
BNP2TKI 1,751,000
Notes: the calculation does not include fees paid to local governments
Source: calculated from the field interview
Table 8: Structure of cost per departing Indonesian worker

Type of cost Cost per worker in IDR Percentage Distribution
Recruitment Cost 4.0 million 30.8
Licensing cost (legal) 1.7 million 13.1
Informal Payments 1.2 million 9.2
Air fare and others 6.1 million 46.9
Total cost 13.0 million 100.0
Sources: calculated from the field interview
center (BLK) must be paid by firms. BNP2TKI and Manpower Ministry are supposed to offer at least
some of its services for free because its operation is financed by the government budget.
174
To obtain protection abroad, each migrant worker must pay USD 15 for DP3TKI (Protection,
Supervising and Monitoring Fund) to the treasury. The proceeds are supposedly used by the
government to monitor and protect migrant workers abroad. However, without giving something in
return, government agencies are tantamount to pure rent-seekers. In the case of death, dispute etc that
requires for migrant workers to be sent home, the costs of travel and accommodation are borne by the
Ministry of Foreign Affairs and not directly by either Manpower Ministry or BNP2TKI. Employers
or foreign agency can also send back workers deemed as incompetent workers. In this case all costs
are paid by sending firms.
Table 9 shows the distribution of the value of business across actors in the migrant industry.
For some actors it is hard earned incomes or profits, but for others it may be just pure rents.
Interestingly, the share of value going to sending firms is smaller than that to government agencies.
This can be explained by the existence of two competing authorities in the industry which create
dualism, delay and wasteful costs. The share of value going to sending firms is only slightly higher
than the informal payment. Again this may come from uncertainty in regulations. Lack of
transparency and uncertainty are breeding ground for corruption. So even without including informal
payments, in effect the Government has been taking an increasing share of the profits from sending
firms.24
Table 9: Distribution of value for 432,000 departing Indonesian workers (informal sector)
Actors Amount Share (%) Number of players
(USD Million)
Migrant workers 800 80.3 432,000
Sending firms 60 6.0 500
Government agencies 80 8.0 2
Informal payment 56 5.7 n.a
Total 996 100.0
Sources: authors; calculation based on the field interview
This suggests that migration is a temporary and positive phenomenon for many workers that
provides them with an opportunity to save for a variety of purposes, including starting up a small
business. At present, the investments made by returning migrants may not be enough to absorb all the
surplus rural labor, and so it is important to have a thriving formal sector for employment in
24
It is also worth to comment on insurance. Insurance premium is set up by the government uniformly without regard to
risks associated with the country of destination. Every departing worker is required to pay insurance of IDR 400,000. The
idea actually originated with workers in Middle East. Because insurance is prohibited it is then handled by a consortium of
insurance firms of which workers pay premium in Indonesia. In Taiwan, Hong Kong and Singapore host families pay
insurance for their domestic helper so additional insurance from Indonesia is just a waste of money. The procedure to choose
an insurance consortium has never been transparent. The biggest problem is that some firms in the consortium do not have
representative abroad so when an incident happen the possession of an insurance policy is virtually useless. The case usually
has to be taken care of by the Indonesian embassy or the respective sending firm. When there is a change of a minister of
manpower a new consortium whose members are different usually takes over. So again the insurance policy becomes
worthless. This is just another example of a pure rent-seeker in the industry.
175
neighboring cities and towns. But in the short term, rural people in Indonesia will continue to seek
work abroad. Temporary work abroad will always represent an attractive means for accumulating
savings and experience quickly.
3. CONCLUSION
In this study we have examined Indonesia’s regulatory, institutional and governance structure
for managing international labor migration. The Indonesian migrant industry has evolved from
oligopolistic in the late 1980s and the early 1990s to near monopolistic-competition in 2004. The
most visible impact of this change is the dwindling profits for migrant recruitment firms. As inter-
firm competition has increased, the interplay between actors in the industry has changed. The
Indonesian government has also become more assertive with respect to protecting the human rights of
migrants. Law 39/2004 was launched in 2004 with the purpose of regulating the migrant industry,
formalizing the recruitment process, and providing more protection for migrant workers.
The most important impact of the new law is its formalization of the recruitment process. On
one hand, the law has succeeded in providing some protection and transparency to migrants. On the
other hand, it has also increased the costs of recruitment to migrant firms and eroded the margin of
profits for these firms. The end result has not always been in the best interests of international
migrant workers. For some destinations, particularly in East Asia, workers’ salaries are now deducted
for between 8 and 18 months to cover the sending costs of recruitment firms. This has had an adverse
effect on the net welfare of migrant workers. The creation of a new agency, BNP2TKI, is another
hallmark of Law 39/2004, but the implementation of this law has created a dualism of authority
within the Manpower Ministry. This adds confusion and uncertainty to the industry, adding more
costs to the industry which are ultimately borne only by migrant workers. A clearer division of labor
is needed within the Manpower Ministry, but due to the huge rents involved this may need political
intervention from the highest levels of government.
176
Chapter 8: The Philippine Labor Migration Industry For Health And
Educational Services: Regulatory And Governance Structures
TERESO S. TULLAO, Jr., Ph.D.25,
MITZIE IRENE P. CONCHADA26,
and JOHN PAOLO R. RIVERA27
De La Salle University, Manila, Philippines
ABSTRACT: Analyzing the effects of the temporary migration of Filipino workers on national
development must be viewed in terms of the key factors affecting international migration. These
factors include push and pull forces, labor and demographic asymmetries, and the forces of
globalization. For example, labor shortages in the labor-receiving countries have raised the expected
rates of return to temporary Filipino migration. Although the expected private returns to education
and international migration are high, decisions to go to school and migrate in the Philippines are
usually accompanied by huge social costs. Because individuals do not consider these social costs,
they continue to invest in education and this has led to a large misallocation of resources in the
Philippines. To reduce the social costs of migration in the country, there is a need for the
government to improve deployment procedures for working abroad and to increase the probability of
external employment.
1. OVERVIEW OF PHILIPPINE INTERNATIONAL LABOR MIGRATION
International labor migration began as a temporary employment measure in the Philippines in

the 1970s, but it has now become an almost a permanent fixture in the country affecting the social,
economic and cultural make-up of the entire population. According to the Philippine Overseas
Employment Administration (POEA), there were 1,422,586 registered Filipinos deployed (working)
abroad in 2009. The remittances sent home by these Overseas Filipino Workers (OFWs) have a huge
effect on the economy. According to the Bangko Sentral ng Pilipinas (BSP), remittance inflows to
the Philippines reached over USD 17 billion in 2009. These large financial inflows have stimulated
the economy and helped to improve the well-being of households through increased expenditures on
housing, education, and consumer durables.
It is difficult to measure with any precision the exact number of Filipinos working abroad.
Figures from the POEA only include those people who are working abroad with registered and
official contracts, while the number of Filipinos laboring abroad on an irregular and unofficial basis is
unknown but probably quite high. With these caveats in mind, Table 1 shows the major destinations
of land-based OFWs – Filipinos working abroad with official contracts -- by country over the last ten
years. The table shows that the Middle East represents the leading destination of Filipino migrant
workers. In 2009 the Middle East (including Saudi Arabia, UAE, Kuwait and Qatar) accounted for
72 percent of the total number of OFWs. In the same year Asia (including Hong Kong, Taiwan and
25
Full Professor, School of Economics, De La Salle University (DLSU), 2401 Taft Avenue, 1004 Manila, Philippines. E-
Mail: tereso.tullao@dlsu.edu.ph or tstullao@yahoo.com
26
Assistant Professor, School of Economics. DLSU, 2401 Taft Avenue, 1004 Manila, Philippines. E-Mail:
mitzie.conchada@dlsu.edu.ph or missirene@rocketmail.com
27
Assistant Professor, School of Economics, DLSU, 2401 Taft Avenue, 1004 Manila, Philippines. E-Mail:
john.paolo.rivera@dlsu.edu.ph or johnpaolo_rivera@yahoo.com
177
Table 1: Deployment of Land Based OFWs by Top Continental Destination from 2002 to 2009
(New Hires and Rehires)
Year 2002 2003 2004 2005 2006 2007 2008 2009
Middle East
Saudi
193,157 169,011 188,107 194,350 223,459 238,419 275,933 291,419
Arabia
UAE 50,796 49,164 68,386 82,039 99,212 120,657 193,810 196,815
Kuwait 25,894 26,225 36,591 40,306 47,917 37,080 38,903 45,900
Qatar 11,516 14,344 21,360 31,421 45,795 56,277 84,342 89,290
Total
306,939 285,564 352,314 394,419 462,545 487,878 631,828 669,042
Deployment
Asia
Hong Kong 105,036 84,633 87,254 96,693 96,929 59,169 78,345 100,142
Taiwan 46,371 45,186 45,059 46,737 39,025 37,136 38,546 33,751
Singapore 27,648 24,737 22,198 28,152 28,369 49,431 41,678 54,421
Total
292,077 255,287 266,609 259,209 222,940 218,983 219,598 260,995
Deployment
Europe
Italy 20,034 12,175 23,329 21,267 25,413 17,855 22,623 23,159
UK 13,655 13,598 18,347 16,930 16,926 9,525 9,308 7,071
Total
45,363 37,981 55,116 52,146 59,313 45,613 51,795 47,409
Deployment
Americas
Canada 3,535 4,006 4,453 3,629 6,468 12,380 17,399 17,344
USA 4,058 3,666 3,831 7,752 11,443 9,401 8,050 6,248
Total
11,532 11,049 11,692 14,886 21,976 28,019 31,916 31,146
Deployment
178
Table 1: Deployment of Land Based OFWs by Top Continental Destination from 2002 to 2009
(New Hires and Rehires)
Year 2002 2003 2004 2005 2006 2007 2008 2009
World
Total
891,908 651,938 704,586 740,360 788,070 811,070 974,399 1,092,162
Deployment
Source: Philippine Overseas Employment Administration (POEA)
Singapore) accounted for another 21 percent of OFWs. In 2009 less than 7 percent of OFWs went to
work in Europe and North America.
The large concentration of Filipino migrant workers in the Middle East can be traced to the
oil boom of the 1970s and the more recent massive development projects in that area of the world
(Tullao and Rivera, 2008). In Asia, the sizeable numbers of OFWs can be attributed to changing
demographics and the growing need for domestic helpers, factory workers and professionals in Hong
Kong, Taiwan and Singapore. While much smaller in size, the flow of Filipino workers to North
America can be linked to the historic colonial ties between the United States and the Philippines in
the early 20th Century. In more recent years, the composition of Filipino migrants to North America
has changed with the liberalization of immigration policies in response to the tight labor market for
professionals in that region of the world (Tullao, Cortez and See, 2004).
The Philippines sends workers abroad from a wide range of occupational groups. Table 2
shows that unskilled workers – that is, production workers and service workers -- accounted for 77
percent of all OFWs deployed in 2009. Skilled workers – that is, professional, medical and technical
workers – accounted for another 14 percent of deployed OFWs. With the global demand for skilled
workers rising, there has been a concomitant rise in the demand for education in the Philippines. This
is especially noticeable in the fields of nursing and teaching (Tullao, Cortez and See, 2004).
179
Table 2: Deployment of Newly Hired Overseas Filipino Workers (OFWs) by Skills
Major Occupational
1992 1995 2000 2005 2006 2007 2008 2009
Groups
Administrative and
289 339 284 490 817 1,139 1,516 1,290
Managerial Workers
Agricultural Workers 2,023 981 526 350 807 952 1,354 1,349
Clerical Workers 5,369 3,441 2,367 5,538 7,912 13,662 18,101 15,403
Production Workers 95,062 82,508 57,807 74,802 103,584 121,715 132,295 117,609
Professional, Medical,
Technical and Related 72,230 43,629 78,685 63,941 41,258 43,225 49,649 47,886
Workers
Sales Workers 2,701 1,990 2,083 4,261 5,517 7,942 11,525 8,348
Service Workers 82,267 81,028 91,206 133,907 144,321 107,135 123,332 138,222
Others 0 0 7,662 996 3,906 10,613 494 1,645
Total 259,941 213,916 240,620 284,285 308,122 306,383 338,266 331,752
Over the years, the educational qualifications of Filipino migrant workers have been rising.
According to a study by Alburo and Abella (2002), 31.2 percent of OFWs are high school graduates
while 43.8 percent have completed college degrees. According to Tullao (2008), even production
workers are now required to have higher education and skills. College undergraduates and those with
lower educational attainment can still manage to find jobs abroad, but often they have to settle for
lower paying positions as domestic helpers, factory workers, and construction workers (Tullao and
Rivera, 2008).
With regards to the deployment of newly hired OFWs, Table 3 shows that household service
workers still represent the largest occupational group for 2009. In that year, household service
workers accounted for 21 percent of all newly hired overseas workers from the Philippines. In 2009
professional nurses and caregivers accounted for only 7 percent of newly hired OFWs, but the
number of workers in these categories is rising.
180
Table 3: Deployment of Newly Hired Overseas Filipino Workers (OFWs)
by Major Occupation Groups
Occupational Group 2008 2009
Household Service Workers 50,082 71,557
Waiters, Bartenders, and Related Workers 13,911 11,977
Char workers, Cleaners, and Related Workers 11,620 10,056
Professional Nurses 11,495 13,465
Caregivers and Caretakers 10,109 9,228
Laborers / Helpers General 9,711 8,099
Plumbers and Pipe Fitters 9,664 7,722
Wiremen Electrical 8,893 9,752
Welders and Flame-Cutters 6,777 5,910
Caretakers Building 6,610 5,127
Other Skills 199,394 178,859
Total Deployment of New Hires 338,266 331,752
Source: Philippine Overseas Employment Administration
Over the years international labor migration has created a culture of migration in Philippine
society. The outflow of Filipino labor has encouraged neighbors, relatives, and friends to seek
employment abroad. This culture of migration has also created some perverse effects on the internal
labor market by increasing the reservation wage of the working-age members of households receiving
remittance income. The culture of migration has also affected the type of higher educational programs
that households desire to pursue. In many cases, households with migrants prefer to take courses in
disciplines where the probability of migration is high.
2. MANAGING TEMPORARY LABOR MIGRATION
The 1970s ushered in an increase in demand for temporary migrant workers, spurred by the
economic boom in the Middle East. The Philippine government immediately took advantage of these
opportunities by adopting the Labor Code of the Philippines (LCP) in 1974. The LCP provided the
needed legislative, political and administrative infrastructure for the systematic deployment of
overseas workers (Agunias, 2008). It also created a large bureaucracy composed of three institutions,
namely the Overseas Employment Development Board (OEDB), the National Seaman Board (NSB),
181
and the Bureau of Employment Services (BES). The first two institutions were responsible for
developing the market for overseas workers, recruiting workers, and securing the best possible
employment terms for land-based and sea-based workers. The third institution (BES) was charged
with regulating private recruitment agencies and functioned as a temporary government-run
employment agency (Agunias, 2008). Currently, these three institutions have now been merged into
the POEA.
The POEA, an attached agency of the Department of Labor and Employment (DOLE), has
the authority to manage temporary overseas employment, including the regulation of private
recruitment agencies. In managing the temporary migration of workers, the POEA coordinates with
three other agencies within DOLE: (a) the National Labor Relations Commission (NLRC), which is a
quasi-judicial body that adjudicates on compensation claims; (b) the Technical Education and Skills
Development Authority (TESDA), which is an agency in charge of developing labor skills and
providing technical training for Filipino workers and migrants; and (c) the Overseas Workers Welfare
Administration (OWWA), which is a financial agency that manages a welfare fund for migrant
workers.
In the Philippines most recruitment of people to work abroad is done by private recruitment
and manning agencies. According to Aguinas (2008), private agencies handled the deployment of 94
percent of all overseas workers in 2007, while 5 percent of workers were directly hired and 1 percent
was government hired.
In order to regulate the participation of these private recruitment agencies, POEA created
strict entry rules. All recruitment and manning agencies, foreign employers, and foreign governments
have to meet a set of standards before participating in overseas deployment including nationality and
financial requirements.
The recruitment industry in the Philippines is composed of two parts: sea-based and land-
based sectors. According to Soriano (2009), there are around 350 recruitment agencies dealing with
sea-based OFWs, which are responsible for placing approximately 270,000 ratings and cruise line
service personnel and 10,000 officers annually. This yields an average of 800 placements per agency.
On the other hand, the number of land based recruitment agencies is approximately 1,010 and they
place some 350,000 OFWs annually. Land-based agencies are more complicated than sea-based
agencies because they deal with disparate countries, laws, regulations, business practices,
occupations, and employers. Also, land-based agencies are subject to Philippine laws and regulations
administered by POEA.
Soriano (2009) has identified some of the characteristics of land-based agencies. First, many
of these agencies do not charge any fees from accepted applicants since they are paid by their foreign
employer clients. However, some of the smaller agencies do charge applicants the legal fee
equivalent of one month’s salary. Second, some land-based agencies are nominally Filipino owned
but are actually foreign owned that charge fees to successful applicants. According to Soriano (2009),
one of the reasons behind the exorbitant fees is the presence of overseas “brokers or agents.”
With respect to the fees collected by recruitment agencies, Soriano (2009) pointed out that
some agencies have a salary fee deduction program for successful applicants while other agencies
require that fees be paid in full prior to migration. To ensure payment, recruitment agencies seek
funds from sources which are sometimes from unregulated lending companies. In some cases, the
foreign employer pays the fees of the recruitment agency, especially for professionals and skilled
workers.
182
Despite nationality and capital requirement barriers, recruitment agencies seem to be a
lucrative type of business as evidenced by the increasing number of illegal and irregular recruiting
agencies. These illegal agencies often appear to be foreign owned firms that are not in compliance
with existing rules and regulations (Soriano, 2009). The presence of illegal recruitment agencies may
be attributed to the prospects of earning handsome profits from such things as charging service fees
for processing an application and/or collecting placement fees after an applicant has been successfully
accepted.
The POEA is charged with regulating the growing number of recruitment agencies in order to
avoid cut-throat competition. However, the government has never stated the number of recruiting
agencies that it desires. While competition is necessary in a private-sector-driven society, too much
competition between agencies can increase the probability of worker abuse. Policymakers fear that in
an overcrowded market, some agencies will not be profitable and will therefore try to cover their
losses by charging exorbitant recruitment fees and/or colluding with employers (Agunias, 2008).
Meanwhile, foreign employers hiring through private agencies have to meet certain entry
requirements. Agunias (2008) highlighted that foreign employers are required to undergo an
accreditation procedure before they can hire Filipino workers. The employer has to identify an agency
that would serve as its representative in the Philippines. Moreover, the employer is required to draft a
master employment contract that satisfies POEA’s minimum standards. The employment contract
must include the number of workers needed, job descriptions, and the salary of each position. In the
accreditation procedure, foreign employers are required to work with the respective recruitment
agencies to submit to POEA valid proof of the business or projects and the documents showing that
the necessary visas are available.
Likewise, POEA must ensure that workers deployed abroad are technically and medically
qualified. All workers should possess a prescribed level of technical qualification and physical,
medical, and psychological fitness. Health requirements vary by destination and type of occupation.
Technical competencies are determined at testing centers accredited by TESDA, while physical and
medical fitness are determined at medical clinics and facilities accredited by the Department of
Health (DOH). In addition to TESDA’s role, the Commission on Higher Education (CHED) and the
Professional Regulation Commission (PRC) work together toward skills standardization, assessment,
and certification. CHED supervises, monitors, and regulates degree programs while PRC administers
professional examinations for these programs (Agunias, 2008).
3. DEMAND FOR EDUCATIONAL SERVICES
Determinants of Demand for Educational Services
As more educated Filipino workers are deployed overseas, the demand for education,
particularly higher education, becomes significant in the management of temporary labor migration.
A number of studies have examined the factors that affect the demand for education including
employability, domestic economic progress, availability of credits and income, and rate of return
(Lillard and Willis (1994), Biblarz and Raftery (1999), Binder and Woodruff (1999), and Borromeo,
Castiilo and Lopez (2007)). In addition, the studies of Hauser and Daymont (1977), Plug and
Vijverberg (2001) have examined how the financial capability of households affects the demand for
higher education.
Demand for Higher Education and Migration
183
A recent study by Tullao and Rivera (2008) found that remittance income generated by
OFWs in the Philippines is more likely to be invested in higher education, rather than in basic
education, because of the limited public provision of higher education. In their study, Tullao and
Rivera (2008) estimated a demand function for higher education that includes the variable average
remittances. Results showed that average remittances have a positive and a statistically significant
effect on the demand for higher education in the Philippines, as measured by the enrollment level in
such academic programs as accounting, business, education, engineering, and nursing. Similarly, De
Guzman, De Vera and Layno (2009) have highlighted the importance of remittances in the course
selection decision of a student in the Philippines, specifically the oldest child in a household with an
OFW member. Results suggest that remittances are statistically significant in influencing the
probability that a student will take higher education, specifically health related courses. The attractive
compensation packages for Filipino migrants in the fields of health, nursing and care giving create
powerful incentives for students to enter these fields.
The strong relationship between overseas employment and the demand for education has
major implications for various aspects of human resource development (HRD) in the Philippines.
According to Tullao and Rivera (2008), one of the key issues here is the ability of educational
institutions and students to respond to global demand. Given the weaknesses of the system of higher
education in the Philippines, higher education institutions (HEIs) may not have the capacity to
adequately prepare an increasing number of students, and this will in turn affect their chances of
getting high paying jobs in the overseas market. For this reason, it is important that the country
continue to address problems relating to both the quality and the quantity of academic programs,
particularly in the area of health services.
Demand for Technical and Vocational Educational Services and Migration
In recent years the number of migrating skilled professionals has been increasing in the
Philippines, and the government sees nothing wrong with this process. According to Manalansan
(2003), “the labor and health departments are even instituting mechanisms to enhance the country’s
capability to send “globally-competitive” professionals abroad.” Within the Philippines, CHED is
charged with ensuring that nursing graduates meet international standards. To ensure that
professionals – especially nurses and caregivers – are competitive abroad, TESDA is mandated to
accredit and monitor technical and vocational schools. To make sure that the country meets the
standards of foreign employers, TESDA has established partnerships with developed countries (like
Canada) that seek to raise the quality of education of Filipino professionals through certification
measures. Certification is a powerful and useful tool for Filipinos seeking to work abroad because
through certification they can show that they possess the skills needed to compete in the global labor
market.
184
Demand for Nursing Education
Because of the high compensation packages for nurses and health-care professionals abroad28,
there is a huge demand in the Philippines to pursue higher education in nursing and health-related
fields. According to the Association of Deans of Philippine Colleges of Nursing (ADPCN), the
estimated monthly salaries of Filipino nurses in various destination countries are as follows. In the
UK, nurses are paid an estimated USD 3,000.00 per month. In USA, the monthly salary of nurses is
estimated at USD 4,000.00 to USD 6,000.00. In Saudi Arabia, nurses receive USD 700.00 to USD
1,500.00 per month. In Singapore, nurses are paid USD 1,100.00 on the average. These salaries paid
abroad are huge when compared to how much nurses are paid in the Philippines, which is estimated at
USD 180.00 to USD 220.00 per month.
In the Philippines, the increasing demand for nursing education has led to a large expansion
in the number of nursing schools: from 40 in the 1970s to 269 in 2005 (Tiongson, 2008). According
to CHED, as of academic year 2008 to 2009, there were already 436 HEIs offering nursing courses.
To ensure that these schools meet certain minimum quality standards, the CHED has established a
moratorium on the establishment of new nursing colleges. According to the Federation of
Accrediting Agencies of the Philippines (FAAP), out of the 436 recognized nursing schools, only 54
are accredited by accrediting agencies as of 2008. Furthermore, out of the 54 accredited schools, only
8 nursing schools have Level III accreditation status, 38 have Level II accreditation status, and 8 have
Level I accreditation status.
Table 4 shows that the growth in nursing schools has led to a concomitant rise in the number
of nursing graduates in the Philippines over the last ten years. Much of this increase has been due to
the rising demand for nurses overseas. However, it should be noted that none of these nursing
graduates possess any assurance of getting employed abroad unless they can pass the licensure
examinations in the Philippines and abroad.
Table 5 shows the passing rate for the Philippine National Licensure Examination (NLE) for
nurses in recent years. The passing rate for the nursing board has declined from 55.8 percent in 1998
to 43.1 percent in 2008. In addition, according to Paquiz (2009), 1 out of every 5 nursing schools
registers a zero passing mark which is indicative of the questionable quality of education in many
nursing schools.
What is revealing among the examinees of the NLE is the presence of a large number of
“second-coursers”. According to PRC, aside from the regular nursing graduates, there were also
several categories of courses or degrees of second-coursers who took the NLE in 2004 to 2005. Out
of 8,603 examinees that were second-coursers, 6,477 passed. The largest proportion, 34 percent or
2,206, of those who passed have a Doctor of Medicine as their first degree. On the other hand, 798 or
12.32 percent are graduates of Bachelor of Science (BS) in Physical Therapy. Likewise, 381 or 5.88
percent are graduates of BS in Medical Technology. This suggests that even doctors and other non-
28
It is important to note that the figure for the salary of nurses outside the Philippines must be valued according to the value
of a dollar of income earned abroad with the dollar equivalent earned in the Philippines. Since the cost of living abroad,
especially in developed countries, is substantially higher and for many Filipinos, there are negative cultural implications in
being abroad rather than in the Philippines, foreign income must be adjusted in accordance the concept of purchasing power
parity (PPP).
185
Table 4: Nursing Enrolment and Graduates in the Philippines
Academic Year Nursing Enrolment Nursing Graduates
1998 – 1999 25,256 11,097
1999 – 2000 25,388 8,097
2000 – 2001 27,142 5,483
2001 – 2002 49,995 4,409
2002 – 2003 92,106 5,425
2003 – 2004 178,626 8,477
2004 – 2005 263,387 13,120
2005 – 2006 248,652* 13,933*
2006 – 2007 153,479* 10,329*
2007 – 2008 87,242* 9,021*
2008 – 2009
* Estimated by the authors using Autoregressive Moving Average (ARMA) Methodology with data
generating process ARMA (1,1)
Source: Commission on Higher Education (CHED)
186
Table 5: Passing Rate for Philippine Nursing Board Exam
Year Number of Examinees Number of Passers Passing Rate
(%)
1991 - - 66.0
1992 - - 58.4
1993 - - 63.8
1994 - - 62.2
1995 - - 57.1
1996 - - 54.2
1997 19,552 9,776 50.0
1998 17,099 9,541 55.8
1999 - - 50.4
2000 - - 49.6
2001 - - 53.6
2002 - - 44.8
2003 15,611 7,582 48.2
2004 25,221 12,581 49.9
2005 49,676 25,951 52.2
2006 82,153 37,533 45.7
2007 132,637 60,928 45.9
2008 64,459* 27,765* 43.1*
*Covers first half of year only
Source: Osteria (2006) for 1991 to 1996; Commission on Higher Education (CHED) for 1997 to 2002;
Professional Regulation Commission (PRC) from 2003 to 2008.
187
nursing health professionals go back to school to study nursing, take the board examinations, and
explore the opportunities of working abroad as nurses.
Aside from the passing the Philippine NLE, Filipino licensed nurses must also pass foreign
licensure examinations. Entry into the practice of nursing abroad, especially in USA and its
territories, is strictly regulated. To ensure public protection, each jurisdiction requires a candidate for
licensure to pass an examination that measures the competencies needed to perform safely as a newly
licensed, entry-level nurse. The National Council of State Boards of Nursing (NCSBN) develops two
licensure examinations, the National Council Licensure Examination for Registered Nurses (NCLEX-
RN) and the National Council Licensure Examination for Practical Nurses (NCLEX-PN) that are used
by state and territorial boards of nursing to assist in making licensure decisions. Likewise, there is
also the Commission on Graduates of Foreign Nursing Schools (CGFNS), which is an immigration-
neutral non-profit organization, internationally recognized as an authority on credentials evaluation
pertaining to the education, registration, and licensure of nurses and other health care professionals
worldwide.
According to Osteria (2006), nurses who wish to work in USA and obtain an immigrant visa
have to pass several tests. The initial procedure is filing the I-140, proof of CGFNS score of 400 or a
nursing license in the preferred state of destination. A nursing license in all states requires passing the
NCLEX examination. Many states also require passing the CGFNS exam by all foreign nurse
applicants. According to the NCSBN, 58.5 percent or 8,376 out of 14,312 first-time Philippine-
educated candidates who took the NCLEX-RN passed in 2006. On the other hand, according to the
CGFNS Philippines Test Data Analysis, out of 11,103 Filipino nurses who took the CGFNS exam in
2005, 4,685 or 42.2 percent passed. Moreover, the average passing rate for Philippine educated
applicants taking the CGFNS examination between 1978 and 2005 was 39.2 percent.
The pressure of passing local and international licensure examinations for nurses has led to a
boom in review centers such as the Millennium-Professional Review Network (M-PRN). These
review centers provide excellent review programs and prepare nursing graduates to become
emotionally and psychologically competent to pass the various regulatory tests. From the
systematized process of personalized predictor tests administration to psychological preparation,
nursing students refresh the skills and theories they have learned back in school.
The Cost of Nursing Education
The demand for nursing courses is significantly affected by the financial resources of the
household. Nursing courses are typically expensive, with tuition fees ranging from PHP 55,000.00 to
PHP 85,000.00 per semester in Metro Manila (De Guzman, De Vera and Layno, 2009) Additionally,
the indirect costs of education such as textbooks, miscellaneous fees, medical diagnostic sets, and
other tools add up to their educational expenditures (Marquez, 2005). In this analysis, aside from
tuition fees and fees mentioned above, the cost of nursing education and migration includes
transportation expenses, review fees, placement and processing fees, travel cost and opportunity cost.
188
The Rate of Return of Investment in Nursing Education
One of the leading reasons for the huge demand for nursing programs in the Philippines is the
high rate of return to nursing education. The rate of return for investing in nursing education is the
rate of discount at which the expected present value of all the cost incurred by a typical nursing
student in order to acquire rights to be employed as a professional nurse abroad will be equal to the
expected present value of all the benefits and earnings that this individual will receive as a
professional nurse abroad as seen in Equation 1:
T
Bt T
Ct
t  0 (1  r )
t
 
t  0 (1  r )
t (1)
1
where r is the rate of return; is the present value interest factor; Bt is the benefits received by
(1  r) t
a typical professional nurse at time t; Ct is the relevant costs incurred by a typical nursing student at
time t; and T is the length of time until a typical nurse reaches retirement. Note that inflation together
with the intricate decision making process of individuals are held constant to simplify the analysis. It
is generally assumed that the nursing aspirant will follow a timely pacing of career development. The
timeframe will start at the first year of the nursing program, which has duration of 4 years.
Assuming that the typical nursing student will graduate on time, the fifth year will be
dedicated for review and registration for the Philippine NLE. Furthermore, assuming that the typical
nursing graduate will pass the NLE, the sixth and seventh year will be devoted to a two-year hospital
experience as required by international nursing standards. Additionally, on the sixth year, the typical
NLE board passer will register for the NCLEX and/or CGFNS, to be taken on the seventh year, to
work abroad as a professional nurse. Likewise, assuming that this typical NLE board passer passed
the NCLEX and/or CGFNS, and then he/she will be eligible to work abroad as a professional nurse.
Hence, this typical professional nurse will spend for placement fees, processing fees, and travel costs
in order to work abroad as soon as he/she finishes the two-year hospital experience by the end of the
seventh year. However, it must be noted that this typical nurse is already earning the prevailing salary
in the Philippines for the sixth and seventh year. Further assuming that this typical nursing migrant
will be employed and will start working as a professional nurse abroad on the eighth year, he/she will
begin to recover the costs incurred before working abroad by earning the prevailing salary for
professional nurses abroad. However, there is still a cost involved, which is the opportunity cost of
working abroad measured by the prevailing salary for professional nurses in the Philippines.
However, this study is particularly interested with the incidence of failing the NLE. Although
Paquiz (2009) mentioned that there is really no limit with the number of times the NLE can be taken,
this will prolong the length of time before a professional nurse recovers the cost of investment in
nursing education and profession. Yet, a typical laborer is only given up to age 60 to work and earn.
Hence, delaying the length of time before a typical professional nurse can start working, specifically
abroad, due to failures in licensure and accrediting examinations will decrease the rate of return to
investment in the nursing profession.
In this study we have considered the incidence of failing the NLE up to three times only.
Table 6 shows the rate of return for different scenarios of nursing education and migration
incorporating the possibility of failing the NLE. The best-case scenario gives an estimated 17 percent
rate of return. Table 6 shows that any form of delay in working abroad as a nurse tends to lower this
best-case rate of return. In addition to failing the NLE, working abroad can be delayed by failures in
189
the nursing coursework, filing leave of absences in school due to personal matters, failing the NCLEX
and/or CGFNS, and failing to acquire domestic hospital experience.
In Table 6 the worst-case possible rate of return for nursing education and migration is
defined as failing the NLE thrice, failing the NCLEX and CGFNS twice, and obtaining hospital
experience two years after passing the NLE. Under these circumstances, the rate of return to nursing
education and migration is reduced to 11 percent for ages up to 60 years and to 8.7 percent for ages
up to 40 years. Hence, as far as nursing education is concerned, even if an individual delays the
earning period, there is still a high rate of return from investing in nursing given the opportunity to
work overseas.
Table 6: Rate of Return for Investment in Nursing Education and Migration
Rate of Return Rate of Return

State of the World
(%) up to Age 40 (%) up to Age 60
Successfully finished the 4-year nursing program;

Successfully passed the NLE; Successfully passed
1 16.35 17.11
NCLEX and/or CGFNS; Successfully acquired 2-year
hospital experience; Successfully employed abroad.

Failed the NLE once; Successfully passed NCLEX
2 14.44 15.44
and/or CGFNS; Successfully acquired 2-year hospital
experience; Successfully employed abroad.

Failed the NLE twice; Successfully passed NCLEX
3 12.82 14.10

Failed the NLE thrice; Successfully passed NCLEX
4 11.41 12.98

Failed the NLE thrice; Failed the NCLEX and/or
5 CGFNS twice; Successfully acquired 2-year hospital 8.70 11.00
experience two years after passing the NLE;
Successfully employed abroad.
190
Demand for Nursing Position in Local and International Hospitals
A major roadblock in the migration of Filipino nurses is the difficulty of obtaining hospital
experience in the Philippines. This is due to the limited number of positions in private and public
hospitals as well as the huge supply of nurses in the country. According to Paquiz (2009), during the
period 1998 to 2007 the supply of nurses in the Philippines was estimated at 173,536, while the
demand for experienced nurses reached 58,000 in the local market and 111,766 in the international
market. In other words, in 2007 there was an estimated surplus of 3,770 licensed nurses.
Furthermore, with the addition of 27,765 board passers in June 2008, the surplus reached 31,535
licensed nurses in 2008.
The estimated surplus of licensed nurses is a problem because of the insufficient number of
positions for these nurses in local hospitals. Not all Filipino nurses can go to work abroad, and those
who do go to work abroad tend to be the most experienced (Table 7). Although the migration of
experienced nurses abroad should open positions for new nurses to acquire hospital experience, the
number of vacant nursing positions is still not enough to accommodate the remaining supply of
nurses. Since the nursing industry wants to accommodate everyone, this means that those who cannot
be accommodated must wait in queue.
Table 7: Annual Deployment of Professional Filipino Nurses
Year Number of Nurses Year Number of Nurses
1992 5,747 2001 13,536
1993 6,744 2002 11,867
1994 6,699 2003 8,968
1995 7,584 2004 8,611
1996 4,734 2005 10,718
1997 4,242 2006 13,525
1998 4,591 2007 9,178
1999 5,413 2008 12,618
2000 7,683 2009 13,465
191
The queuing time before nurses are deployed for work is becoming a problem because the
number of those in the queue keeps rising. While waiting, nurses are doing work not at all related to
nursing in order to survive. For instance, they are working in call centres, spas, banks and department
stores (Paquiz, 2009). Moreover, the problem does not end here. Not all of those nurses who are
deployable to the international labor market are being deployed. This leaves the Philippines with a
large number of licensed and experienced nurses who are either unemployed or underemployed. For
this reason, there is a need to address the growing problem of inadequate work for trained nurses in
the Philippines.
Brain Drain in the Health Sector
According to PRC, the Philippines has 483,332 ever registered nurses as of 2008. This
number is based on the cumulative NLE passers per year, less the number of retired nurses. Because
of the huge global demand for licensed nurses and caregivers, many of these nurses in the Philippines
have migrated to a wide number of developed countries (Table 8). Aside from increasing the training
costs of inexperienced nurses, this massive outflow of nurses could threaten the viability and
productivity of the health care sector in the Philippines. Likewise, the health care sector in the country
may face an impending crisis due to the heavy migration of its most trained health professionals. For
these reasons, the social costs of educating the nursing needs of the rest of the world need to be
quantified. These costs could be huge considering that public funds were used to educate the brightest
students in the Philippines and their exit from the country may drain the country’s human resources.
As the more experienced nurses migrate, the quality of health services in the Philippines may
suffer. For example, a recent report by the Alliance of Health Workers (AHW) reveals that over the
last two years, 17 percent of nurses in 11 hospitals went to work abroad, This suggests that that there
will come a time when “operating rooms are staffed with novice nurses, while the more experienced
nurses will work double shifts” (Manalansan, 2003). Since nurses cannot be prevented from seeking
employment overseas, the sectors that carry the burden of external migration need to be compensated
in order to arrest any possible deterioration in health services in the Philippines.
As noted above, there are many nurses and other medically trained workers who do not
qualify for overseas work because of a lack of hospital ward and practical experience. According to
Soriano (2009), “many of the new nursing graduates wind up paying the hospital for a job in a
desperate attempt to get the necessary ward experience.” This circumstance is a global rarity and
reflects badly on the country’s health system. Consequently, client countries are becoming
increasingly wary of the health education in the Philippines.
192
Table 8: Destinations of Deployed Professional Filipino Nurses
Country 2001 2002 2003 2004 2005 2006 2007 2008 2009
Australia 0 2 15 8 1 8 - - -
Bahrain 7 57 21 43 34 62 - - -
Canada 7 51 25 14 21 7 19 527 346
Ireland 1,529 915 207 190 297 248 127 35 -
Jordan 36 0 38 120 2 38 - - -
Kuwait 182 108 51 408 191 340 393 458 423
Libya 9 411 52 10 23 158 66 104 276
Qatar 143 213 242 318 133 140 214 245 133
K.S.A 5,045 5,688 5,740 5,640 4,627 5,640 6,633 8,848 9,965
Singapore 413 334 326 166 129 73 276 667 745
Taiwan 9 129 200 5 2 2 174 231 202
U.A.E 243 405 226 218 670 768 616 435 572
United Kingdom 5,383 3,089 1,544 800 546 145 38 28 165
U.S.A 304 316 196 373 3,853 202 186 649 242
Rest of the World 226 149 85 298 189 5,694 521 1,022 396
TOTAL 13,536 11,867 8,968 8,611 10,718 13,525 9,178 12,618 13,465
Distortions in the Demand for Education and Training
The possibility of reaping the large wage differentials across national boundaries through
migration has encouraged thousands of Filipinos to invest in nursing education. The surge in the
number of nursing schools in the Philippines may be viewed as a supply response to this
phenomenon, but it has also created a number of distortions in the educational sector. According to
Ronda (2008), the main problem is the proliferation of schools offering poor quality nursing
education. According to a Commission on Audit (COA) report from 2001 to 2005, only 42.2 percent
of the nursing schools across the Philippines managed to have at least 50 percent of their graduates
pass the PRC licensure exams, with 7.2 percent of these schools failing to have even a single passer.
193
Filipino students demand training and education in nursing because of the high expected
private rates of return to education and migration. However, these private decisions are usually
accompanied by huge social costs. Because individuals do not consider these social costs, they
continue to invest in nursing and health education even if there is a risk of failing the licensure exams
one or several times. Nursing students will continue to take these exams until they are to migrate and
reap the returns on their investments. This process has created distortions in terms of a misallocation
of resources and has created a large pool of unemployed and underemployed nurses.
The expected private rate of return to nursing education was estimated at between 11 and 17
percent (Table 6, page 191), which is very high given the prevailing rates of return in other academic
programs. These high private rates of return continue to attract people and resources to nursing
education. However, the surge of enrollment in nursing programs has gone beyond the optimal level
of investment and new graduates of these programs are unlikely to enjoy such high rates of return
given the high failure rates in local and international licensure examinations, the limited capacity of
domestic hospitals to train new nurses, and the low probability of overseas placement.
The Philippine higher education system is characterized by high attendance rates showing
that there are private rates of return to education similar to those in developed countries. According to
Joshi (2004), in the Philippines the private rate of return to higher education has been around 12
percent while the social rate of return to higher education has been around 11 percent. These figures
are consistent with the rates of return estimated by Psacharopoulos (1994) and Psacharopoulos and
Patrinos (2002), who found in developed countries that the private rates of return to education were
11.6 and the social rates of return were 10.5 percent.
In the Philippines the thousands of nursing graduates who are eager to pass the licensure
examinations has created a highly profitable business climate for the establishment of review centers
that coach students to pass these exams. The market is huge, competition stiff and returns so lucrative
that these review centers are willing to invest thousands of dollars in various forms of advertisement
to attract potential clients. Instead of resources being used in teacher training and graduate education
with high social rates of return, huge sums are being spent on advertising to promote the performance
of various licensure review centers.
The high private return to nursing education and migration has likewise caused a staff
substitution among health professionals that threatens the quality of health service in the country in
the medium and longer term. Many doctors, dentists, pharmacists and other health professionals are
taking up nursing as a second degree to increase their probability of migration and overseas
employment. This shift among health professionals as well as the exit of highly experienced nurses
may lower the quality of hospital care in the Philippines.
Moreover, hospital care may also be at risk since many Filipino hospitals have become de
facto training grounds for nurses seeking overseas employment. Since hospitals abroad require 2 to 3
years of home country hospital experience, there is a strong demand for placements in Philippine
hospitals. This environment has created an incentive system where local hospitals have no inclination
to improve the compensation of nurses since these nurses will tend to migrate after their 2 to 3 years
stint with the hospitals. As a result, there is a fast turn-over of nurses in domestic hospitals and this
may have serious consequences on the quality of health and hospital care.
Another manifestation of distortion in resources is the creation of a large pool of unemployed

and underemployed nurses. There are uncertainties and risks involved in nursing. Specifically, not all
graduates of nursing programs are able to pass the national licensure exams. For example, only 43
percent passed from among the 64,459 examinees during the first licensure examination in 2008. The
194
wastage in resources does not end in the large failure rate in the national examinations. Not all nurses
who have passed the local licensure examination and the NCLEX are able to get the 2 to 3 year work
experience in public and private hospitals. According to key informants, only around 22 percent of
the new graduates are absorbed by the local hospitals. Others wait for their turn in a long queue which
is becoming longer over time.
Although the risks and uncertainties in the nursing field are borne by individuals, the costs of
these risks and uncertainties end up in the creation of a large pool of unemployed and underemployed
nurses. This ultimately becomes a major social burden. However, there are factors that can reduce
this social burden and the present massive investment in nursing education and training in the
Philippines. For example, it is possible that a narrowing of the income gap between labor sending and
receiving countries can somehow reduce the link between nursing and migration. It is also possible
that government efforts to tax nursing and subsidize other educational programs can reduce the
attractiveness of the present massive private investment in nursing. There is a need to reduce the
attractiveness of nursing education so that social costs are reduced and benefits in other fields are
realized (Tullao, 1982).
4. CONCLUSION AND POLICY RECOMMENDATIONS
Given the high private rates of return to international labor migration, and the highly
globalized labor market, the optimal strategy for the Philippines government is to manage temporary
labor migration so that benefits are enhanced and social costs mitigated. In terms of enhancing the
positive contributions of migration, remittances can be channeled to more productive uses particularly
in tempering its impact on the overvaluation of the real exchange rate. The government can also take
steps to augment the supply of sectors producing human capital, specifically in the area of expanding
and improving higher education in the health sector. Through these supply enhancing activities the
price effect of the appreciation of the real exchange rate due to remittance inflows may be eased.
Although private returns to education and migration are high in the Philippines, decisions to
go to school and migrate are usually accompanied by huge social costs. Because individuals often do
not consider these social costs, they continue to invest in migration because the expected private
internal rates of return are still quite high. As a result, thousands of students take on a multitude of
risks until they are able to migrate and to reap the returns on their investments. This process, while
good for the individual, is costly to society, because it has created distortions in terms of a
misallocation of resources and has created a large pool of unemployed and underemployed.
To mitigate the social costs of migration, the government needs to improve deployment
procedures for working abroad and to enhance the protection mechanisms of workers already abroad.
Since the processing fee for migration abroad is too high, this fee should be paid through various
financing mechanisms over time. Given the unscrupulous actions taken by many unlicensed recruiting
agencies, there is a need for POEA to demonstrate more caution in licensing new recruitment
agencies. Also, given the uncertainties in passing licensure examinations and in gaining domestic
work experience, there is a need for POEA to provide more realistic information to potential migrants
regarding the difficulties of securing overseas work.
To address the distortions created by the outflow of nurses, there is a need to internalize the
cost of education and migration. To a great extent, the underemployment and unemployment of
nurses in the Philippines has been a mechanism for internalizing the cost of investment in education
and migration. However, the widespread underemployment and unemployment of thousands of
nurses in the Philippines makes under-utilization of these trained people a huge social cost. To reduce
these social costs, the government needs to pursue a variety of regulatory and development measures.
195
On the regulatory side, the government can implement a moratorium on the establishment of new
nursing schools and programs. In addition, the government can implement a strict accreditation
mechanism for nursing schools to ensure quality programs. On the development side, CHED can
revise nursing curriculums to reflect domestic and foreign employment opportunities for nurses
including the self-employment of nursing graduates and licensed nurses.
Protecting OFWs at destination countries can be enhanced by the proper government
implementation of employment contracts. This can be done by explicitly stating this objective in the
various bilateral labor agreements that the Philippines government signs with other countries. It is
also important for the government to actively monitor the welfare of overseas Filipino workers.
In terms of policy reforms, the inability of the Philippines government to reduce the outflow
of Filipino migrant workers may lead to a renewed focus on the supply dimensions of migration
rather than its demand aspects. One way of reducing the outflow of migrants would be to lessen the
wage rate differentials between the Philippines and other countries by pursing rapid economic growth
in the Philippines. Another way of reducing the outflow of migrants would be to promote a more
enlightened population program in the Philippines that could moderate increases in the supply of
labor. If these policies could be effectively implemented over time, temporary labor migration would
become more of a choice rather than a necessity for many Filipino workers.
196
REFERENCES
Agunias, D. R. (2008). Managing temporary labor migration: Lessons from the Philippine model.
Program on Migrants, Migration and Development. Migration Policy Institute
Alburo, F. & D. Abella. (2002). Skilled labor migration from developing countries: Study on
the Philippines. International Migration Paper 51. International Labor Office (ILO).
Biblarz, T. & A.E. Raftery. (1999). Family structure, educational attainment and socioeconomic
success: Rethinking the pathology of matriarchy. American Journal of Sociology, 105(2), 321-365.
Binder M. & M. Woodruff. (1999). Intergenerational mobility in educational attainment in Mexico.

SSRN Working Paper Series 166388, Social Science Research Network.
Borromeo, M.R.V., J.C.E.D. Castillo & E.N. Lopez. (2007). Selected demographic variables
affecting the educational attainment of children in the Philippines. Undergraduate Economics Thesis.
De La Salle University.
De Guzman, E.M.N., E.N.G. De Vera & I.C.P. Layno. (2009). Deciding their destinies: The impact of
remittances and other socio-economic determinants on course selection decision of the eldest sibling
in Pasay and Eastern Samar, Philippines. Undergraduate Economics Thesis. De La Salle University.
Hauser, R.M. & T.N. Daymont. (1977). Schooling, ability, and earnings: Cross-sectional findings 8 to
14 years after high school graduation. Sociology of Education, 50(3), 182-206.
Joshi, K.M. (2004). An exploration of private sector financing of higher education in the Philippines
and its policy implications for India. Retrieved from
www.asianscholarship.org/asf/ejourn/articles/Kishore%20Joshi2.pdf
Lillard, L. & R. Willis. (1994). The family and intergenerational relations. The Journal of Human
Resources Special Issue, 29(4), 1126-1166.
Manalansan, E. (2003). Brain drain refrain. Retrieved from http://bulatlat.com/news/2-50/2-50-

braindrain.html
Marquez, R.C.R. (2005). Medical schools rake in profits but health system is in crisis. Retrieved from
http://www.bulatlat.com/news/4-51/4-51-health.html.
Osteria, T.S. (2006). Policy context of Filipino nurses’ migration to the United States and Canada.
Yuchengco Center Occasional Paper No. 4. De La Salle University.
Paquiz, L.P.G. (2009). Why ang nars? Manuscript.

Philippine Overseas Employment Agency (POEA). (2007). POEA Annual Report.
Plug, E. & W. Vijverberg. (2001). Schooling, family background, and adoption: Does family income
matter? IZA Discussion Paper No. 246. Bonn, Germany: Institute for the Study of Labor.
Psacharopoulos, G. (1994). Returns to investment in education: A global update. World Development,

22(9),1325-1343.
Psacharopoulos, G. & H. Patrinos. (2002). Returns to investment in education: A further update.

World Bank Policy Research Working Paper No. 2881.
197
Ronda, R.A. (2008). Diploma mills’ hurting RP nursing education – COA report. The Philippine
Daily Inquirer.
Soriano, L.B. (2009). The OFW economic engine. Philippine Reality and Required Reform Arising
From The Global Financial Crisis.
Tiongson, F.L. (2008). Boom in nursing education began quality decline. The Sunday Times.
Tullao, T.S. (1982). Implications of the brain drain. DLSU Dialogue, 16(1).
Tullao, T.S. (2007). Impact of overseas Filipino labor migration on macroeconomic stability, the
labor market and human resource development. Proceedings of the Fourth Ambassador Alfonso
Yuchengco Policy Conference. Yuchengco Center. De La Salle University-Manila.
Tullao, T.S. (2008). Demographic changes and international labor mobility in the Philippines:
Implications for business and cooperation. Philippine Pacific Economic Cooperation Committee.
Tullao, T.S. & J.P.R. Rivera. (2008). The impact of temporary labor migration on the demand for
education: Implications on the human resource development in the Philippines. East Asian
Development Network.
Tullao, T.S., M.A. Cortez & E.C. See. (2004). The economic impacts of international migration: A
case study on the Philippines. East Asian Development Network.
198
Chapter 9: Vietnam’s Regulatory, Institutional and Governance Structure
for International Labor Migration
NGUYEN HUYEN LE
Institute of Labor Science
and
DANIEL MONT
University College of London
ABSTRACT: Vietnam considers labor export to be an important tool for job creation, poverty
reduction, and sustainable economic development. The government has promulgated many policies to
support labor exports including incentives exist to encourage workers to go work abroad and various
types of training and preferential loans. Some local officials believe that remittances have
transformed their communities, and qualitative interviews suggest that working overseas leads to
improved incomes for both migrant families and migrant home districts. There are however important
governance and institutional issues that can lead to negative outcomes from migration and the
dominant causes behind this is fraud. Many brokers and other agents do not comply with the 2006
Law on Overseas Workers, and some local officials are too weak to deal with this situation. One
reason that such fraud continues to survive is that many people still lack knowledge of the labor law
and overseas work procedures. Officials are often reluctant to disseminate information about
overseas labor recruitment and are often not very effective at distinguishing between legitimate and
illegitimate sending companies. Service enterprises are not always effective at reaching out to
potential migrants. Potential workers lose the benefit of receiving official information and instead,
have to rely on brokers and hearsay. This chapter discusses recommendations to provide more
information to workers and improve the enforcement of workers’ rights.
1. OVERVIEW OF INTERNATIONAL MIGRATION IN VIETNAM
Vietnam’s economic growth and progress in reducing poverty has been impressive in recent
years. However, compared to the rising labor force, employment growth has been somewhat low, and
with most workers still in agriculture, surplus labor in rural areas remains abundant. To address these
conditions, the Government of Vietnam (GOV) is promoting its overseas work program to alleviate
pressure on employment and to allow its citizens the opportunity to earn capital from working abroad.
In fact, a recent poverty reduction program aimed at the 62 poorest districts in Vietnam explicitly
includes training and subsidizing migrants to go work abroad.
As a result, the number of international migrant workers from Vietnam has grown
significantly. What started out as a pilot program for 810 workers in 1992 has grown to over 85,546
workers in 2010. As can be seen in Table 1 (next page), the growth has been steady for the last 10
years, and was only reduced slightly by the recent global economic crisis. Still, the total stock of
international workers from Vietnam is only estimated to be slightly more than 500,000.
According to Table 2, most Vietnamese migrants -- 69 percent -- are unskilled. By

occupation, most of these migrants work in manufacturing and housekeeping (Table 3). The most
recent figures suggest that over 75 percent of all Vietnamese migrants work in these twofields.
199
Table 1: Number of Workers Migrating Overseas, 2001-2010
Of which: Female
Year Total
2001 36168 7704

2002 46122 10556
2003 75000 18118
2004 67447 37741
2005 70594 24605
2006 78855 27023
2007 79503 28278
2008 86990 28598
2009 73028 22020
2010 85546 28573
Total 540679 233216
Source: Department of Overseas Labor (DOLAB), Government of Vietnam
Table 2: Overseas workers by skill level, 2005-2009.

By skill 2005 2006 2007 2008 30/6/2009 Total
Total 71823 86931 96085 95682 25971 376492
University 28 63 129 60 45 325
Skilled 11007 20141 33204 38331 11765 114448
Unskilled 60788 66727 62752 57291 14161 261719
Table 3: Overseas workers by occupation, 2001-2005.

No By occupation 2001 2002 2003 2004 2005 Total
1 Manufacturing 22725 23064 20940 13636 51361 131726
2 Construction 3008 1986 13539 1600 839 20972
Textile and
3 1099 2127 2212 1494 886 7818
Garment
4 Housekeeper 3576 9589 29649 44134 9820 96768
5 Nursery 143 200 337 111 551 1342
6 Sailor 1220 1244 1028 1081 1201 5774
7 Fishery 1277 4456 3775 2611 2514 14633
8 Other 3120 3456 3520 2780 3422 16298
Total 36168 46122 75000 67447 70594 295331
Most migrant workers from Vietnam work in East Asia. Table 4 shows that 42 percent of
Vietnamese workers are in Taiwan and another 34 percent are in Malaysia. In the past, more
200
workers went to former Soviet bloc countries – in part because of previous political ties. However,
migration to the former Soviet bloc has declined in recent years, and migration to the Middle East has
increased. However, recent problems in Libya have prompted the early return of thousands of
Vietnamese workers.
While some workers from Vietnam migrate illegally, there are no good estimates of the
number of illegal and undocumented migrants from Vietnam. However, the GOV does have
information on workers who did not return home after the completion of their work contracts. In
2010, approximately 150,000 workers fell into this category, with the largest number of Vietnamese
workers (10,711) remaining in Taiwan, followed by South Korea (4200).
Remittances sent home by migrant workers represent an important part of foreign exchange
in Vietnam. In 2007 it was estimated that international remittances amounted to $5.0 billion, and
accounted for about 8 percent of GDP.1 These remittances are important to both economic growth
and poverty alleviation in Vietnam.
Country Context
Since 1986, when Vietnam embarked on the "Doi Moi" process of moving to a “socialist-
oriented market economy”, Vietnam has made important economic advances. During the period
2001-2010 Vietnam's economy grew at an annual average rate of 7.3%. By 2008, GDP per capita had
quadrupled to 1,000 USD, and Vietnam left the list of the poor countries to join the group of low
middle income countries. Even during the recent economic crisis, GDP growth in Vietnam remained
above 5%.
Nevertheless, surplus labor remains a major problem in the country. Population is increasing
dramatically and the growth in the labor force is outpacing that of employment. According to the
2009 census, the population of Vietnam was 85.8 million, an increase of over 33 million from 1979.
The growth rate of the labor force is currently 2.4% per year, which is higher than that for both
population growth (1.7%) and employment growth (2.2%).2.
1
World Bank, Migration and Remittances Factbook.
2
Census of Population and Housing, Labour Employment Survey, General Statistical Office.
201
Table 4: Destination of Overseas Workers in Vietnam, 2001-2010
Destination
Year Total Saudi
Taiwan Japan Korea Malaysia Lao Cambodia Singapore Thailand Others
Arabia
2001 36168 7782 3249 3910 23 21204
2002 46122 13191 2202 1190 19965 9574
2003 75000 29069 2256 4336 38227 1112
2004 67447 37144 2752 4779 14567 8205
2005 70594 22784 2955 12102 24605 8148
2006 78855 14127 5360 10577 37941 10850
2007 79503 23640 12187 26704 16972
2008 86990 31631 6142 18141 7810 23266
2009 73028 21667 5456 7578 2792 2604 9070 1769 195 0 21897
2010 85546 28499 4913 8628 11741 2729 5903 3615 164 36 19318
Total 540679 229534 35285 83428 184375 5333 14973 5384 359 36 140546
202
During the period 2000-20101, the growth elasticity of employment in Vietnam was only 0.32, which
means that employment increased by only 0.32% for every 1% rise in GDP. Compared with other
countries in Southeast Asia this is a low rate of employment growth.
In Vietnam over half of the labor force remains in the agriculture sector. But while the
proportion of agricultural labor declined from 64.1% in 2000 to 52.5% in 2008, the number of
agricultural workers in the country actually increased. Thus, domestic employment has not kept pace
with labor supply, and surplus labor is abundant in rural areas where productivity remains low.
Moreover, agricultural land is being reduced as part of the urbanization process. This “pushes” rural
migrants to seek employment, and has made international migration a priority of the Government.
To reduce pressure on employment, Vietnam has developed policies to encourage work

abroad for a limited period of time under labor contracts (also known as exported labor). In 2010 the
GOV has a goal of exporting 87,000 workers. To achieve this target, the GOV has instituted a
number of policy reforms. For example in November 2006 the Vietnamese Congress passed the Law
on Vietnamese employees working on overseas contracts. This law stipulates that services associated
with overseas contracts can only be undertaken by companies or organizations which are licensed by
the government.
Although some progress has been made, labor exporting has met a number of problems in
Vietnam. Most importantly, the demand for overseas work far exceeds the supply of jobs arranged by
the labor export companies. Moreover, there are also many loopholes in the management of labor
export services that undermine the benefits received by workers. In many cases, prospective workers
have had to accept addendum costs from brokers that are much higher than those prescribed by law.
Objectives
This paper will analyze the regulatory and institutional arrangements governing the recruitment,
use and repatriation of Vietnamese labor to and from foreign countries. In particular, it will examine:
 The role of Government agencies in arranging contracts with foreign employers

 Current policies on recruiting workers to work abroad for a limited period
 The role and quality of regular recruitment agencies
 The nature of private channels for arranging overseas labor
 The experience of exported laborers: Opportunities and challenges of exported laborers,
including the recruitment process, cost, procedures, work conditions, and the impact of
migration on families in Vietnam
An assessment will be made of the costs and benefits of this activity and the factors that have shaped
its development. Finally the paper will present recommendations for improving policies and
procedures.
1
Authors’ estimation from GSO data.
203
2. THE INSTITUTIONAL AND POLICY FRAMEWORK OF LABOR EXPORT IN
VIETNAM
Brief history
Labor export policies have evolved over time in Vietnam. The first policies date to 1980
when the country began sending workers abroad to other socialist countries. Vietnam and other
governments made cooperative agreements to exchange laborers and exports under the principal of an
international division of labor. Governments selected workers, managed them, dealt with logistics,
and provided workers with settlement upon their return. The goal was to promote economic growth in
both sending and receiving countries.
As Doi Moi progressed in the 1980s, international migration became more market oriented.
In 1988 Vietnam abolished its centralized, subsidized system and established a number of public
service enterprises to manage overseas labor, sign contracts and to help workers go abroad. Most of
these organizations were state owned. Beginning in the early 1990s, the Government promoted a
market mechanism involving contracting directly with firms, in order to respond to labor supply and
demand. State management was separated from labor export services.
In the 1990s labor export enterprises were established and licensed to perform labor export
services. Simultaneously, the Government promulgated regulations to support enterprises in becoming
more market oriented. In 1998 the Politburo directed that labor export (and the exchange of experts)
“must be expanded and diversified in form. The labor export market should be in accordance with
market mechanisms which are managed by the State, to meet foreign countries’ demands for
qualifications and professions” (Directive No. 41-CT/TW, September 22, 1998).
The results were disappointing. Labor export enterprises were weak and the reputation and
quality of Vietnamese migrant workers was low. Moreover, labor contract violations, fraud and
illegal migration were commonplace. Overall, management of many enterprises was considered
inadequate, especially when it came to protecting the rights of overseas workers.
The 2006 Law on Overseas Workers.
In November 2006 the Law on Overseas Workers (effective July 1, 2007) was passed in order
to encourage labor export by addressing the problems mentioned above. This law attempts to protect
the rights and lawful interests of overseas workers, service enterprises, and related organizations. It
promotes investment in the labor export market by supporting managerial, vocational, and foreign
language training, and it also provides preferential treatment in credit policy for social welfare
beneficiaries to work abroad. Finally it encourages employees who have professional qualifications
and technical skill to work overseas.
Specifically, the 2006 law:
 Establishes procedures for services enterprises in regard to licensing and fees, including
conditions for suspensions and revocation of those licenses, as well as the responsibilities for
service enterprises in the case of a revoked license, dissolution or bankruptcy.
 Establishes the requirements for labor contracts, as well as the rights and obligations of
overseas workers.
 Establishes brokerage fees and regulates service enterprise and workers deposits, and the
Overseas Worker Support fund. This fund supports and promotes overseas workers. The fund
204
is pooled from contributions from enterprises, workers and state, and is used mainly for
expansion of the labor export market and for improving the quality of labor force.
 Sets up a program for fostering knowledge among workers
 Establishes a policy for return workers
To complement this law, the Government promulgated specific policies to make conditions for
overseas work more advantageous, including
 Policy for employees to borrow money for working abroad. The Bank of Agriculture and
Rural Development now provides loans to cover expenses at more favorable conditions than
the commercial banks overseas workers had used in the past. Workers in poor households
and beneficiaries of social welfare are able to borrow money with preferential interest rates at
the Bank for Social Policy.
 Policy of vocational training support. Poor people and beneficiaries of social welfare
obtaining apprentice to work abroad are exempt from course fees. Workers are also provided
necessary knowledge (basic life skills, integration into new working and life environment
abroad, etc.) before working abroad.
 Policy of information dissemination: Promoting dissemination of policies and legislation on

work overseas in accordance to the decision of Prime Minister. .
 Policy of administrative simplification. For example, the granting of passports has been
streamlined.
With the passage of this law, labor exports from Vietnam started to grow. From 2006 to
2008, Vietnam sent nearly 250,000 labors to work abroad, averaging 83,000 annually. The number of
migrant workers produced during this period equaled the number of all workers sent abroad during the
previous five-year period. Also, the number of sending enterprises grew rapidly from 50 enterprises
in 2003 to 164 in 20092.
Through these efforts, Vietnam increased its market share in traditional markets such as
Japan, South Korea, Taiwan, Malaysia, Libya, and other countries (see Figure 1). In 2008, Vietnam
sent 12,000 new employees and over 6,000 returning employees to South Korea. It also opened new
markets for migrant workers, in such places as Brunei, Singapore and several Middle Eastern
countries (UAE, Qatar and Saudi Arabia).
2
Progress report of DoLAB, 2009
205
Figure 1: Number of overseas workers by destination, 2001-2010
100000
90000
80000
70000
Total
60000
Malaysia
50000
Korea
40000
Japan
30000
Taiwan
20000
10000
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Source: Data from the Department of Overseas Labor DOLAB
Vietnam’s Overseas Workforce
Although exact data is lacking, most overseas migrants from Vietnam come from rural areas.
According to the Department of Overseas Labor (DOLAB) most international migrants are also in
their 20s and early 30s, and about 80% are married. Approximately 63% are female.
Typical labor contracts for migrant workers range from three to five years, sometimes with
the option to be renewed. Of course, some workers return early, either because their jobs disappear
(e.g., the recent experience in Libya) or because the work conditions are not as expected.
Unfortunately, the DOLAB does not have detailed data on these issues. However, a recent pilot study
by the GOV sheds some light on the overseas migrant population. The study examined overseas
workers returning to Vietnam in the period 2004-2008 who originated from four rural, high-migration
provinces. According to the study, most migrants (88.3%) were employed prior to migration, mainly
in low-skilled agricultural jobs. A large majority (about 70%) were working in the household
economy. Female migrants were much more likely to be unemployed than men prior to migration
(11.4% compared with 3.7%), and women were only one-sixth as likely to be employed in wage
labor.
During focus groups the migrant workers were asked about their reasons for going to work abroad.
The primary reason is economic, especially the desire to escape poverty. Forty-seven percent of the
workers said their primary reason for migrating was to earn more income and nearly 39% stated that
they wanted to escape poverty. Just over 5% wanted to improve their skills and gain professional
experience.
Working overseas is a dream for most Vietnamese workers. Most people who want to work
abroad are ready to do anything just as long as their income is higher than at home. It is this strong
desire to go work abroad that makes prospective migrants susceptible to the unsavory practices of
brokers and labor supply centers.
206
The 62 Poorest Districts Program
In an effort to reduce rural poverty and unemployment, the GOV has recently launched an
overseas work program aimed at the country’s 62 poorest districts. This program uses a number of
different incentives to try to boost the number of workers from these districts who go to work abroad.
Supports offered by this program include
 Subsidized training. This includes full tuition support for one year of education beyond
primary, learning materials, food and living expenses while at school and transportation to
school. Foreign language training, cultural training, and information on the Overseas Worker
Law are also provided.
 Travel related expenses. This includes the costs of a required medical examination, passport,
visa, and police report.
 Preferential credit policies: The Bank of Social Policy provides preferential loans to all poor
workers in these districts to cover any remaining expenses, including overseas transportation
and service fees. Vocational schools for labor export are able to borrow at preferential rates.
Non-poor people in these districts get 50% off of the cost of their training and are allowed to
borrow from the Social Policy Bank. In addition, there is a risk coverage policy that covers deaths,
accidents, and contracts ending due to no fault of the worker in the first year. There is also a public
relations campaign to advertise the program. After completing their contracts, overseas workers
receive assistance when they return in finding a job or starting their own business.
While this program just began in May 2009, early indications are that it might be a success.
For example, in Dak Rong District in Quang Tri Province there had only been 39 overseas workers
between 2006 and 2008. However, by November 2009 that district already had produced 149 new
migrants. When these 149 were recruited, the licensing company was looking for 400 workers. One
thousand applied to go work abroad, but only 168 qualified (based on the hiring company’s
requirements). The main disqualification was not having the required minimum 9th grade education.
As of December of 2010, 30 sending enterprises were participating in this program with 70

supply contracts in such countries as Malaysia, UAE, Libya, Laos, Japan and others (Table 5). In
2010 these enterprises had identified over 3600 overseas workers, up from only about 900 the
previous year. Among these migrants, 93% were from poor people from various ethnic minorities.
Table 5: Participants in Migration Program of 62 Poorest Districts Program, 2010

Destination Number Percentage
Malaysia 2400 53.3
UAE 550 12.2
Libya 390 8.7
Laos 200 4.4
Others 960 21.0
Source: Department of Overseas Labor, Government of Vietnam
Vietnam’s regulatory, institutional and governance regime for managing labor export.
Decree 126 of the Government in 2007, “Guidance on Implementing Regulations on the Law
on Overseas Workers,” defined the responsibilities of agencies involved in sending laborers to work
overseas as follows:
207
Ministry of Labor, Invalids and Social Affairs (MOLISA). MOLISA is given the primary
responsibility for overseeing overseas workers, both in terms of management and strategic
development. They are tasked to co-ordinate with other agencies in drafting legislation, policy, and
regulations, and also in developing labor markets aboard.
Other responsibilities include negotiating international agreements, developing worker

training programs (and guidelines) and granting licenses to service enterprises. They also organize
and guide the registration of these enterprises and supervise the implementation of service contracts,
as well as settling complaints relating to dispatching workers overseas on labor contract basis
(including mediating the worker's or enterprise's complaints)
Ministry of Foreign Affairs (MFA). Together with MOLISA and other ministries, the MFA
is tasked to promote overseas work and propose possible changes in overseas worker guidelines.
MFA’s overseas consulates are responsible for consular protection, protection of overseas workers’
rights in the host country in accordance with the Vietnamese law, the host country laws and
international treaties. In coordination with other functional agencies, they conduct research and supply
information to develop labor market in host countries. They also direct and guide diplomatic missions.
Ministry of Public Security. The Ministry of Public Security assures the legality of
international travel by granting passports, investigating and prosecuting violations of the law by
workers and organizations, receiving workers expelled from host countries
Ministry of Health. The ministry of health sets and enforces the health standards necessary
for overseas work. They also work with MOLISA to evaluate the general health of overseas workers,
and supervise and handle violations of the health facilities.
Ministry of Finance. MOF coordinates with the MOLISA and other ministries to provide a
financial regime for Vietnamese laborers to work abroad. This includes setting cost norms, charges
and fees, and funds, as well as establishing procedures their management.
State Bank of Vietnam. The State Bank works with MOLISA and other ministries to
promulgate the credit policies associated with overseas migration, providing incentives for social
welfare beneficiaries to seek international employment. Their responsibilities include guiding the
credit agencies that lend to overseas workers.
Provincial People’s Committees (PPCs). PPCs manage the State’s responsibilities within
their respective localities. In the process of propagating policy on labor export, they recommend
workers, coordinate with service enterprises’ recruitment efforts, and oversee implementation of
activities on dispatching worker overseas. They are supposed to ensure that the laborers and
enterprises both exercise their rights and perform their contractual.
Sending Companies
In Vietnam the main engine in the overseas worker process is the service agencies. These
agencies locate jobs overseas, recruit and train workers, organize labor contracts, and help transport
workers to their new jobs. With only one exception, all of these service agencies are private-for-
profit entities. The Government sets up and enforces the policies and legal arrangements, but the
process runs through the service enterprises, also known as “sending companies.” According to law
regulations, the sending companies have to license and must directly organize service activities to
employees to work abroad. If they send more than 100 laborers to a particular destination, the
sending companies must also have a representative stationed there.
The number of sending companies has increased in recent years, rising from 139 in 2005 to
170 in 2010. The government is concerned about the degree of oversight conducted by the companies
so they limit them to a maximum of three branches. Out of the 170 companies currently operating, 51
208
have additional branches, of which only 12 enterprises have two branches and only4 enterprises have
three branches. Most of these sending companies (105) are located in Hanoi, with 32 companies
located in Ho Chi Minh City, 10 in Haiphong and the rest in eight other provinces.
All sending companies must have sufficient legal capital – 1 billion VND deposited at a
commercial bank -- to grant service permits to work overseas. The General Director of Overseas
Labor has the right to use these funds to resolve any problems with the enterprise. The business
licenses of the sending companies are only for organizing service activities to employees to work
abroad.
Only one non-profit service enterprise exists under the auspices of MOLISA. This center
was established in 2004, to provide licenses for foreign workers in South Korea. Vietnam and South
Korea have signed an agreement for sending laborers to South Korea, and this is the only center for
which Vietnamese workers can get processed to work in that country.
MOLISA sits on top of the process of exporting migrants because it is involved both with
negotiations and treaty signing with receiving countries and with the promotion, recruitment, and
licensing of migrants and sending companies. Their mission is to regulate and promote the growth of
the overseas labor market.
Figure 2: Recruitment Process - Sending laborers to work abroad.
MOLISA MOFA Government of

destination
Bank Local Government
The Employers in
Broker Destination
Sending
Worker
Company
Health Public Security

service
Note: Non- official way (illegal)
Official way (legal)
The Law stipulates that service enterprises can only charge for brokerage services and money
to implement labor export activities, and that this fee should not exceed one monthly salary per person
for a one year labor contract. If the market demands a higher rate, the sending companies must inform
MOLISA and decide upon a proper fee in agreement with the Ministry of Finance.
Service enterprises collect a lump sum brokerage fee from workers before they emigrate. If
workers have to return ahead of time through no fault of their own (e.g., natural disasters, war,
209
business bankruptcy), they can qualify for 50% reimbursement of these fees if they have not finished
50% of the contract
Service enterprises also receive service fees from workers, not to exceed a monthly salary (or
benefits) under contract for one year of work. As with the brokerage fee, monthly salary used to
determine the fee cap is the basic salary and does not include: money to work overtime, bonuses and
other benefits. Particularly for officers, crew members train maritime transport, monthly wages under
the contract is used to determine a service fee that is the wage service includes basic wages and
supplemental wage.
Sending companies cover all costs associated with vocational training, foreign language
training, clinics, and other costs according to regulations. The only formal costs that workers have to
pay are brokerage fees and service charges. Besides these, both workers and enterprises have to make
unofficial payments, often referred to as “informal”, especially for the high income markets such as
Taiwan, South Korea, Japan.
In order to cover these costs, the Government has a preferential credit policy. Before leaving
to go abroad, workers are allowed to borrow from banks a specified amount of money at a low rate of
interest.
Since they are often located in large cities, sending companies are not always successful at
finding prospective migrants, who typically reside in rural areas. For this reason, sending companies
often use private, informal brokers, for whom there are no provisions in law. These brokers often
"move with the times", rapidly adapting themselves to the situation to find and recruit rural laborers to
go work abroad. These brokers are typically a breeding ground for fraud because they extract large
rents from prospective migrants. The services and procedures offered by brokers represent a type of
black market that is neither regulated nor supervised by the government.
In addition to extracting rents, brokers often do not follow legal procedures. For instance,
brokers sometimes produce "fake papers" for sending workers overseas, or sometimes send workers
abroad without the proper visa requirements. In these instances, workers may find a job abroad, but
that job is usually very risky because no organization will take responsibility for their stay abroad. A
comparison between the formal and informal routes to employment is shown in Table 6.
Table 6. Labor Export Services Process: Formal vs. Informal

Formal services enterprises (licensed) Informal enterprises (organizations and
individuals
Step 1: Obtaining information about labor demand
Source of information: Source of information:
Foreign brokerage enterprise and employers; Mass media, other overseas workers,
personal relationships
Type of information received:
- Number of job openings, by industry, occupation and Type of information received:
place of work - General information on wages and hours
- Wages and bonuses (if any) and on costs of migrating, often highly
- Working conditions, including working hours, rest inaccurate because of multiple and informal
time sources
- Duration of contract
- Accommodation;
- Dispute settlement system
- Health care and social insurance regimes
- Conditions for contract terminations (earlier than
bargained period); severance pay; responsibility to pay
round trip transport cost from Vietnam to the
destination; responsibilities of participants if the
210
Formal services enterprises (licensed) Informal enterprises (organizations and
individuals
worker die when he/she was abroad; Responsibility to
help workers send money back home.
Step 2.Labor recruitment
Provide information to potential migrants Provide information to potential migrants
-Contact local Governments for approval and - Personal introductions
coordination with Provincial Steering Committee for - Contact through acquaintances and other
recruitment at district level personal relationships
- List on bulletin board at enterprises’ headquarters
- Publish on website and in mass media
- Send information to local Governments and domestic
job services to distribute
Step 3. Consulting and training for employees

- Consulting: Consultants from service companies will - Consulting: Not performed because of
consult and receive labor export demand information. incapacity in cases where sending workers
- Training: Service enterprise provides directly or directly overseas, or
contract out: foreign language, vocational, cultural and -Training: Transferred to service enterprises,
procedural when acting as an intermediary
Step 4. Labor Contract
- Service enterprises submit labor supply contract to - Contract made directly with worker with no
Department of Overseas Labor, which appraises and oversight or supervision when sending
grants approval directly overseas
- Transferred to service enterprise when
acting as an intermediary
Step 5. Sending laborers to work abroad
- Profile: Enterprises co-ordinate with workers, create - Not be performed because of incapacity or
a file on profile, loan records, medical examination, transferred to service enterprises.
passport, etc. This task will be done in cooperation
with police agencies, health care centre in accordance
with the current law.
- Procedures (plane ticket, etc) to go abroad
Step 6. Monitoring and supporting for workers overseas
- Representatives of service enterprises in the host Not be done in cases when brokers send
country support workers to solve potential problems workers directly, but transferred to service
with their employers when necessary if there are more enterprises when they act as intermediaries.
than 100 workers
Workers who are recruited illegally by brokers are often cheated, face higher costs, and are
exposed to greater risks. As background for this paper, a series of group discussions and in-depth
interviews were held with 30 migrant workers. Among them were 22 people looking to career
services enterprise through brokers. Only eight of them were recruited directly by service enterprises
in their respective localities.
The experience of these 30 migrant workers with brokers is mixed. Some brokers appear to be
honest and reputable and are merely extracting rents in return for providing access to information and
services more efficiently than would occur if they did not exist. However, there are also many
brokers that get money illegally from services enterprises and workers and offer poor services, which
also damages honest efforts to recruit workers.
211
Interview: Chief of Labor Office, Department of Labor Invalid and Social Affairs (DOLISA) in Nghe
An
"There are many centers, enterprises or individuals (brokers) who report for duty as intermediaries
looking for sources of labor service enterprises. Also, there are many enterprises that need these
workers. Consequently, disputes happened. It is difficult to manage. For example, Broker A recruited
10 workers for the Enterprise C. For each referred worker, C would pay a certain amount to A. Along
with that, enterprise D also hired broker A to look for workers and paid a higher amount. Therefore,
A only provides 5 workers for C, left 5 others to provide for D. This occurred continuously, so,
enterprises offer different prices, it leads market to turmoil, and scrambling over each other.
In order to recruit workers, some enterprises (often through brokers) provide information that
is misleadingly optimistic. And if things do not work out, workers can be left hanging. According to
the law, sending companies that have engaged potential workers can only keep them waiting a
specific length of time. If the time limit expires and the sending companies have not sent the laborers
abroad, they must clearly inform the workers of the reasons for the delay. Moreover, within 15 days
the companies must reimburse workers for all costs, including records costs, physical examinations,
accommodation during training courses, and any visa or document charges. Unfortunately many
service companies do not comply with this rule, and many prospective migrants are themselves
unaware of this law.
Interview: Worker in Nghe An.
"Last time, an enterprise from Hanoi came. They advocated for and gave information on working
abroad. Being introduced by a commune officer, I agreed to go. All procedures were done by the
enterprise, including the loan documents. However, after more than a year, I’m still here and don’t
know why. My cousin has gone to work overseas. I’m in debt and have no information from that
enterprise."
Many enterprises take advantage of legal loopholes to achieve their goals. This problem
occurs most frequently in the recruiting process. Labor export law stipulates that each enterprise is
allowed to open up to three branches. However, the branches often then open affiliated centers or
links to brokers to create an informal network. These informal brokers form a multi - level process
for potential workers, which are ripe for fraud and abuse.
For example, once brokers bring potential migrants to a sending company, the sending
company will create paperwork indicating that they signed the migrant directly, with no mention of
the brokerage firm. In this case, the potential migrant will probably be made to pay twice: first, the
worker will have to pay a fee to the sending company who will in turn pay a percentage of that fee to
the broker; and second, the potential worker will have to pay a fee to the broker who will then pay a
percentage of the fee to the sending company. The actual amount varies from client to client.
Interview: General Director of Labor Export enterprise in Nghe An
“For Nghe An and Thanh Hoa - where labor is a relatively abundant resource – there are hundreds of
labor supply brokers. They collect fees, with the commitment by labor-based contracts associated with
labor export enterprises.
"Last year, enterprise X came, recruited in the whole district and found 39 workers. Some people
have gone to Malaysia but most of them are still here. For me, they said that my procedures were
completed things were only waiting to fly but six months have passed and nothing happened. I asked
the district authorities, they said they didn’t know, that the enterprise didn’t submit them. However,
that enterprise had rented an office in the district centre, clearly. They hung their sign, had their own
212
address and telephone number. Now that signs is still there, but closed, nobody answer telephone
number. District government has not known and has not been responsible to handle”.
For their part, labor sending companies face problems which are created by local government
officials. The Law on Overseas Workers stipulates that sending companies can recruit labor
throughout the whole country. However, when these companies want to recruit labor in any particular
province, they have to be "licensed" by the local government, in order to let local governments know
about their activities, and help people avoid scams. This creates a burden for sending companies, as
reflected in the following newspaper article: "According to many labor export enterprises, when they
recruit laborers through the career guidance centre of Department of Labor Invalid and Social
Affairs in some central provinces, they usually have to pay 500.000 to 2.000.000 VND per worker for
commissions. Otherwise, it would be difficult to recruit labor, not even if licensed in the provinces. To
avoid this "Village custom", instead of the hard work permit, some enterprises choose to develop
centipede-foot-shaped, known as “decoy-duck”. These people are paid at a fixed rate per worker.
"If there are many labor export enterprises that all recruit in the same locality, some local officials
will assist depending on the money they receive. If an enterprise does not know about this “village
custom", they won’t recruit any workers even if their order is good because they are out of favor with
local officials.
"… in other words, some local government officials have been brokers” – Director of Labor Export
Enterprise in Ho Chi Minh City said
In the end, the workers suffer. They have to pay higher costs and accept higher risks. When
workers lose out to these risks, distrust in labor export grows, making recruitment more difficult for
lawful enterprises.
Interview: Service enterprise (in Hanoi)
"Some districts of the Central Provinces even said "no" to labor export. Some workers went to work
overseas unluckily. When they went to the host country they lacked employment and had to return.
Some others completed their procedures but had to wait for a long time. Service enterprises and
brokers promised over and over but they couldn’t go to work overseas. People complained to the
government but the government couldn’t solve because enterprises disappeared. Furthermore, after
that, it is difficult for other enterprises to recruit workers to work overseas difficult because people
have lost their belief.
Deep interview: Service enterprise (in Ho Chi Minh City).
"There are localities where we came to recruit workers, people and governments were apathetic. They
said that recruitment was very difficult. There had been other enterprises which had come to recruit,
and workers participated enthusiastically at first. They had been trained vocationally, and in foreign
language and many more procedures. However, they couldn’t go and had to wait for so long, being
promised over and over. At the end, they were too tired to believe. They won’t trust anymore. Now,
we must recruit all the cases failed before and pay for them by ourselves, including accommodation
until they will be able to go. We only receive their money when they leave.
These examples show that prospective migrant workers do not understand the 2006 Law on
Overseas Workers. They believe in brokers, and take a risk. A survey of the Labor Social magazine
in coordination with the Department of Legal Affairs – MOLISA -- about information on the labor
law on a sample of 220 migrant workers found that less than one-half of all workers knew anything
about the labor law before they went abroad. Thirty-six percent of workers found out about the labor
law from the staff of sending companies, which was equal to the share of f workers who learned of the
labor law through personal relationships and brokers.
To overcome these weaknesses, the scheme promoting labor export for the 62 poor districts in
Vietnam has paid great attention to dissemination of information on the 2006 Law on Overseas
213
Workers. The Department of Overseas Labor is the unit which prints propaganda leaflets and posters,
and coordinates with local governments to get the word out.
Figure 3: Percentage of workers with labor export information by information channel
Know Do not know
50.45 60.45 66.36 64.09

80.00
97.27
49.55 39.55 33.64 35.91

20.00
2.73
Source: Data survey of the Labor – Society Magazine
In an effort to improve awareness of the labor law, and to recruit more overseas workers, a
new labor export scheme has been piloted in 28 of the 62 poorest districts of Vietnam. About 2800
workers in these districts have registered to work abroad, and nearly 1000 workers have already been
sent abroad. Their incomes abroad seem to be quite good. While the average monthly wage in
Vietnam is between 1.5 and 2.5 million VND, in Malaysia overseas workers can earn approximately 6
million VND per month. According to this pilot program, which lasts until 2020, prospective migrant
workers do not have to pay any costs or take out any loans to go work abroad. Moreover, when they
return to Vietnam after three years, they will be supported with more than 100 million VND to re-
integrate into the community. Although small, this new pilot program is providing an important
bridge out of poverty for a small number of families. The General Director of Department of Overseas
Labor believes that this program can be very beneficial for workers in poor districts.
As described in the recruitment section, many overseas workers have to pay large, illegal fees
to brokers in order to work abroad. Group discussions and interviews with returned migrants show
that 22 out of 30 workers had to pay informal costs with a minimum 10 million VND per person. One
person had to pay 30 million VND but still has not yet been to work overseas.
Labor group discussion, An Giang:
"My neighbor knows a person who went to work in Taiwan. She asked me if I wanted to work in
Taiwan as a housekeeper with basic salary was 500 USD per month without bonuses, accommodation
to be paid by the employee. The contract period was 3 years and can be extended if done well. She
said procedures would be done by a proper enterprise and I had no need to be worry at all provided
that I gave her 50 million VND (about 3000 USD). She would discuss with the service enterprise. I
trusted my neighbor and gave her 30 million VND. However, 6 months have passed, and I’m still
here. She tells me that I had to wait. I don’t know what’s going on.
214
Many migrants are ill-informed about the costs – both legal and illegal -- of going to work
abroad. According to the Labor Social magazine interview with 220 overseas workers, only 166 out
of 220 workers (75%) actually knew what it would cost to go work abroad, and only 89 out of 220
workers (40%) knew the procedures for getting bank loans to work overseas. Given these figures, it is
easy to understand why some workers are cheated and others had to pay extra, informal costs to get
abroad.
As noted above, all workers going to South Korea must go through the non-profit agency
established for this purpose. All workers have to pass a Korean test, and then submit applications
online to Korean companies. The cost for the entire process is about 700 USD, but most workers also
have to pay informal costs to brokers, ranging from 7,000 to 10,000 USD. However, if successful,
emigrants to Korea can work and save quite a large sum of money before the expiration of their work
contract...
Interview: Worker who went to work in Korea successfully:
"Before going to Korea to work, I was introduced through acquaintances. They had done all the
procedures. I only took medical report and the Korean test. Total cost was 100 million (more than
6000 USD). I worked in Korea as an electronic assembly maker, after three hardworking years, I
saved a few hundred million VND to do my business.
Impact on migrants to work overseas and their families /prospective migrant
Since the incomes earned by migrant workers are high, the remittances sent home by these
migrants might be expected to have a large impact on the living standards of migrant households and
communities in Vietnam. However, information on the impact of remittances from overseas workers
on local socio-economic development in Vietnam is limited (see Cuong and Mont, 2010). In
communities where there are a large number of overseas workers, local authorities tend to believe that
remittances have a great impact on the living standards of households, creating a great source of
financial investment for the promotion of production and business in rural areas.
Findings from a survey of returned migrants in four provinces in Vietnam found that returned
migrants earned more per month than before migration. Returned migrants earned 1.72 million VND
per month in Vietnam, which is significantly higher than their pre-migration income of 822,000 per
month. Moreover, most migrants found employment upon their return to Vietnam: 85% of returned
males and 80% of returned females reported being employed upon their return to Vietnam. At the
same time, the percentage of migrants being employed in agriculture fell from 70% before migration
to 54% after migration. These figures suggest that returned migrants are using their savings from
abroad to enter non-agricultural work.
However, the survey also found that returned migrants are generally not using the skills that
they learned abroad. Yet an important caveat must be made here. The sampling procedure for this
survey of returned migrants did not follow those migrants who left the countryside and moved to
urban areas. This means that the migrants who might have gained the most in new skills were not
included in the survey.
Findings from in-depth interviews and focus groups with overseas workers indicate that, in
some cases, there are also social negative effects incurred by household members who stay at home.
One example of these negative effects is reduced child care. When migrants – especially women -- go
to work abroad, sometimes the supervision of the children left at home suffers.
215
Impact of labor export on workers and their families
Interviews with returned migrants suggest that most remittance money sent home to families
in Vietnam is spent on housing (repair and rebuilding) and repairing roads. Remittance-inspired
spending on housing and roads makes rural villages look cleaner and more spacious.
Interview: Vice chairman, district level, Nghe An province
If you come to the communes that have many people who go to work overseas, you will see that it very
different. Many new buildings have grown up. They built houses, made roads, which not only benefit
those households but also the communities. Those villages look nicer and cleaner.
This finding was verified by a brief quantitative survey. According to this survey, after
paying back pre-migration loans, remittances are mainly spent on physical and human capital
investments: building houses (53.3%), investment in child education (24.2%), and business
investment (17%).
It is interesting to note that business investment figured so prominently in the quantitative

survey. Most of the returned workers in the survey reported that they were able to amass significant
savings from overseas work and to use these savings to invest in small businesses at home. A
number of workers stated that working overseas not only helped them earn more money, but also
enabled them to acquire knowledge and skills that increased their confidence in establishing new
businesses. Unfortunately, the only quantitative data to document the significance of this effect
comes from the small survey in the four provinces. DOLAB intends to broaden this data collection to
include more areas of Vietnam.Interview,
Interview Ms. K, An Giang province
"Before I went to work in Russia, I’ve known tailoring already. When I was in Russia, I worked as a
tailor. After returning, I opened a small shop; hired some others to work together. Now, my family’s
economy is fairly good.
Negative impacts:
For many overseas workers, the whole process of working overseas, including looking for
information, going through various migration procedures, working abroad, and finally returning home
is a long risky process.. The level of risk and losses varies across occupation, destination country,
educational and technical skills, and gender. There are also a number of failed workers among migrant
workers from Vietnam. Economic losses suffered by failed workers can be very serious because
most of the victims are poor. To get abroad, migrants typically have to borrow money from relatives,
friends, and banks to pay the sending enterprise and brokers. If they fail to get overseas work or have
to return early, the economic consequences of these loans can be devastating. Specifically, paying the
principal and interest on migration loans can become a huge burden to poor rural families. Rural
migrants with large unpaid debts can also become severely depressed.
To avoid such problems, the goal of GOV agencies is to detect and handle migrant workers’
problems before they become a burden. In recent years the number of Vietnamese workers who leave
their contracts early in South Korea and Japan has been declining. For example, in previous years the
rate of attrition for Vietnamese trainees in Korea and Japan was 20-30%. However, now this rate is
only 2%. Similarly, the success rate for trainees in Japan has been improving. In 2004, about 900
out of 3000 Vietnamese trainees relinquished their jobs. In 2008 this dropped to a rate of under two
percent (112 trainees out of 6142).
216
After their return: DOLISA (the provincial counterpart to MOLISA) is responsible for
informing workers about domestic recruitment demand, and assisting workers find suitable jobs. The
Government encourages enterprises to receive and employ returned workers. Returned workers
register at a local job introduction center where they receive assistance in finding a job. The
Government hopes that after working abroad, the returned workers have more skills, and are thus
more attractive to employers. If overseas workers have been able to amass capital, another choice for
them is to invest in their own production and business firms.
Interview, Mr. H, Hai Duong province
"I went to work in Korea and worked for an electronic assembly company. Before that, I’ve known
little about this job. After returning, I have more capital. I opened an electronics repair shop and
hired a mechanic. My wife was a tailor. Now, she can take care of our children. I rebuilt the house,
opened a clothing shop for my wife to sell, changed the home and shop, to sell more outside Hanoi.
Clothes are taken from Ha Noi.
However, re-integration into the domestic labor market is not always easy.
Interview, Ms. K, An Giang province
"Working overseas was quite hard, work overtime to expect a much higher income. Now, I’m home. I
don’t know what to do. Working for other people does not pay. I have some extra money, so I’ll find
out what I should do. If I have a chance to work overseas again, I’ll take it.
3. CONCLUSION AND RECOMMENDATIONS
The Government of Vietnam considers labor export to be an important tool for job creation,
poverty reduction, and sustainable economic development. For this reason, the Government has
promulgated many policies to support labor export policy, and has adapted these policies through
various stages of Vietnam’s development. Currently, many incentives exist to encourage workers to
go work abroad, including various types of training and preferential loans. Additional incentives are
also being offered in a new program designed at promoting overseas labor migration in the poorest 62
districts in Vietnam. As a result of these policies, Vietnam is currently sending about 85,000 migrants
to work abroad per year.
Remittances sent home by overseas workers now total over $5.0 billion a year and have important
effects on the living standards of migrant households in Vietnam. Some local officials believe that
remittances have transformed their communities, and qualitative interviews suggest that working
overseas leads to improved incomes for both migrant families and migrant home districts. However,
the net impact of remittances needs to be examined with more quantitative rigor before any definitive
statements can be made about the impact of these resource flows on Vietnam.
With respect to migration, the dominant cause for negative outcomes is fraud. Many brokers
and other agents do not comply with the 2006 Law on Overseas Workers, and some local officials are
too weak to deal with this situation. As a result, overseas migrants must make large and illegal
payments to brokers and agents to go work abroad, especially in high income markets like Taiwan,
South Korea and Japan.
One reason that such fraud continues to survive is that many people still lack knowledge of
the labor law and overseas work procedures. Local authorities are often reluctant to disseminate
information about overseas labor recruitment and are often not very effective at distinguishing
between legitimate and illegitimate sending companies. Service enterprises also are not always
effective at reaching out to potential migrants. Potential workers lose the benefit of receiving official
information and instead, have to rely on brokers and hearsay. There is no effective mechanism for
217
providing migrants with the proper access to information to help them make good decisions about
costs and to avail themselves of the protections provided by law.
Recommendations
This paper sets out a few key recommendations aimed at improving the implementation of the
current Law on Overseas Workers, in particular targeting weaknesses in public awareness and in labor
export management.
1. For labor export administrative agencies (such as MOLISA, MOFA, and MOF) the key
issues are awareness raising, better access to information, and improved management and
enforcement.
a. Institute a public awareness campaign aimed at informing prospective overseas

workers of their rights, and how to access the official export labor system.
b. Developing an effective information sharing system for shareholders, including local
agencies, departments, authorities to improve access to information about export
labor opportunities and procedures.
c. Streamline procedures for sending workers overseas in order to reduce the role now
being played by illegal brokers.
d. Step up monitoring and enforcement of legal provisions related to the recruitment and
employment of overseas workers.
e. Develop an effective and viable management mechanism in countries where there are
a large number of Vietnamese workers in order to provide more timely legal and
social services to overseas workers
f. Conduct surveys to assess overseas working experiences, the socio -economic
impacts of working overseas, and the experiences of returned workers.
2. For related agencies and enterprise the key issues are improved management and more
stringent enforcement of workers’ rights.
a. Enterprises should be more be pro-active in expansion of formal service channel to

workers directly, reducing the informal services to minimize risks and financial costs,
time for workers
b. Enterprises need to have positive measures to increase the quantity and improve the
quality of overseas workers to meet the demand of foreign labor market.
c. Enterprises/ organizations/ labor suppliers engaged in dispatching labor overseas are
encouraged to streamline their process, covering various steps from provision of
information, job introduction, training, sending worker overseas, follow -up the
changes to ensure that workers have adequate and suitable job
d. Enterprises should have an action plan for improvement of follow up the status of
Vietnamese overseas workers, to develop good mechanism cooperation with
concerning agencies to settle the problems faced by workers
e. Keep up to date information on overseas employment opportunities including:
responsibilities and legal obligations, working condition, wage, living condition,
working time, fringe benefits, wage management mechanism... for workers during the
process of working overseas
f. Strictly comply with the requirements on transparency of financial obligations,
including fees, costs for various services and other obligations.
218
REFERENCES
Circular No. 08/TTLT-BLDTBXH-BTP on July 11, 2007 of Ministry of Labor, Invalids and Social
Affairs and the Ministry of Justice on guarantees contract and guarantees contract payment for
laborers to work overseas under contracts;
Circular No. 16/2007/TTLT-BLDTBXH-BTC on April 9, 2007 of Ministry of Labor - Invalids and

Social Affairs and the Ministry of Finance guiding the brokerage and services fees;
Circular No. 11/2008/TTLT-BLDTBXH-BTC on July 21, 2008 of MOLISA - Ministry of Finance

guiding the management and use of work overseas support funds;
Circular No. 17/2007/TTLT-BLDTBXH-NHNN on April 9, 2007 of Ministry of Labor - Invalids and

Social Affairs and the State Bank provide for the management and use of enterprises deposit and
deposit of laborers to work overseas under contracts;
Circular No. 21/2007/TT-BLDTBXH on August 10, 2007 of Ministry of Labor - Invalids and Social
Affairs guiding some articles of Decree providing detailing and guiding the implementation of labor
code regarding Vietnamese labors working overseas under contract.
Decision No. 18/2007/QD-BLDTBXH on July 18, 2007 of the Minister of Labor - Invalids and Social
Affairs promulgate the program fostering knowledge required for workers before going to work
overseas;
Decision No. 19/2007/QD-BLDTBXH on July 18, 2007 the Minister of Labor - Invalids and Social
Affairs promulgate regulations on the organizational structure of sending labor to work overseas
activities and specialist responsible structure for fostering knowledge required for workers before
going to work overseas;
Decision No. 20/2007/QD-BLDTBXH on 02/8/2007 by the Minister of Labor - Invalids and Social
Affairs on certificate of fostering knowledge required for workers before going to work overseas;
Decree No. 126/2007/ND-CP dated August 1, 2007 of the Government providing detailing and
guiding the implementation of labor code regarding Vietnamese labors working overseas under
contract.
Decree No. 144/ND-CP. 10/9/2007 of the Government on handling administrative violations in

sending laborers to work overseas;
Decision No. 144/2007/QD-TTg on August 31, 2007 by the Prime Minister on the establishment,
management and use of working abroad support funds;
21. Decision No. 61/2008/QD-LDTBXH on August 12, 2008 of the Minister of Labor Invalids and
Social Affairs on brokerage fees at some markets;
Law of Vietnamese workers going to work overseas under contract, 10th National Assembly, 10th
session of 72/2006/QH11 November 29, 2006.
http://www.dolab.gov.vn/
http://www.tuoitre.vn
Cuong, N. V., and D. Mont, “Economic Impacts of International Migration and Remittances on
Vietnam’s Development,” Working Paper, 201
219
Chapter 10: Singapore’s System for Managing Foreign Manpower
YAP MUI TENG

Institute of Policy Studies,
Lee Kuan Yew School of Public Policy,
National University of Singapore, Singapore
ABSTRACT: This paper examines Singapore’s system for managing the recruitment and use of
foreign manpower which currently forms more than one third of its total workforce. As
background, some data are presented on the numbers and selected characteristics of these foreign
workers. This is followed by an overview of Singapore’s foreign manpower policy and the main
drivers of this policy. The management of foreign workers involves a series of work passes with
different benefits and obligations depending on worker qualifications and skills. Levies,
dependency ceilings and source country restrictions have also been introduced to reduce
Singapore’s reliance on low-cost, low-skilled foreign workers. Despite these restrictions, the size
of the foreign workforce in Singapore has grown significantly because of economic growth and the
need for the country to augment its potentially diminishing population and workforce. The paper
also discusses issues that have arisen from the utilisation of foreign manpower, and ways that
have been proposed to solve these issues. Some sticking points still remain. The paper ends with
recommendations for information sharing and the effective implementation of policies by all
stakeholders.
1. INTRODUCTION
Singapore ranks as one of the countries with the highest proportions of foreigners in its
population and among its workforce in the world. Foreigners (not counting those who have become
citizens and permanent residents) make up over one third of its workforce, and over one quarter of its
five million population in 2009. Over the years, the country has developed a rather elaborate system
for managing the inflow and utilisation of foreigners to augment its potentially diminishing population
and workforce.
This paper examines Singapore’s system of managing its foreign manpower. It includes four
main sections: 1) an overview of foreign workers in Singapore; 2) a review of Singapore’s foreign
manpower policy; 3) a discussion of the regulatory framework, processes and enforcement; and 4) an
assessment of the issues that have arisen. The paper ends with a section on conclusions and
recommendations.
2. OVERVIEW OF FOREIGN WORKERS IN SINGAPORE
Foreigners accounted for a substantial 35.2% of Singapore’s workforce of 2.99 million in

December 20093. Over the last ten years (1999- 2009) the proportion of foreigners in Singapore’s
workforce increased from 30.1% to 35.2% (Table 1). Over this decade the absolute number of
foreigners working in the country grew from 621,000 to more than one million. These figures do not
include the growing number of foreigners who have obtained permanent resident (PR) status and
citizenship in Singapore and are considered part of the local or resident workforce
3
Ministry of Manpower. Labour Market Report 2009 (released in Singapore on 15 March 2010).
http://www.mom.gov.sg/Home/MRSD/Documents/GLM/qtlmr094.pdf (accessed 16 March 2010).
220
Table 1 Employment Trends by Residential Status 1998-2009 (as at end-Dec each year)
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Total 2,022.7 2,062.6 2,171.1 2,171.0 2,148.1 2,135.2 2,206.6 2,319.9 2,495.9 2,730.8 2,952.4 2,990.0
Local 1,399.6 1,441.3 1,499.7 1,501.0 1,520.4 1,535.3 1,585.2 1,648.7 1,739.6 1,830.0 1,894.7 1,936.5
Foreign 623.2 621.4 671.5 670.1 627.8 599.9 621.4 671.2 756.3 900.8 1,057.7 1,053.5
Foreign as
30.8 30.1 30.9 30.9 29.2 28.1 28.2 28.9 30.3 33.0 35.8 35.2
% of Total
Notes: Numbers are in thousands. Locals comprise Singapore citizens and Permanent Residents.
Source: Figures for 2008 and earlier calculated from Employment Change, Labor Market 2008. Figures for 2009 are obtained from Labor Market 2009.
Both sources are published by the Ministry of Manpower.
221
Nearly half of Singapore’s foreign workforce in 2009 can be found in the services sector and
about a quarter each in manufacturing and construction. Foreigners dominate the construction sector,
with more than six of ten workers in the construction sector being a foreigner. In comparison, about
half of the manufacturing workforce and about a quarter of service workers are foreigners.
Detailed information on the foreign workforce is not readily available in the public domain.
This is because such information is considered highly sensitive. Reportedly, the vast majority of
foreign workers are Work Permit (WP) holders, generally the less skilled with basic monthly salaries
not exceeding S$1,800. Numbering 856,000 persons, these WP holders constitute 81% of the foreign
workforce. The remaining 197,000 or 19% are Employment Pass (EP) holders, generally
professionals and highly qualified workers, and mid-level skilled workers on S Passes1. Reflecting
the government’s policy to recruit higher quality foreign workers, the proportions of EP and S Pass
holders have increased slightly over the years with a corresponding decline in WP holders (more
below).
Information on the occupational distribution of the foreign workforce is available only for
2000 based on the results of the 2000 Census of Population. Estimates from that census show that
most foreigners (75%) work in three unskilled and semi-skilled occupations: cleaners, laborers and
related workers; plant and machine operators and assemblers; and production, craftsmen and related
workers categories (Table 2). About 12% of foreigners are senior officials, managers and
professionals, while another 5% are associate professionals and technicians. Compared to the resident
workforce, foreigners are more likely to work in less skilled occupations. It is likely that some of the
foreigners at the top of the occupational spectrum have taken up permanent residency or citizenship
and are not counted among the foreign workforce.
Given its geographical proximity, shared history and culture, Malaysia has traditionally been
the primary source of migrant foreign workers in Singapore. However, in recent years the source
countries for Singapore’s foreign workforce have expanded to other countries in the Southeast Asian
region, North and South Asia, and elsewhere. Estimates of the size of foreign population groups by
nationality vary widely, depending on the source, and are probably not very reliable.
While there are illegal foreign workers in Singapore, the issue of illegal immigration is not as
serious in Singapore as in some neighboring countries. The reason for this is that Singapore is a very
small country with a relatively long coastline. Apprehensions of illegal workers can be made at
border checkpoints or along the coastline. As a result of these factors, as well as strict enforcement by
Singapore authorities, the number of illegal immigrants in Singapore has declined from more than
10,000 in 2001 to 1,800 in 20092. Singapore also has some problems with over-stayers, that is,
foreign workers who entered the country legally but did not leave the country after the expiration of
their permits. In recent years, the number of over-stayers in Singapore has also declined, from 12,000
in the 1990s to 3,700 in 20093.
Illegal deployment is an issue that can be distinguished from illegal immigration. It refers to
the deployment of foreigners to jobs, sectors and employers other than the ones stated in the work
pass. There are also workers who contravene the conditions of their social visit passes by working
illegally. Within the country, the food and beverage and construction sectors appear to be the main
users of illegal workers and immigrants. According to a recent report, “over the past three years,
1 Zakir Hussain, “Foreign worker levy should be raised”, The Straits Times 2 February 2010, accessed 2 February 2010.
2
Immigration and Checkpoints Authority Press Release 17 February 2005; Immigration and Checkpoints Authority
Statistics 2009.
3
Teh Joo Lin, “Broken dreams, tough life for overstayers”, The Straits Times 2 August 2010.
222
Table 2 Foreign and Resident Workforce (aged 15 and older) by Occupation, 2000
Foreign Workforce* Resident Workforce
Number Per cent Number Per cent
Total 612,233 100.0 1,482,579 100.0
Senior Officials and Managers 37,455 6.1 211,835 14.3
Professionals 36,334 5.9 150,265 10.1
Associate Professionals and Technicians 29,699 4.9 283,361 19.1
Clerical Workers 17,884 2.9 213,588 14.4
Sales and Service Workers 28,963 4.7 182,966 12.3
Agricultural and Fishery Workers 292 0.0 1,158 0.1
Production, Craftsmen and Related Workers 159,690 26.1 106,753 7.2
Plant and Machine Operators and Assemblers 66,482 10.9 178,752 12.1
Cleaners, Laborers and Related Workers 234,881 38.4 101,149 6.8
Workers Not Elsewhere Classified 553 0.1 52,752 3.6
Source: Census of Population 2000 Statistical Release

Note: * computed by taking the difference between total and resident workforce.
223
about 0.1% of the foreign workforce has been involved in illegal deployment cases each year
... (and) an average of 190 errant employers was convicted for illegal deployment”1.
Given the size of Singapore’s foreign workforce today, it is difficult to imagine that the
country once aspired to achieve a totally Singaporean workforce. In the early 1990s Singapore’s
leaders warned of an over-reliance on cheap unskilled foreign workers and its impact on Singapore
society2. However, since that time the number and share of foreigners in Singapore’s workforce have
grown dramatically, and especially in the last five years. The next section looks at Singapore’s
foreign manpower policy, the factors that drive this policy, as well as the process and enforcement
aspects.
3. SINGAPORE’S FOREIGN MANPOWER POLICY
Overview
Singapore adopts a two-pronged policy with regard to migrant foreign workers, one for the
less skilled, and another for skilled and professional workers and entrepreneurs/investors. The
country has always welcomed the latter category of foreigners (skilled and professional migrants),
whom it believed could make an important economic contribution to the country. Skilled migrants
are allowed, and in fact encouraged, to sink their roots in Singapore and become permanent citizens.
On the other hand, the lower skilled are viewed as guest workers who are allowed to work in the
country under stringent conditions and expected to be repatriated after their jobs have been completed
or their contracts terminated. Unskilled workers cannot bring their families or sink their roots into the
country -- or at least not until they have upgraded their skills to a level that qualify them for
permanent resident status3.
Migrant workers were initially thought of as a temporary solution to a situation of labor

shortages in Singapore and as a kind of buffer that could ameliorate the impact of changes in labor
demand due to economic cycles4. As elsewhere, however, migrant workers have become a permanent
feature of the Singapore landscape, with the recognition from the mid-1980s onwards that foreigners,
both professionals and the less qualified, were needed to augment the domestic workforce. The
Economic Committee appointed by the government in 1985 to study the causes of Singapore’s first
recession recognised that it was unrealistic for Singapore to be rid of unskilled workers and
recommended that they be allowed in on a “revolving” basis while efforts should be made to retain
skilled workers.5 The view since then, as articulated by various government ministers and
committees, has been that any manpower constraint would hobble economic growth and cause
businesses to relocate elsewhere, thereby dampening job opportunities and lowering the standard of
living for Singaporeans6. As economic growth became more volatile in the first half of the 2000s
decade (Figure 1), Singapore’s leaders have decided to take advantage of all opportunities for growth
and thus have allowed a sharp increase in the number of migrant workers.
1
Hawazi Daipi, “Illegal employment of foreign workers”, Oral answers to questions, 26 April 2010.
2
Pang Eng Fong, Tan Chwee Huat, and Cheng Soo May, “The Management of People” in Management of Success; the
moulding of modern Singapore, ed. Kernial Singh Sandhu and Paul Wheatley (Singapore: ISEAS, 1989), 128-143.
3
Examples of former Work Permit holders who have achieved citizenship include a former Malaysian who has
little formal education but has since become a stylist to local and international stars, and a former Indian and
former Bangladeshi who upgraded their skills while in Singapore. The number of foreigners who have similarly achieved
status change is not known.
4
Pang and Lim 1982; Pang, Tan and Cheng 1989; Hui, Weng Tat, “Foreign Manpower Policy in Singapore” in Singapore
Economy in the 21st Century; Issues and Strategies, ed. Koh Ai Tee, Lim Kim Lian, Hui Weng Tat, Bhanoji Rao and Chng
Meng Kng (Singapore: McGraw-Hill Education), 29-50.
5
Ministry of Trade and Industry, Report of the Economic Committee, The Singapore Economy: New Directions, February
1986,
6
The most recent iteration of this is Prime Minister Lee Hsien Loong’s National Day Rally speech delivered on 29 August
2010.
224
Figure 1: Singapore’s Economic Growth Rate
Real Growth Rate 1965-2009

16
14
12
10
8
Per cent
0
1965 1970 1975 1980 1985 1990 1995 2000 2005
-2
-4
Source: Singapore Department of Statistics. Time Series on GDP at 2005 Market Prices and Real Economic Growth Rate
However, in more recent years there has been a growing recognition that the rate of growth of the
foreign population in Singapore is not sustainable. The current view, following the recommendation
of the Economic Strategies Committee (ESC) appointed in the aftermath of the 2008-2009 recession,
is to maintain the share of the foreign workforce at about one third of the total workforce and to not
let this share grow significantly over the long term7. This one-third share is not an absolute limit so
that businesses have the flexibility to hire more foreigners when needed to take advantage of
growth opportunities (see below). The new foreign workers are also supposed to be of a higher
quality than those in the past8. Employers are urged to invest in training their foreign workers along
with their local employees so as to raise their productivity levels9.
Drivers underlying Singapore’s foreign manpower policy
Economics has always been the main driver of Singapore’s foreign manpower policy. This is
still the case today. As mentioned above, since the early 2000s, the government’s strategy has been
to maximise growth in good economic times. This growth has been fuelled mainly by increasing the
workforce through importing more foreign workers, because domestic population growth in
Singapore has slowed due to low, below-replacement fertility. Although probably unfair, this growth
strategy has been referred to by critics as a “maximise growth at any cost” strategy. Notably, even
during the global economic crisis in 2008-2009 and growing complaints of competition and the rising
cost of living, Singapore’s leaders continued to maintain that the country needed to remain open to
7
Economic Strategies Committee, Report of the Economic Strategies Committee, High Skilled People, Innovative
Economy, Distinctive Global City, February 2010, p 7, para 30. Available online at
http://www.esc.gov.sg/attactments/ESC%20Report.pdf.
8
Janice Heng, “Foreign worker influx is good”, The Straits Times 21 July 2010.
9
Economic Strategies Committee 2010.
225
foreigners to remain globally competitive.10 The social impact concerns raised earlier appeared to
have taken a backseat.
Nevertheless, in an apparent acknowledgement of the disquiet felt among Singaporeans about

the large foreign presence in the country, Prime Minister Lee Hsien Loong, announced in August
2009 that the government would slow the pace of its intake of foreign workers. This was the first
such announcement since the 1980s. In recent years Singaporeans have complained about the
negative aspects of the influx of foreign migrant workers, including competition for jobs, housing and
school places, as well as the lack of basic English skills among frontline service personnel. As
mentioned above, while the move now is to slow down the pace of the intake of foreign workers,
foreign workers will continue to make up about 30% of the workforce. In the meantime, the
government has also sharpened differences in the educational and healthcare subsidies and other
benefits available to citizens and permanent residents. This is in addition to the differentiation already
made between citizens and permanent residents on the one hand, and foreigners on the other.
Specifically, the cost of services provided to permanent residents has been raised in what has been
dubbed a “citizens first” policy.
Demography is another factor that has driven Singapore’s foreign manpower policy. While
too rapid population growth, in particular, the high fertility level, was considered detrimental to
Singapore’s development effort at the time of independence, the country has experienced more than
thirty years of below-replacement and declining fertility rates. The cumulative impact of these forces
is among the most important factors accounting for Singapore’s aggressive drive to increase
immigration to “top up” its population and augment its workforce. While Singaporeans have been
provided with financial and other support measures to form families and have more children, this
effort has yielded little positive results. Since 2003 the total fertility rate in Singapore has been below
1.3 births per woman.11 Migrant workers and new immigrants are justified as necessary to make up
for the shortfall of births and for continued economic growth and prosperity. Without immigration,
Singapore’s population is projected to begin declining in 202012.
Another aspect of Singapore’s demography is the multi-ethnic composition of its population,

comprising Chinese, Malays, Indians and “Others”. This is the result of its history of immigration13.
In order not to upset the existing ethnic balance, migrant workers were initially restricted to sources
that reflected the ethnic origins of the original population. However, with globalisation and the
growing need for talent, this focus on ethnicity has been relaxed somewhat, and the proportions of
Indians and “Others” among the resident population have increased14. Issues of ethnicity remain a
sensitive issue in Singapore as seen by the fact that Malay community leaders have recently expressed
concern about the declining share of Malays in the resident population15. This decline, it may be
noted, is partly due to recent fertility declines among Malays to below the replacement level.
Regulatory Framework
The management of the recruitment and use of foreign manpower in Singapore is governed
by the Immigration Act, the Employment of Foreign Manpower Act (EFMA) and the Employment
10
Aaron Low, “Does restricting foreign talent help Singaporeans?”, The Straits Times 28 February 2009; Peh Shing Huei,
“Foreign talent vital to S’pore’s progress”, The Straits Times 8 March 2009; Chuang Pek Meng, “S’pore needs skilled
foreigners to grow and must continue welcoming them: SM”, Business Times 23 October 2008; Kor Kian Beng, “S’pore has
to be open to globalisation”, The Straits Times 18 October 2008).
11
Yap, Mui Teng, “Ultra-low fertility in Singapore: Some observations”, in Ultra-low Fertility in Pacific Asia: Trends,
causes and policy issues, ed. Gavin Jones, Paulin Tay Straughan and Angelique Chan. Abingdon, Oxon: Routledge, 2009,
160-180.
12
Wong Kan Seng, DPM’s Speech on Population at the Committee of Supply, 5 February 2009.
13
For a description, see P. Arumainathan, Report on the Census of Population 1970, vol 1, Singapore: Department of
Statistics, chapter5, 31-40.
14
Singapore Department of Statistics. Census of Population 2010 Advance Census Release. August 2010. Available at
http://www.singstat.gov.sg/pubn/census2010.html.
15
Zakir Hussain, “Malay leaders worry about numbers”, The Straits Times, 2 September 2010.
226
Agencies (EA) Act. These three acts regulate the entry and exit of foreigners and spell out the rights
and obligations of employers and migrant workers and those of the employment agencies involved in
the recruitment of migrant workers16.
Immigration control is the purview of the Immigration and Checkpoints Authority (ICA),
which is a division of the Ministry of Home Affairs (MHA). Immigration permits are issued by ICA
which also conducts checks on immigration offenders, that is those who have entered the country
without the relevant passes and permits or who have overstayed their immigration permit. The ICA
works closely with two other divisions in the MHA, the Singapore Police Force (SPF) and the Police
Coast Guard (PCG) in its enforcement of Singapore’s immigration laws. It also works closely with
the Ministry of Manpower (MOM) in the detection and prosecution of the illegal immigrants and
over-stayers and those who employ or harbour immigration offenders.
The Immigration Act provides that if an immigration offender is found at a work place, the
“occupier” of the work place shall be presumed to have permitted the immigration offender to enter or
remain there and to have knowledge that he is an immigration offender. The Act provides for
penalties including heavy fines and/or imprisonment for anyone who knowingly shelters an
immigration offender. Landlords and employers are required to check the authenticity of their
potential tenants and employees’ passes. Housing agents are also liable for their roles as
intermediaries.
Regulation of the employment of foreigners in Singapore, including the rights and obligations
of employers, intermediaries and the migrant workers themselves, falls under the Ministry of
Manpower (MOM). MOM issues work passes and enforces the regulations spelled out in the EFMA
and its schedules of work pass conditions, including overseeing the wellbeing of migrant workers
while they are in Singapore. It also regulates employment agencies involved in the recruitment of
foreign workers into the country, as set out in the Employment Agencies Act and its schedule of rules
and licensing conditions. MOM also works with other agencies in Singapore including the housing,
physical planning and environmental authorities, the labor movement, non-governmental welfare
organisations, and embassies of the sending countries in carrying out its work.
Work passes are a key instrument for regulating the employment of foreigners in Singapore.
The EFMA requires foreigners who wish to work in Singapore to possess work passes issued by
MOM’s Work Pass Division (Table 3, next page). The three main types of work passes are the
Employment Pass (EP), Work Permit (WP) and S Pass.
Employment Pass (EP)
EPs are issued to foreign professionals, specialists, middle management and highly qualified
foreigners. They may be issued with P1, P2 or Q1 Passes, depending on their basic monthly salary,
qualifications, skills and experience (Table 4).
As with all work passes, employers are required to apply for EPs on behalf of the foreigners
they wish to employ and the passes issued are specific to the employer and the foreigner. Should the
EP holder change job, the new employer will have to apply for a new pass on his/her behalf. An EP
holder must leave the country upon expiration of his/her contract or EP unless issued with a
Personalised Employment Pass (PEP) which allows the holder to remain in Singapore for up to six
months to look for a job. PEP applicants must have been P or Q1 Pass holders with some years of
work experience in Singapore or foreigners with salaries that qualify for a P1 Pass. PEPs are valid for
five years and are non-renewable. PEP holders enjoy the privileges provided by their previously held
status and are subject to a minimum annual salary requirement of S$30,000 (i.e. at least S$2,500 per
16
Selected categories of migrant workers are also covered by legislations that govern the employment of local
workers, such as the Employment Act and the Workmen’s Compensation Act.
227
month, the minimum salary that qualifies them for an EP). The purpose of the PEP is to facilitate
Singapore’s quest to attract and retain foreign talent.
There is no ceiling on the number of EP holders that a company may employ, nor is there any
restriction on the source countries that they may come from. Employers are also not required to pay a
foreign worker levy for the EP holders they employ.
Work Permit (WP)
Foreigners who command monthly salaries of no more than S$1,800 (generally the low-
skilled and mid-skilled workers) can be brought in to Singapore to work under the WP or R Pass17.
Access to such workers is, however, restricted to sectors that have difficulties recruiting Singaporean
workers, specifically, the construction, manufacturing, marine, process and services sectors (Table 5).
Dependency ceilings, specifying the ratio of foreign to local workers that a firm may employ, vary by
sector depending largely on the difficulty of attracting local workers. There is also source country
restriction for better integration of the foreigners (Table 6).
Employers are required to pay a foreign worker levy for each WP holder hired. The levy acts
as a price mechanism to regulate the demand for such workers who, because of their willingness to
accept lower wages, could out-compete local workers in terms of cost to employers. The levy payable
depends on the worker’s skills level and the industry sector where he/she is deployed. From July
2010, the levy will be raised in stages (Table 7) to encourage companies to reduce reliance on such
workers and invest instead in boosting skills and productivity18.
17
The current salary ceiling was adopted in 2004 when the S Pass was introduced.
18
Tharman Shanmugaratnam, “Budget Speech 2010 Towards an Advanced Economy: Superior Skills, Quality
Jobs, Higher Incomes”, Accessed
http://www.mof.gov.sg/budget_2010/speech_toc/download/FY2010_Budget_Statement.pdf
050510.
228
Table 3: Divisions in Ministry of Manpower involved in Foreign Manpower Matters
Ministry of Manpower
Work Pass
Foreign Manpower Division
Management Division
The Division
Core functions of the facilitates and
FMMD include: International Manpower regulates the
Management and Manpower Planning and Policy employment of
protection of foreign Division Division foreign nationals in
manpower, particularly Singapore … through
To facilitate the To develop a globally the administering of
in: illegal employment/ competitive workforce
deployment; attraction of skilled three types of Work
global manpower which meets industry Passes. Mission: to
accomodation/ needs, through
employment conditions/ including overseas develop a foreign
Singaporeans. it balancing sources of manpower admission
physical wellbeing; manpower supply, as
abandonment/ runaway. works with industry framework that meets
to understand well as an efficient the needs of the
Professionalism of the and flexible labour
Employment Agency manpower needs and Singapore economy.
oversees a network market.
industry through
overseeing EA licensing of Contact Singapore
rules and conditions; offices that promote
enforcement of EA Act Singapore as an
and Regulations; and attractive job Foreign Workforce
developing and destination. Policy Unit
implementing the points The unit is involved
demerit system. with broad policy
Strengthening work to facilitate a
enforcement capabilities flexible and
through strengthening responsive labour
work processes related to market. At present,
prosecution and this includes a review
enforcement; and of the broad policy
conducting audits on framework for foreign
employment agencies manpower, as well as
and employers of Work work on wage
Pass holders. restructuring.
Employment Inspectorate Department

Department focuses on policing illegal employment,
illegal deployment and other violations of MOM’s
foreign manpower regulations.
Well-Being Department
Department focuses on management and protection of foreign
manpower, particularly in the areas of accomodation, working
conditions, physical wellbeing and abandonment/ runaway.
Department also processes Employment Agencies’ licenses and ensures
that Eas comply with the Employment Agency Act Rules and License
Conditions.
Planning and Organisation Development Department

Department educates and raises awareness of employers, foreign workers
and the public on FMMD policies. It also develops policies and doctrines
and build capabilities within the division.
229
Source: MOM website.
Table 4 Types of Passes, Eligibility and Benefits
Criterion and Restrictions Salary Range(1) Dependent Long Term

Social Visit Pass
Type of Pass Pass
1. P Pass
a. P1 Pass  For professionals, Basic salary Eligible Eligible

managers, executives and >$7000
specialists.
b. P2 Pass  For professionals, Basic Eligible Eligible

managers, executives and salary
specialists. >$3500
2. Q Pass
a. Q1 Pass  For those who possess Basic salary Eligible Not

recognize qualifications >$2500
or skills and years of eligible
experience.
3. S Pass  For middle level skilled Basic salary Basic salary ≥ Not
manpower. $2500 eligible
≥ $1800 eligible
4. R Pass (Work Permit)
a. R Pass  Foreign workers who do Basic salary Children of WP Not

not qualify for S Passes Holders who wish
 Source country ≤ $1800 to study in eligible
restrictions apply Singapore
 Companies employing R
national schools
pass holders subject to
levy & dependency must pass a
ceiling. qualifying test.
b. R Pass  For FDWs who wish to Not Not

work in Singapore
(FDW) households. eligible eligible
Source: Ministry of Manpower website www.mom.gov.sg
(1) With effect from 1 July 2011, the minimum salaries will be raised to S$8000 for P1 Pass, S$4000 for P2 Pass,
S$2800 for Q1 Pass and S$2000 for S Pass.
230
Employers are also required to post an S$5,000 security deposit for each non-Malaysian foreign
worker hired. The deposit will be refunded upon the cancellation of the Work Permit and repatriation
of the foreign worker. From January 2010, only half of the security deposit would be forfeited if the
employer had made “reasonable” effort in locating their foreign workers who have absconded.
Foreigners without a Work Permit are prohibited from entering or remaining at the work place and
will be repatriated upon settlement of all outstanding wages or monies. The “occupier” of the work
place is presumed to know about the presence of such persons and subject to penalty for
contravention. In fact any foreigner found on any premises (including dormitories) is presumed to be
employed by the occupier of the premises unless proven otherwise. This is to stem the problem of
illegal employment.
Under the conditions of the WP, the foreign worker may only work for the employer
specified in the permit. Moreover, the worker may not be deployed in jobs or sectors other than that
specified in the Work Permit. This is to prevent exploitation of the foreign worker. Any breach of
these conditions constitutes illegal employment. Employer responsibility also includes providing safe
working conditions and prompt payment of salary, and in general making sure that the provisions of
the Employment Act19 are complied with. However, foreign domestic workers (FDWs), like their
local counterparts, are not covered by the Act.
Employers are also responsible for their WP holders’ upkeep and well-being, including
providing appropriate medical care and accommodation. Medical insurance coverage has been raised
from $5,000 when it was first introduced in 2008 to $15,000 in January 2010. WP holders are also
entitled to compensation for injuries sustained in the course of their job under the Workmen’s
Compensation Act. Employers are also to maintain a register of the addresses of their foreign workers
and report changes to MOM.
WP holders are not allowed to bring their family with them on Dependant’s or Long Term
Visit passes. However, the children of WP holders are allowed to study in Singapore national schools
if they pass the qualifying tests. The conditions of Work Permit also prohibit marriage between WP
holders and Singapore citizens or permanent residents without the prior approval of the Controller of
Work Permits. Failure to do so would result in repatriation and a non-permanent ban on re-entering
Singapore. Foreign domestic workers may not become pregnant or deliver a child while employed in
Singapore.
S Pass
S Passes are issued to foreigners classified as mid-level skilled workers. Based on a points
system, applicants are assessed according to a range of factors such as their qualifications, skills,
employment type, work experience, and a basic monthly salary of at least S$1,800. In 2009, the
educational qualification for eligibility for an S Pass was raised from upper secondary or a diploma to
a degree or diploma in line with the government’s policy to raise the quality of the workforce and also
as a response to concerns that a liberal S Pass policy could take away job opportunities for qualified
Singaporeans who were facing prospects of retrenchment and job loss20.
There is a quota on the number of S Pass holders that a company may employ, which is
currently at 25% of total workforce. The S Passes granted are part of the quota of the company’s total
WPs. Employers pay a foreign worker levy though at a much lower rate than for WP holders they
employ. They must also purchase medical insurance for their S Pass employees.
S Pass applicants with fixed monthly salaries of more than S$2,500 a month (the equivalent
of a Q Pass) may apply for Dependant's Passes for their family members.
19
The Employment Act covers the basic terms and working conditions of all employees in Singapore except those employed
in managerial, executive or confidential positions earning above S$2,500 per month, seamen and domestic workers.
20
Aaron Low, “Stricter criteria for semi-skilled foreigners”, The Straits Times 14 February 2009.
231
S Pass holders who earn less than this amount are not allowed to bring their dependants with them.
The children of S Pass holders may, as with the children of R Pass holders, study in mainstream
public schools in Singapore if they passed the relevant tests.
Other Passes
In addition to the above, a range of other passes also allow foreigners to work in Singapore
for various durations. The EntrePass is issued to foreign entrepreneurs interested in setting up
businesses in Singapore and intending to be involved in the daily operation of the business.
EntrePasses fall into two categories, namely P and Q Passes, and holders are accorded privileges
similar to those of EP holders. Short Term Employment Passes allow the entry of foreigners who are
working on certain projects or assignments. Student Pass holders are permitted to work without the
need to apply for a work pass, but within certain stipulations. Foreign students from selected schools
or academic institutions are permitted to take on full-time employment during their school vacation or
part-time employment during the school term. Those awaiting their examination results may also take
up employment provided they are able to attain permission from the Controller of Work Permits to do
so. Dependent’s Pass holders are allowed to work so long as a Letter of Consent has been granted by
the Ministry of Manpower, but dependants of S Pass holders are required to apply for work passes.
Long Term Social Visit Pass holders are only allowed to work if they are similarly successful in
obtaining the relevant work passes.
The procedures for acquiring work passes are as follows. In general, an employer, either on
his own or with the assistance of an employment agency, identifies suitable candidates, and then
makes an application to MOM’s Work Pass Division. If the application is approved, an In-Principle
Approval letter would be issued. The issuance of the actual work pass takes place only after
requirements, such as the furnishing of the security deposit and proof of medical insurance by
employers, and the foreign worker passing of the requisite tests (including medical examination by a
doctor in Singapore), have been completed. Failure to pass the tests could result in repatriation.
Enforcement of Singapore’s foreign workers regulations post-entry is carried out by the

FMMD (see Table 3, page 232). Its Enforcement Inspectorate works with other enforcement agencies
in policing violations of MOM’s foreign manpower regulations21. It carries out checks on illegal
employment, illegal deployment of foreign workers to jobs or employers other than the one stated on
their Work Permits and other violations of the Ministry’s regulations. The Well-Being Department
focuses on the management and protection of foreign manpower in the areas of accommodation,
working conditions, physical well-being as well as abandonment and abscondment. The FMMD
works closely with other divisions in MOM such as Occupational Safety and Health Division, Labor
Relations and Workplaces Division to ensure that worker safety and other workplace conditions are
met. The department also processes Employment Agencies’ licences and ensures that these comply
with the Employment Agency Act and its accompanying rules and conditions.
Regulation of employment agencies which are an integral part of the management of the
deployment of foreigners in Singapore fall under MOM’s FMMD (see Table 3, page 232). These
agencies are subject to the Employment Agencies (EA) Act and its subsidiary legislation which
require that they be licensed and furnish a security deposit with the government. Violation of the
Employment Agency licensing conditions could result in the revocation of the license and forfeiture
of the security deposit. From April 2011, fines and prison terms for errant and unlicensed
employment agencies will be raised. It will also be an offence for employers to use the services of
unlicensed agencies and for licensed agencies to make employment-related applications on behalf of
unlicensed ones. A newly created Commissioner for Employment Agencies will have the power to
suspend and reinstate EAs. Measures are also introduced to improve the professionalism and
accountability of EAs. EAs placing foreign workers and foreign domestic workers FDWs will also be
subject to certain minimum service standards pertaining to their relationship with employers. Beyond
21
MOM website.
232
these, MOM encourages EAs to evolve and adopt higher standards and best practices and it will work
with the industry to develop a new accreditation scheme. MOM will remove the caps on fees charged
to employers. The agency fees charged to workers, however, may not exceed one month’s salary for
each year of contract, subject to a maximum of two months’ salary. EAs will have to refund 50% of
the fees charged to the worker if the contract is terminated by the employers within six months; no
refund will be made if the worker absconds or terminates the contract prematurely. EAs will be
required to issue to receipts to workers for all monetary transactions.
Issues and Assessment
Overall, Singapore’s foreign manpower policy has evolved, from one of using foreign
manpower, particularly lower-skilled and unskilled foreign workers, as a stop-gap measure to
acknowledgement of their place as an integral part of the Singapore workforce. Unfortunately, few
studies have been done examining the contribution that these foreign workers have made to GDP
growth in Singapore. However, many commentators, including economists, have acknowledged that
foreign labor has enabled Singapore to grow beyond its potential22.
One question that is frequently asked is whether the deployment of low-cost migrant workers
has depressed the wages of lower-income Singaporeans. According to the economist Manu
Bhaskaran, “(T)he problem is not that one group of people is seeing slower growth of income
compared to another group. What is worrying is (that) there is a group of people in Singapore, and
quite a large one, whose real income has actually fallen in the last few years. And this is much more
worrying than just saying the (income) gap has widened”23. For its part, the government has argued
that migrants have helped Singapore grow and created more jobs for Singaporeans, and that
Singaporeans’ wages grew over the period 2006-2008 when the foreign workforce was growing most
rapidly24.
It is too early to assess the effectiveness of the recently announced drive to reduce
dependency on migrant workers, particularly the low-skilled ones, and to increase productivity with
higher quality locals and foreigners. This drive is perhaps as much a political move as an economic
one following the growing discomfort of Singaporeans regarding the influx of foreigners into their
country. According to a paper published by DBS (Development Bank of Singapore) Research Group,
a key determinant will be the elasticity of demand for such labor. The group was of the opinion that
“it is not easy to replace low skilled foreign workers with higher skilled local workers (since) these
are jobs which Singaporeans shun in the first place”, although it also added that “some level of
substitution is possible in the mid-skilled category”25.
There is an apparent “mismatch” or difference in expectations on the parts of Singaporean

employers and workers. Employers charge that local workers are too picky and turn their backs on
jobs requiring shift and weekend work and that they avoid working in the retail sector because they
believe that this sector is only for those who cannot find jobs elsewhere. Employers also claim that
local workers have turned down job offers when the wages offered were below what was deemed a
“living wage.” While professing to prefer local workers if they are available, employers have also
noted that foreigners are more willing to put in extra hours and have better service attitudes26.
22
See, for example, Hui Weng Tat, “Foreign Manpower Policy in Singapore”, in Singapore Economy in the 21st Century;
Issues and Strategies, ed. Koh Ai Tee et al. (Singapore: McGraw-Hill, 2002), 29-50; Tan Kong Yam and his colleagues at
the Ministry of Trade and Industry estimated that foreign talent contributed 37% and foreign workers about 4% to quarterly
GDP growth over the period 1990-2000 (see Has foreign talent contributed to Singapore’s economic growth? An empirical
assessment, 2002).
23
Manu Bhaskaran, “Singapore Economy: Medium-Term Outlook” in Singapore Perspectives 2007; A New Singapore” ed.
by Tan Tarn How, Singapore: IPS and World Scientific, 2007, p 40.
24
Tharman Shanmugaratnam, Budget Debate Roundup Speech 2010, paraC7 p 18,
http://www.mof.gov.sg/budget_2010/download/FY2010_Budget_Debate_Round_Up_Speech.pdf, accessed 14 April 2010.
25
DBS Research Group, Singapore: The economics of the Foreign Worker Levy Hike, 17 March 2010.
26
Cassandra Chew, “Foreign workers not cheaper”, The Straits Times 5 September 2010, accessed 4 September 2010 (US).
233
Because of its low birth rate, there are also real limits to expanding the domestic supply of
labor in Singapore. Delaying retirement has been proposed as one way to increase the domestic labor
supply. In this regard, the government has said that it will introduce re-employment legislation in
2012 to require employers to offer opportunities for re-employment to their staff who have reached
the current retirement age of 62 years. The extension will be to age 65 in the first instance, with
possible future extension to age 67. More employers have reportedly already begun offering re-
employment to their older workers ahead of the proposed legislation27. Attracting the return of
women who have left the workforce for domestic responsibilities has also been suggested. However,
encouraging the return of women to the workplace seems to be left to the National Trades Union
movement, in particular its Women’s Development Secretariat. Programs have been instituted to
equip women to return to the workforce. However, these programs lack the legislative support of the
initiative on older workers. A question that may be asked is whether these older workers and women
are substitutes for the foreign workers hired. They are unlikely, for example, to be able to replace
migrant workers in the construction and marine sectors.
In Singapore levies and dependency ceilings have been used to control the number of foreign
workers (Table 5 and Table 6 next page). However, these levies and dependency ceilings have had to
be revised upwards many times over the years, suggesting that either they have not been effective or
not set at sufficiently high levels. The use of levies has been argued as affording greater flexibility
than the use of dependency ceilings alone, because employers who need more foreign workers are
able to get them by paying the higher levy. This is also preferred to a tender system which provides
less certainty to employers. It is seen as a better way to allocate labor28.
As noted earlier, the management of foreign manpower in Singapore involves multiple

government agencies besides the Ministry of Manpower. As is the practice with other multi-
dimensional national issues, an inter-ministerial committee was set up in 2009 to oversee foreign
worker issues. It is not possible to know the relationship between agencies or how issues are resolved
because such information is not publicly available. Discussions are typically carried out behind
closed doors and not publicly accessible. Generally, it can be assumed that they work towards
cohesion and social harmony for the economic prosperity of the nation. An example of co-operation
is the arrangement between MOM and ICA whereby migrant workers with outstanding issues who are
being forcibly sent home by their employers will be able to inform immigration officers at the point of
departure of their situation. In such instances, the worker will not be sent home but will instead be
referred to MOM for assistance29.
The recent influx of foreigners and the apparent inadequacy of infrastructure to meet the
needs of the population (resulting in complaints of overcrowding and rising costs) appear to signal a
lack of coordination between agencies. However, it has also been argued that the influx had not been
anticipated and thus was not provided for in physical planning30. This was due partly to the speed
with which the government has had to respond to the heightened demand for workers during the years
of high economic growth. Another reason given was that the government decided to take advantage
of the surge in applications for permanent residency in the aftermath of the 2008-2009 economic crisis
to augment the population, arguing that it has not always been easy to attract immigrants.
Over the years, the government has made various responses to the concerns of the
Singaporean public. It has acted on the concerns raised over municipal and law and order issues
arising mainly from the weekend gatherings of foreign workers at certain locations by raising police
and voluntary vigilance and encouraging responsible behaviour. In response to residents’ concerns
about locating a foreign workers’ dormitory in a middle class residential neighbourhood, the
27
Amanda Lee, “64% rehiring older workers”, The Straits Times 6 August 2010.
28
Alicia Wong, “Quota cut and thrust”, Today online 3 March 2010.
29
Hawazi Daipi, Parliament 13 February 2009.
30
Zaini Hassan, “Take rational approach to problems: SM Goh”, The Straits Times 7 September 2010.
234
Table 5 Approved Sectors, Levies and Dependency Ceilings by Sector (before July 2010)
Dependency Ceiling Category of Foreign Levy Rate ($)

Sector
(DC) Worker Monthly Daily
Up to 40% of the total Skilled 150 5
workforce Unskilled 240 8
Above 40% to 55% of Skilled 150 5
the total workforce Unskilled 280 10
Manufacturing
Above 55% to
65% of the
*Skilled/Unskilled 450 15
total
workforce
1 local full- Skilled 150 5
time worker ****Experienced &
Construction 300 10
to 7 foreign exempted from MYE
workers Unskilled 470 16
1 local full-time worker Skilled 150 5

Marine
to 5 foreign workers Unskilled 295 10
Skilled 150 5
1 local full-time worker ****Experienced &

Process 300 10
to 7 foreign workers exempted from MYE
Unskilled 300 10
Up to 30% of the total Skilled 150 5

workforce Unskilled 240 8
Services Above 30% to 40% of

the total workforce
Above 40% to 50% of
the total workforce
Normal rate 265 9

Domestic Worker NA
**Concessionary rate 170 6
***25% of the
S Pass Holder Skilled 50 2
total workforce
Notes:*Skilled workers belonging to the categories of above 55% to 65% of the total workforce in the Manufacturing Sector
and above 30% to 50% of the total workforce in the Service Sector are not eligible for the skilled levy rates.
**An employer of a Foreign Domestic Worker (FDW) is eligible for low levy for each FDW if he/she satisfies Conditions
A, B, C or D below at the time of application for levy concession: (A) (i) the employer or spouse has a child/grandchild who
is a Singapore Citizen below 12 years old; or (B) (i) the employer or co-residing spouse is a Singapore Citizen who is aged
65 years old or above; or (ii) the employer or spouse is a Singapore Citizen and the other party is a Singapore Permanent
Resident aged 65 years old or above, and both are living together at the same registered address as in NRIC; or (C)(i) the
employer or spouse has a parent, parent-in-law, grandparent or grandparent-in-law who is a Singapore Citizen aged 65 years
old or above, and is living with him/her at the same registered address as in NRIC; or (ii) the employer or spouse is a
Singapore Citizen and has a parent, parent-in-law or grandparent or grandparent-in-law who is a Singapore Permanent
Resident aged 65 years old or above, and is living with him/her at the same registered address as in NRIC or (D) (i) With
effect from 15 September 2007, the FDW levy concession will be extended to FDW's employers with disability or who have
family members with disability and require a full-time caregiver's assistance in Activities of Daily Living. For more
information on the eligibility criteria and application procedures, please visit the Centre of Enabled Living (CEL) .
235
For each condition that is satisfied, the employer is eligible for levy concession for one FDW. However, each household is
only eligible for levy concession for a total of two FDWs at any one time. *** For all sectors, S Passes' Dependency Ceiling
(DC) of 25% will be counted within the Work Permits' DC.
**** The monthly levy rate of $300 only applies to Non-Traditional Sources (NTS) or People's Republic of China (PRC)
workers in the construction and process industries, who have been exempted from the requirement of Man-Year Entitlements
(MYEs). To be exempted from MYE, the foreign worker must have at least two years' working experience in his respective
industry and his employer has applied for a Work Permit without any Prior Approval support. Source: Ministry of
Manpower website “Foreign Worker Levy Rates”
http://www.mom.gov.sg/publish/momportal/en/communities/work_pass/work_permit/application/requirements/foreign_wor
ker_levy.html
Table 6 Foreign Worker Levies and Dependency Ceilings 2010-2012
Sector Dependency Ceiling (DC) Worker Category Monthly Levy Rate

($)
Manufacturing Basic Tier / Tier 1: Up to 35% of total Skilled 160

workforce
Unskilled 260
Tier 2: above 35% to 55% of total Skilled 180

workforce
Unskilled 280
Tier 3: above 55% to 65% of total Skilled(1) 450

workforce
Unskilled(1) 450
Construction 1 local full-time worker to 7 foreign Skilled and on 160

workers MYE(2)
310
Experienced &
exempted from 470
MYE(3)
Unskilled
Marine 1 local full-time worker to 5 foreign Skilled 160

workers
Unskilled 300
Process 1 local full-time to 7 foreign workers Skilled and on 160

MYE(2)
310
Experienced and
exempted from 300
MYE(3)
Unskilled
Services Basic Tier / Tier 1: Up to 25% of total Skilled 160

workforce
Unskilled 260

workforce
Unskilled(1) 300
236
($)

workforce
Unskilled 450
Notes:
(1) Skilled workers belonging to the categories of above 55% to 65% of the total workforce in the Manufacturing
sector and above 25% to 50% of the total workforce in the Services sector are not eligible for the skilled levy
rates.
(2) The MYE allocation quota for Construction and Process sectors will be reduced by 5%.
(3) The monthly levy rate of $310 only applies to Non-Traditional Sources (NTS) or People’s Republic of China
(PRC) workers in the Construction and Process industries, who have been exempted from the requirement of
Man-Year Entitlements (MYEs). To be exempted from MYE, the foreign workers must have at least two years’
working experience in his respective industry and his employer has applied for a Work Permit without any Prior
Approval support.
S Pass Levies

($)
All sectors Basic Tier / Tier 1: Up to 20% of total Skilled 100

workforce(1)

workforce(1)
Notes: (1) For all sectors, S Passes Dependency Ceiling of 25% will be counted within the Work Permits’ Dependency
Ceiling. Source: MOM website.
on GDP at 2005 Market Prices and Real Economic Growth Rate (%)
government made some modifications but still proceeded with its plan to convert an unused school
into a dormitory. As mentioned, since 2009, the government has also responded to Singaporeans’
unhappiness with the influx of foreigners by sharpening the difference in the benefits received by
citizens, permanent residents and foreigners. However, it has also to be careful not to make the
conditions for foreigners so onerous as to deter their interest in coming to work in Singapore.
There is a strong tripartite relationship between the government, employers and the labor
movement in Singapore which to all intents and purposes has served the country well. The scope of
this collaboration extends to matters involving foreign manpower. The government has been
responsive to business’ or employers’ need for access to migrant workers, primarily because this
could have major economic implications and because of the reliance on foreign investment for
Singapore’s economic growth. Singapore’s attractiveness as a place for investment by foreign MNCs
is also due to the ease of access to skilled workers, the result of the foreign talent policy. It is also
easy for the foreign MNCs to bring in foreign talent. The government has also helped businesses
during economic downturns by lowering the foreign worker levy to reduce their operational cost. The
labor movement under the leadership of the National Trades Union Congress (NTUC) has also been
responsive to the needs of government and employers. Instead of confrontation, negotiation has been
the key. This relationship is perhaps helped by the fact that government ministers and politicians of
the ruling party are also leaders of the NTUC. As Members of Parliament, they are able to voice their
237
concerns and that of their constituents in Parliamentary sessions, although when an issue is put to the
vote, this is generally along party line unless the Whip is lifted (but this done so only on rare
occasions). An example of the labor movement’s collaboration with the Singapore National
Employers Federation is the setting up of the Migrant Workers Centre by both these parties to provide
support for foreign workers. The latter is a contact point and has a hotline for access to information
on foreign workers’ rights and how to seek help. It also provides humanitarian support to stranded
workers, organises basic training in areas like English and social activities to help fit in.
Government agencies have also responded to foreign workers’ needs when problems have
surfaced, for example, when they protested about salary arrears or when they are found to have been
abused. MOM helps foreign workers recover monies owed by employers. Foreign workers may also
approach MOM for dispute settlement. However, these cases could be just the tip of the iceberg.
Voluntary organisations that serve foreign workers have been concerned about the vulnerability of the
low skilled foreigners, and in particular, the foreign domestic workers who are required to live with
their employers. In this regard, it may be noted that Singapore’s Penal Code has been revised to
inflict heavier punishment on abusers of foreign domestic workers (FDWs). Some sticking issues
remain unresolved. The issue of rest days for FDWs and their inclusion under the Employment Act
remain unsettled. Accreditation bodies for employment agencies engaged in placing FDWs have a
standard contract whose terms are to be agreed on between employer and worker. The contract
includes agreements on wages, rest days and payment to be paid by the employer in lieu of rest days.
However, enforcement of the contract actually depends on the worker making a complaint to the
ministry if there is malpractice or abuse. A postage paid envelope addressed to MOM is given to each
first-time FDW at their safety awareness course31. The FDW, usually with a debt to repay, may be
willing to accept lesser conditions or are simply too afraid to risk termination of employment and
premature repatriation.
Another area of concern has been with regard to recruitment fees. As mentioned earlier,
recruitment or employment agencies are an integral part of the scene, especially at the lower end of
the occupational spectrum. These agencies work with counterparts in sending countries who identify
potential workers who are interested in seeking employment in Singapore. As mentioned, the
Employment Agencies Act sets caps on the amount of fees that Singapore agencies can charge
workers and employers. However, the larger component of the placement fee is the charges accruing
to recruitment agencies in the home countries. Anecdotal evidence suggests that these fees range
from $2,000-$10,000, depending on the source country and the number of intermediaries involved32.
TWC2, a voluntary organisation that assists foreign workers, found in their study of Indonesian
FDWs that the recruitment fees charged have risen substantially over the years. This in turn means
that foreign workers are taking a longer time to repay their migration “loans.” The Singapore
government’s position on this matter is that it has no jurisdiction over transactions that took place
outside its territory33. MOM, however, shares information obtained in the course of its investigations
with the embassies of sending countries34.
The elimination of recruitment malpractices has been adopted by the countries of the
Association of Southeast Asian Nations (ASEAN) as an item on the agenda of its Socio-Cultural
Community35 . Dialogues between member countries have begun with a workshop in Singapore in
April 2010 with the objectives of allowing participants to share challenges and best practices and to
propose recommendations. Among the recommendations at the national level are: enforcement of
existing laws against recruitment malpractices, education and information, making available channels
for reporting grievances, and encouraging countries to enter into government-to-government
memorandum of understanding.
31
“Default in payment of foreign domestic workers’ salaries”. Oral answers to questions. 19 May 2010.
32
“Job placement for foreign workers (Prevention of exploitation)”, Written answers to questions, 23 November 2009.
33
MOM COS [speech 3], 13 February 2009.
34
The Indonesian government has reportedly proposed to cap the recruitment fees payable to Indonesian agencies for FDW
placements in Singapore at $1000 (“Jakarta proposes new rules on hiring maids”, The Straits Times 9 March 2011).
35
See http://www.aseansec.org
238
As noted above, Singapore is not free of problems of illegal immigration and illegal
employment. However, because of strict enforcement, the number of illegal immigrants in Singapore
has declined in recent years. Employers are made responsible for illegal immigrants found on their
premises. They are assumed to be aware of any foreign illegal immigrants or workers present and
penalised accordingly. Construction companies have to fence in their sites and maintain registers at
every entry point. Harbouring and conveying of immigration offenders are crimes punishable by law,
the former including renting of accommodation to illegal immigrants. Illegal immigrants are liable to
be caned. Outreach programs to warn prospective migrants of the penalties of illegal immigration
have included sending information to source countries. Similarly, there are penalties for illegal
employment. An issue that seems to have grown in recent years is the increase in foreigners who
have been caught falsifying educational qualifications in their work pass applications in order to
qualify for a higher level work pass36. Foreigners who do this are likely to be repatriated, and
employers who aid in this process are liable to be fined or jailed.
Recently, the Philippines government has tightened its implementation of an existing law
requiring Filipinas who are departing the country for domestic work to go through the Philippines
Overseas Employment Agency (POEA). The Philippines government also requires that the departing
workers have signed contracts indicating that they will receive the monthly salary and number of rest
days required by the government. While the Philippines, as the labor-sending country, considers
departures without the approval of the agency to be illegal37, from the receiving countries’
perspectives, however, the workers are legal so long as they have met their own requirements. This
example points to the need to harmonise legislation and regulations between labor-sending and
receiving countries.
Tighter implementation of Philippine laws on departure for domestic work overseas has
resulted in a shortage of such workers in Singapore38. This could result in better terms for these
workers. However, the Philippines could also face the problem of being replaced in the global
marketplace by competitors who are willing to accept less pay. In Singapore, the Association of
Employment Agencies has already begun sourcing for domestic helpers from Bangladesh39. In recent
years, Filipinas have lost their dominance in the domestic work market in Singapore by being
replaced by cheaper Indonesians. However, according to maid agencies, the lack of facility in English
could prevent Bangladeshi women from making a similar inroad in the Singapore market.
4. CONCLUSION AND RECOMMENDATIONS
Foreigners currently make up over one third of Singapore’s work force and over one quarter
of its five million population. With such large numbers, it is quite inevitable that there will be
aberrations and deviations from the prescribed norms, from rent-seekers taking advantage of unequal
access to information to risk-takers who, perhaps out of desperation, try to gain access to work
illegally in Singapore.
The reform that seems to be needed most urgently in the foreign manpower system is to
ensure that access to true and accurate information is available to all parties involved in the migration
process: foreign workers, local employers and employment agencies. In this regard, co-operation
between labor-sending and receiving countries is essential. Accurate information on work and
migration opportunities should go down even to the most rural areas in labor-sending countries where
migrant workers typically come from. This should go some way towards solving the problem of
36
“Hundreds with fake degrees nabbed”, The Straits Times 8 August 2008; “More caught for lying in work pass
applications”, The Straits Times 11 February 2011.
37
Melissa Sim, “Harder to hire Filipino maids now”, The Straits Times 18 August 2010.
38
Melissa Sim, “Harder to hire Filipino maids now”, The Straits Times 18 August 2010.
39
Melissa Kok and Kimberly Spykerman, “Maid agencies eye Bangladesh”, The Straits Times 2 September
2010.
239
excessive debt burdens borne by poor, unskilled foreign workers who seek to work in Singapore.
Implementation of policy agreements in both labor-sending and receiving countries is also important
as it would be quite unfruitful if agreements are implemented only at one end.
Singapore has made a start in this area, by issuing information leaflets to foreigners who have
received in-principle approval to work in Singapore. Singapore also makes information available on
the MOM website regarding the regulatory framework, and procedures and administrative costs
involved in obtaining a work pass. This MOM website also includes publications on labor market
conditions in Singapore. One publication, for example, lists the jobs in demand in various sectors, the
salary available in each sector, as well as the qualifications and experience required. The jobs range
from the top professional category offering $7,000 or more per month, to clerical jobs and low-skilled
jobs. This is available for all prospective migrant workers to see. Activists and interested parties
should help make this information more available to potential job seekers in labor-sending countries.
REFERENCES
Bhaskaran, M. 2007. “Singapore Economy: Medium-Term Outlook” in Singapore Perspectives

2007; A New Singapore” ed. by Tan Tarn How, Singapore: IPS and World Scientific
DBS Research Group, 2010. Singapore: The economics of the Foreign Worker Levy Hike, Singapore.
Fong, P.E. Tan Chwee Huat, and Cheng Soo May, 1989. “The Management of People” in
Management of Success; the moulding of modern Singapore, ed. Kernial Singh Sandhu and Paul
Wheatley, Singapore ISEAS
Hui, Weng Tat. 2002. “Foreign Manpower Policy in Singapore” in Singapore Economy in the 21st
Century; Issues and Strategies, ed. Koh Ai Tee, Lim Kim Lian, Hui Weng Tat, Bhanoji Rao and
Chng Meng Kng . Singapore: McGraw-Hill Education.
Ministry of Manpower. 2009. Labour Market Report 2009 (released in Singapore on 15 March
2010). http://www.mom.gov.sg/Home/MRSD/Documents/GLM/qtlmr094.pdf (accessed 16 March
2010).
Ministry of Trade and Industry. 1986. Report of the Economic Committee, The Singapore Economy:
New Directions, February 1986
Shanmugaratnam, T. 2010. “Budget Speech 2010 Towards an Advanced Economy: Superior Skills,
Quality Jobs, Higher Incomes”, Accessed
http://www.mof.gov.sg/budget_2010/speech_toc/download/FY2010_Budget_Statement.pdf
050510.
Yap, Mui Teng. 2009. “Ultra-low fertility in Singapore: Some observations”, in Ultra-low Fertility in
Pacific Asia: Trends, causes and policy issues, ed. Gavin Jones, Paulin Tay Straughan and Angelique
Chan. Abingdon, Oxon: Routledge,
240
Chapter 11: The Management of Foreign Workers in Malaysia:
Institutions and Governance Regime
AZIZAH KASSIM,
Universiti Kebangsaan Malaysia
TERENCE TOO,
STEPHEN C. M. WONG
Late MAHANI ZAINAL ABIDIN

Institute of Strategic and International Studies (ISIS) Malaysia*
ABSTRACT: Malaysia is both a labor-sending and a labor-receiving country, but it receives far
more foreign workers than it exports. In 2009, the number of legally recruited foreign workers was
over 1.9 million or 16.8% of the labor force, with an estimated additional 0.6-1.0 million irregular or
unregistered foreign workers. Given the very large number of foreign workers in Malaysia, it would
seem that the institutions and governance structure for managing these workers should be efficient
and effective. However, foreign worker policy in Malaysia has struggled to keep pace with the
rapidly changing realities on the ground. This paper examines the system for managing foreign
workers in Malaysia, identifying the main characteristics and achievements of this system, and
suggesting ways for improving it. On the whole, it seems clear that Malaysia’s policy institutions and
instruments for managing foreign workers require significant review and reform. In particular, policy
strategies need to be devised to minimize the negative aspects of foreign worker migration and to
ensure that these foreign workers are used in a manner that is fully consistent with the objective of
becoming a high-income economy.
1. OVERVIEW OF INTERNATIONAL LABOR MIGRATION IN MALAYSIA
Malaysia is both a labor-sending and a labor –receiving country, but it receives far more
workers than it exports. In the last ten years, the number of legally recruited foreign workers in
Malaysia has more than doubled, peaking at just over 1.9 million in 2009 or about 16.8% of the labor
force (Table 1). These figures exclude about 38,000 expatriate workers and between 0.6 and 1.0
million irregular workers.1 While the influx of foreign workers into Malaysia contributes positively to
economic growth, it also presents certain problems, the most serious of which is the large and
growing number of irregular migrant workers. This study focuses on unskilled and semi-skilled
foreign workers, both legal and irregular (subsequently referred to as foreign or migrant workers),
who form about 98 percent of the foreign work force in Malaysia.
1
The Department of Immigration estimates that, at any one time, the number of irregular migrants is about one third to that
of legally recruited migrant workers.
241
Table 1. Expansion of Legal Foreign Workers by State, Malaysia, 1999-2009*
Peninsula Sabah & Labuan Sarawak

Year Malaysia
(%)
1999 84.3 10.0 5.7 897,705
2000 80.9 8.4 10.6 819,684
2001 75.2 13.4 11.5 769,566
2002 76.9 14.1 9.1 1,057,156
2003 79.7 12.1 8.3 1,412,697
2004 78.4 13.5 8.1 1,474,686
2005 79.7 13.0 7.2 1,821,750
2006 80.8 11.8 7.3 1,871,038
2007 84.0 10.0 6.0 2,044,805
2008 70.0 15.0 15.0 2,062,596
2009 n.a n.a n.a 1,918,146
Notes: * Based on issuance of work passes, Pas Lawatan Kerja Sementara (PLKS) or in English, Visit Pass (Temporary
Employment), or VP(TE)
Source: Department of Immigration, Putrajaya and Azizah Kassim (2008)
Since the 1970s the number of legally recruited foreign workers has risen considerably, with
actual numbers fluctuating in response to Malaysia’s changing economic conditions. The New
Economic Policy (1971-1990) helped to shift the country from agriculture to manufacturing and to
usher in high rates of economic growth (average annual GDP growth rates of 3.1 percent in the 1980s
and 4.5 percent in the 1990s). As the economy expanded, and FDI-driven industrialization created
new employment opportunities, there was an increasing demand for foreign labor. In particular, as
more educated Malaysians moved into higher-paying jobs in the urban sector, foreign workers moved
to take advantage of vacancies in unskilled production and rural agricultural jobs.
In addition to these domestic changes, Malaysia’s rapid economic growth in the 1980s and
1990s gave rise to large disparities in living standards between Malaysia and neighbouring, labor-
supply countries. Over the years, per capita incomes in Malaysia grew to be about ten times higher
than those of the predominately agricultural economies of Indonesia, Myanmar and Vietnam
(Manning, 2006: 55-56). This development, coupled with political instability in these countries,
helped encourage the massive inflow of foreign workers into Malaysia, both legal and otherwise.
At present, foreign workers in Malaysia come from 14 different source countries. By far, the
most important source country is Indonesia. Because of its physical and cultural proximity to
Malaysia, Indonesia accounts for over 51 percent of all foreign workers in Malaysia (Table 2).
However, in recent years the share of foreign workers from Indonesia has been falling, and the
proportion of workers from other countries – like Bangladesh and Myanmar – has been rising. This
change in the source of foreign workers occurred after a riot by Indonesian foreign workers in Nilai,
Selangor in 2002. The authorities at the time felt that an overdependence of workers from a single
country had the potential to pose a threat to national security. Action was therefore taken to reduce
the quota for Indonesian workers and open Malaysia to other source countries.
242
Table 2. Foreign Workers by Country of Origin (%), 2001-2009
Countries 2001 2002 2003 2004 2005 2006 2007 2008 2009
Indonesia 73.7 71.4 70.0 n.a. 66.7 62.8 56.1 52.6 51.7
Bangladesh 13.7 7.3 8.4 n.a. - 3.6 10.6 15.3 16.6
Philippines 2.2 2.0 1.8 n.a. - 1.3 1.1 1.3 1.3
Thailand 0.3 2.2 1.9 n.a. - 0.7 0.9 1.0 1.0
Pakistan 0.3 0.2 0.2 n.a. - 0.6 0.8 1.0 1.2
Cambodia - 0.2 0.2 n.a. - 0.4 0.5 0.6 0.6
Nepal - 7.4 6.8 n.a. 10.6 11.4 9.3 9.8 9.5
Myanmar - 2.5 2.6 n.a. 4.9 5.8 5.1 7.0 7.3
Vietnam - 3.0 4.1 n.a. 4.5 5.7 5.6 4.3 3.8
Laos - - - n.a. - - - - -
Uzbekistan - - - n.a. - - - - -
India - 3.5 4.0 n.a. 7.6 7.4 7.0 6.3 6.4
Sri Lanka - 0.1 0.1 n.a. - - 0.2 0.2 0.2
China - - - n.a. - 0.1 0.3 0.5 0.5
Others 9.7 0.1 0.1 n.a. 5.7 0.2 2.5 - -
Total1 100.0 100.0 100.0 n.a. 100.0 100.0 100.0 100.0 100.0
Notes: 1Slight discrepancies may occur due to rounding
Source: Data from the Department of Immigration, Malaysia; Azizah Kassim (2005a); Dairiam (2006).
Table 3. Distribution of Foreign Workers by Sector (%), 2001-2009
Sector 2001 2002 2003 2004 2005 2006 2007 2008 2009
Domestic services (maids) 20.3 21.4 18.4 19.4 17.7 16.6 15.4 14.2 13.1
Construction 7.8 13.9 19.9 15.7 15.5 14.3 14.4 14.9 15.6
Manufacturing 36.8 31.5 29.3 32.4 32.1 34.6 35.9 35.3 34.6
Services 7.2 6.1 6.7 6.3 8.8 8.9 9.9 10.3 10.6
Plantations1 27.9 27.1 25.7 26.2 23.5 19.0 16.5 16.2 16.6
Agriculture - - - - 2.5 6.6 8.1 9.1 9.5
Total2 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Notes: 1Includes the agriculture sector for 2001-2004. 2Slight discrepancies may occur due to rounding
Source: Azizah Kassim, 2005; Dairiam, 2006; Kanapathy, 2008 & Department of Immigration, Malaysia, 2006, 2007, 2008.
Malaysia sanctions the recruitment of foreign workers in only six sectors of the economy:
domestic services, construction, manufacturing, services, plantations and agriculture. Of these sectors,
most foreign workers – 34.6 percent in 2009 – are in manufacturing (
Table 3). After manufacturing, construction, plantations, and domestic services comprise the next
three sectors of the economy with the highest numbers of foreign workers.
In Malaysia it is generally acknowledged that foreign workers are needed in such labor-
intensive sectors of the economy as construction and plantations, since the more educated local
workforce tends to shun these sectors. In addition, increasing rates of local female participation in the
formal workplace creates a growing demand for foreign workers in domestic services (i.e. maids) and
services. In 2009 about 24 percent of all foreign workers were in domestic services and services.
While Malaysia’s economy has benefitted from the employment of foreign workers in these and other
sectors, policymakers in the country are increasingly concerned that the number of unskilled foreign
workers has risen significantly in recent years, while the number of skilled foreign workers has
remained relatively stagnant.
As indicated earlier, there is a large pool of irregular migrants and workers in Malaysia. These
irregular workers include: (1) foreign nationals who entered the country without proper documents;
243
(2) children born to foreign nationals in Malaysia whose births were not registered; (3) foreign
workers whose work passes have expired and not been renewed; (4) refugees in Sabah whose work
passes have not been renewed; and (5)work pass abusers, contract defaulters, and overstayers. In
Malaysia, these irregular migrants are officially referred to as “illegal immigrants” (pendatang asing
tanpa izin or PATI), or “illegal workers” if they are employed. These illegal migrants have no rights
and are subject to deportation under the Immigration Act 1959/63.
Table 4. Identified Irregular Migrants (1991-2009)
Ops Nyah 2 Regula-

Year Ops Nyah 1 Amnesty Runaways Total
/Ops Tegas Risation
1992 483,784 483,784
1996 554,941 554,941
1997 413,812 413,812
1998 187,486 187,486
129,746
2000 909,473 72,528 72,528

2002 439,727 439,727
Oct 2004- 398,758 398,758
Feb 2005
2006
20091 161,747 161,747
Total 129,746 909,473 1,614,284 1,025,971 72,528 3,751,902
Source: Azizah Kassim (2008) and data from Department of Immigration Sabah.
Notes: 1. 2009 numbers are for Sabah only.
2. Ops Nyah 1 comprised border control measures including land and sea patrols; Ops Nyah 2/Ops Tegas refers to
operations to root out irregular migrants who failed to participate in regularisation or amnesty exercises.
3. Figures for regularisation and amnesty include dependents of migrant workers (children, wives and elderly parents).
While there are no reliable estimates on the exact number of irregular migrants, an indication
of their size can be gauged from the results of various measures taken against them by the
government. During the period 1992 to 2009, the government identified over 3.75 million irregular
migrants in various amnesties and regularisation exercises (Table 4). However, this figure only
includes a fraction of the total number of irregular migrants in the country since many irregular
migrants are able to avoid detection by the government. The Department of Immigration, Putrajaya
currently estimates that there are between 600,000 and one million irregular migrants in Malaysia.
Table 5. Irregular Workers by Country of Origin & Gender (2004-05)
Country Male Female Total

Indonesia 234,078 113,829 87.3%
India 16,538 1,452 4.5%
Bangladesh 8,414 73 2.1%
Philippines 4,032 3,546 1.9%
China 1,131 2,191 0.8%
Nepal 3,181 37 0.8%
Pakistan 2,876 24 0.7%
Myanmar 1,851 117 0.5%
Vietnam 1,747 192 0.5%
Cambodia 983 708 0.4%
Sri Lanka 773 65 0.2%
Thailand 350 295 0.2%
Others 216 59 0.1%
Total 276,170 122,588 398,758
Note: The breakdown for irregular migrants for the subsequent years is not available.
Source: Laporan Program Khas Pengampunan Tahun 2004-2005, pp: 39, Jabatan Imigresen Malaysia, 1st March 2005.
244
Most irregular migrants come from Indonesia. According to the results of a recent nationwide
amnesty exercise, over 87 percent of all illegal migrants come from Indonesia (Table 5). Indonesia
shares a long common border with Malaysia and so it is easy for undocumented migrants from that
country to come and work in Malaysia.
Table 6. Results of Amnesty Exercise by Economic Sector (29 Oct 2004 - 28 Feb 2005)
Sectors No. %
Construction 110,006 27.6
Plantations 98,887 24.8
Services 78,102 19.6
Domestic Helpers 40,898 10.3
Manufacturing 49,741 12.5
Others 21,124 5.3
Total 398,758 100.0
Source: Laporan Program Khas Pengampunan Tahun 2004-2005, pp: 39, Jabatan Imigresen Malaysia,
1st March 2005.
In terms of job distribution, the majority of irregular migrants work in the construction,
plantation and the services sectors (Table 6). In urban areas, they work mainly as tailors, petty traders,
cobblers, masseuses and gardeners. In the countryside, such as in Sabah, irregular migrants work in
commercial fruit and vegetable farming and on small holdings of rubber and oil palm (Mohd Rizal
2003; Ramlah Daud 2004; and Saira Jool 1998).
2. FOREIGN WORKERS: INSTITUTIONS AND GOVERNANCE REGIME, 1970S-2009
Malaysia’s historical experience in managing foreign workers provides an interesting case

study of policies and institutions at work.
Inception and development
Prior to the 1980s, there were no legal provisions for the recruitment of foreign workers into
Malaysia. Foreign workers were initially confined to the agricultural sectors and only later began
expanding into other sectors, such as petty trading, services and even informal manufacturing
(Abubakar 2002).
Some guidelines on the recruitment of foreign workers were developed in 1984, aimed at
encouraging the legal recruitment of foreign workers and reducing the number of illegal immigrants.
The guidelines covered: (1) the source countries for foreign labor, (2) the sectors and general
conditions under which foreign labor could be engaged, (3) the recruitment procedures for foreign
workers, including the amount of levies, bond, types of visa, etc., and (4) the eligibility, rights, and
responsibilities of foreign workers and employers.
These guidelines viewed the recruitment of foreign workers as a temporary measure to

overcome labor shortages in some sectors of the economy while longer-term domestic measures were
being developed. In pursuit of these guidelines, a number of initiatives were carried out, included the
signing of the Medan Agreement with Indonesia in May 1984, which was designed to regularise the
status of Indonesian agricultural workers in Malaysia.
Regularisation exercises were launched in 1985; however, due to the lack of response from the
workers and employers, another exercise was launched in 1987 and extended to 1989. Additional
regularisation exercises for other sectors were carried out in early 1992 beginning with domestic
services, followed by construction, services and manufacturing. At the same time, the government
245
took steps to establish the necessary institutions to regulate foreign workers such as the Cabinet
Committee on Foreign Workers (CCFW) and a Committee on the Recruitment for Foreign Workers2.
The Ministry of Human Resources (MOHR) was made responsible for matters related to the
welfare and employment of foreign workers, while the Department of Immigration (DOI) under the
Ministry of Home Affairs (MOHA) was made the lead agency responsible for processing the
recruitment of foreign workers and for eliminating irregular migrant workers. With these preparatory
steps completed, the guidelines, which covered the recruitment and employment of foreign workers,
became effective in early 1992. In 1993, there were over 532,000 legal foreign workers in Malaysia
and the figure trebled to 1.7 million in 1997.
The rapid growth in the number of foreign workers led to growing public concern about their
possible negative impact on society. In response, the government strengthened border control and
security, particularly in Sabah, and conducted regularisation and amnesty exercises. Regularisation
drives were carried out throughout the 1990s but these drives were later replaced by amnesty windows
when it was determined that regularisation exercises were of limited impact in reducing
undocumented workers. Many workers would fall back into irregular status when they failed to renew
their work permits.
Consolidation
Problems related to foreign workers were highlighted by the national media in the early 2000s.
These problems included an oversupply of foreign workers and accusations that employers were
discriminating against local workers because foreign workers were assumed to be cheaper and more
compliant. There was also the perception that foreign workers were responsible for the rising crime
rate.
Given these concerns, the government moved to overhaul its foreign worker policy. In 2005
the government introduced the Job Clearing System (JCS) and the One Stop Centre (OSC) to
streamline the worker recruitment process and to ensure that locals were not discriminated against in
the labor market.
The JCS requires all employers applying for foreign workers to first advertise all job vacancies
through a job clearing system. If this system fails to elicit enough responses from locals, then
permission is granted to the employer to recruit foreign workers (Buku Panduan, 2006:23).
The OSC’s Panel for the Recruitment of Foreign Workers includes representatives from the
relevant ministries and agencies. The Panel meets once a month to interview prospective employers to
simplify recruitment procedures, reduce processing time and provide a second layer of checking to
ensure that conditions for the employment of foreign workers are followed. The Panel also monitors
the inflow of foreign workers so that the employment of local workers is not jeopardised. Its
decisions, which may or may not concur with that of the MOHR, are conveyed to applicants within a
day, after which the prospective employer can proceed to pay the levy and apply for visas for the
approved number of foreign workers. This new recruitment procedure takes about a month compared
to three to six months previously.
In 2005 the foreign worker levy was raised in certain economic sectors in order to discourage
foreign employment in those sectors. For example, the levy for the service sector was increased to
RM1, 800, and the levy for the plantation sector was raised to RM540.
In addition, the Special Court for Illegal Immigrants was established in 2006 to expedite the
processing of cases against illegal immigrants. The huge backlog of illegal immigration cases in
2
Later renamed to the Technical Committee for the Recruitment of Foreign Workers
246
Malaysian courts led to severe overcrowding in the holding centres where arrested illegal immigrants
are detained. The Special Court, located at the Immigration holding centres, was established to ease
this situation.
Prior to 2005, foreign workers could only be employed directly by a company (except for
domestic maids), and one of the conditions of the work visa was that the worker could only stay and
work within the sector and company where they were originally hired. This situation prevented
worker mobility and employer flexibility, and created a variety of problems. For example, a job
contract in a particular construction site might be completed earlier than expected leaving many
workers unemployed. Foreign workers could also be made redundant half way through the tenure of
their work permits because of company downsizing. Foreign workers in such a predicament could not
transfer to another sector or company; instead, they were required to return home. However, in
practice many of these workers elected to stay in Malaysia and join the growing pool of irregular and
undocumented workers.
To avoid such outcomes, the government, in 2005, introduced the outsourcing system which allowed
the movement of foreign workers across employers and sectors. Under this system an outsourcing
agency supplies and manages foreign workers who are contracted out to a company or an individual
employer. This system allows indirect employment of foreign workers by a company/employer,
which pays the outsourcing agency for the workers’ services. The outsourcing agency is responsible
for the workers’ wages, housing, and other benefits. If the workers become temporarily unemployed,
the outsourcing agency must give them RM400 a month for their upkeep.
However, outsourcing companies were far from being the solution that the government
desired. Abuses highlighted by the media and nongovernmental organisations led the government to
halt the issuance of licenses for outsourcing companies, and to gradually phase them out. Initially,
there were about 400 outsourcing agencies, but due to inactivity or abuses the number of active
companies declined to less than a hundred.
Crisis redux
With the onset of the global economic crisis in late 2008, new policy instruments were put in
place. In January 2009 the government froze the hiring of foreign workers in specific jobs in the
manufacturing and service sectors. However, the freeze in the manufacturing sector was lifted in July
2009 because of strong demands by foreign multinationals. Additionally, the government issued a
new ruling making it compulsory for employers to pay the cost of the levy in an attempt to discourage
employers from employing foreign workers. Another measure announced was the doubling of the
cost of the levy in the manufacturing, service and agricultural sectors in early 2009. However,
opposition by the employer federation and trade associations caused this measure to be temporarily
shelved.
In February 2010, the government announced the establishment of a commission to study the
recruitment of foreign workers, headed by MOHA and comprising representatives from related
ministries. The commission is supposed to review existing policies, laws, recruitment procedures, and
to provide recommendations to the government to ensure that the hiring of foreign workers is
commensurate with the needs of the country. The findings of this commission are yet to be publicised.
3. INSTITUTIONS AND GOVERNANCE REGIME -THE PRESENT INSTITUTIONS
There are many institutions involved in the recruitment and employment of foreign workers in
Malaysia: The Cabinet Committee on Foreign Workers and Illegal Immigrants is the main body
pertaining to foreign worker policy in the country. This high-powered committee is chaired by the
247
Deputy Prime Minister and comprises Cabinet members from relevant ministries.3 Formerly known as
the Cabinet Committee on Foreign Workers, its name was changed in 2009 to better reflect the scope
of its functions. The Committee meets twice a year and monitors the implementation of policy
measures, and reviews and amends them in response to Malaysia’s economic, socio-cultural, political
and security situations.
Under the MOHR, the Department of Labor (DOL), specifically its Division for Management
of Foreign Workers, must assess an employer’s application for foreign workers to ensure that it is
based on actual labor shortages. Once foreign workers are brought in and employed in the country, all
matters relating to their welfare and employment are handled by this department.
Technically, foreign domestic maids are not under the jurisdiction of DOL and not subjected
to the labor laws of Malaysia. In 2009, measures were taken by the MOHR to put them under the
DOL. The MOHR initially announced plans to overcome the problem of exploitation of domestic
maids, such as making spot checks on private homes to ascertain the welfare status of the maids, but
opposition by some sectors forced the MOHR to abandon it.
Under the MOHA, the primary department involved with foreign workers is the Department
of Immigration (DOI). The Foreign Workers Division of the Department of Immigration is the body
responsible for processing visas for foreign workers to enter into and to work in Malaysia. This
Division also acts as the secretariat of the OSC, which processes prospective employers’ applications
to recruit foreign workers. This Division is also responsible for all matters relating to the recruitment
and employment of foreign maids.
There are a number of agencies responsible for dealing with irregular migrants. Ops Nyah 1,
which tightened border controls, includes the Royal Malaysian Police’s (RMP) General Operation
Force (GOP), which patrols border villages, the Army for jungle borders, and the Marine Operation
Force and the Malaysian Maritime Enforcement Agency (MMEA) to patrol sea borders. To deal with
illegal immigrants already in the country under the Ops Nyah 2 / Ops Tegas, the main agencies
involved are the Enforcement Division in DOI and RELA, the civil volunteer corps. The cooperation
of the RMP, National Registration Department (NRD), local councils and urban authorities is also
sought as and when necessary.
Laws and regulations
Malaysia has instituted a range of instruments through which it applies and enforces its policy
on foreign workers. This section covers some of the key laws and regulations pertaining to foreign
workers in Malaysia.
The Employment Act 1955 is the primary legislation covering employment of workers in
Malaysia, both foreign and local. The Act governs employment practices in Malaysia and sets out the
minimum conditions of employment, including rights and responsibilities of employers and
employees, wages, termination of work and lay-off benefits, shift work and maximum hours of work,
rest days and rest day pay, sick days, annual leave, etc.. The Act empowers the Director General of
the DOL to investigate and take action in the case of discriminatory and unfair treatment of workers.
The Workmen’s Compensation Act 1952 requires employers to provide insurance coverage
for each foreign worker they employ (excluding domestic maids) from officially approved insurance
companies at a stipulated annual cost. This legislation covers foreign workers who are legally
3
These ministries include the Ministry of Home Affairs, Ministry of Human Resources, Ministry of Public Works, Ministry
of International Trade and Industry, Ministry of Foreign Affairs, Ministry of Agriculture and Agro-Based Industry, Ministry
of Finance, Ministry of Plantation Industries and Commodities, Ministry of Rural and Regional Development, Ministry of
Health, and the Ministry of Tourism.
248
recruited as they are not covered by the Social Security Act and provides for compensation in the
event of injury or death.
The Immigration Act 1959/63 and Passport Act 1966 are the two primary pieces of legislation
with regards to the entry and exit of foreign workers, illegal workers and immigration matters. A
foreign worker is only permitted to work in Malaysia when in possession of a valid work permit
issued by the DOI.
The Anti-Trafficking Act 2007 defines human trafficking activities as encompassing both the
exploitation of men, women and children for sexual purposes and for labor, and the sale of children
and human organs. The Act has three main objectives, to prevent trafficking activities, to protect and
provide shelter for trafficking victims, and to prosecute traffickers. It was enforced in February 2008
with the formation of the Malaysian Council for Anti-trafficking in Persons.
Procedures and Costs of Recruitment of Foreign Workers
The procedures and associated costs for the recruitment of workers are described below.
Procedures for recruitment
The first step for employers who intend to hire foreign workers is to make initial efforts to
recruit local workers. Only if there are insufficient local workers can the employer submit an
application to the DOL for a permit to hire a fixed number of foreign workers. After obtaining this
permit, the company or outsourcing agency then submits an application directly to the OSC, which
reviews the application and may make further adjustments to the number of workers to be hired.
The employer then proceeds to hire foreign workers from the source country either directly or
through the services of a recruitment agent. After obtaining the required number of workers, the
employer then liaises with the DOI to obtain Calling Visas for the workers and to pay the levy. After
the submission of the relevant documents and successful completion of a medical examination in the
source country, the DOI issues the foreign worker with a work pass called the Visit Pass (Temporary
Employment), or VP (TE). On arrival in Malaysia, the worker undergoes a second medical
examination. Upon passing the medical check-up, the worker is then placed with the employer.
Renewals of the VP (TE) pass are carried out yearly, subject to the payment of the annual levy and the
foreign worker successfully passing an annual medical checkup. Foreign workers who reach the
maximum five-year period are required to return home for a ‘cooling-off’ period of 3 months before
being allowed to return.4 Domestic maids are not subject to this ruling and can work for as long as she
is fit and her services are required.
Cost of legal recruitment
The costs of recruitment vary depending on the source countries and the job sector of the
foreign worker. Recruitment costs include statutory payments for entry into Malaysia, such as levy,
visa, and deposit fees, as well as other processing and administration fees. The cost of the work pass,
or VP(TE), is RM60 for all sectors, and the processing fee is RM50 for the manufacturing,
construction and service sectors, and RM10 for the remaining sectors. Levies represent the primary
mechanism used by the Malaysian Government to control the intake of foreign workers, while visa
fees and deposits are primarily an immigration matter.
4
The break period was previously set at six months, but was changed to 3 months in early 2009.
249
Table 7. Cost of Levy for Foreign Workers, Peninsula Malaysia, Sabah and
Sarawak, 2009.
No. Sector Fee (RM)
Peninsula Malaysia Sabah/Sarawak
1. Manufacturing 1,200 960
2. Construction 1,200 960
3. Plantation 540 540
4. Agriculture 360 360
5. Services
a) Restaurant 1,800 1,440
b) Cleaning Services 1,800 1,440
c) Cargo handling 1,800 1,440
d) Laundry 1,800 1,440
e) Caddy 1,800 1,440
f) Barber 1,800 1,440
g) Retailing and whole selling 1,800 1,440
h) Textile Merchant 1,800 1,440
i) Scrap iron 1,800 1,440
j) Welfare Homes 600 600
k) Resort Island 1,200 960
6. Others (Special approval) 1,800 1,440
7. Domestic Helper
a) First Domestic Helper 360 360
b) Second Domestic Helper 540 540
Source: Department of Immigration, Malaysia.
Levies are based on the current policy towards the employment of foreign workers in a
specific sector or job type. For example, the foreign worker levy in agriculture and plantations is low
because foreign labor is deemed to be essential in this sector. Likewise, the levy for a first domestic
helper is low to encourage greater female participation in the labor force. The exact costs for
recruitment are often difficult to ascertain because in many cases workers either pay a lump sum fee
upfront to the agent in their respective countries, or they receive a loan from agents or prospective
employers. The costs of these loans are then deducted from the monthly salaries of the workers. In
many cases, foreign workers often have limited knowledge of the specifics of these salary deductions
or the exact amount of the deductions relative to their monthly salaries.
In some cases, bilateral MOUs signed between Malaysia and source countries may provide
cost guidelines for the recruitment of foreign workers. However, the impact of these MOUs is limited
because the costs outlined in the MOU only serve as a guideline and do not have any legal clout. For
example, in the case of Indonesian domestic workers, the 2006 MOU between Malaysia and
Indonesia details a recommended cost structure for the recruitment process where the Malaysian
employer pays an upfront fee of RM6,415. Of this, RM2,700 is ultimately borne by the Indonesian
maid through salary deductions. In other words, the Malaysian employer pays the full recruitment cost
in Malaysia (RM2,415) and part of the recruitment cost in Indonesia (RM1,300).
250
Table 8. Visa Fee and Deposit for Foreign Workers (Excluding Domestic Helpers)
No. Nationality Visa (RM) Deposit (RM)

1. Indonesia 15 250
2. Bangladesh 20 500
3. Myanmar 19.50 750
4 India 50 750
5. Vietnam 13 1,500
6. Philippines 36 1,000
7. Cambodia 20 250
8. Nepal 20 750
9. Thailand Gratis 250
10. Pakistan 20 750
11. Turkmenistan, Uzbekistan & Kazakhstan 20 1,500
12. Laos 20 1,500
50 ( SEV1)
13. Sri Lanka 750
100 (MEV2)
1 2
Notes: Single Entry Visa. Multiple Entry Visa.
Source: Department of Immigration, Malaysia.
While the MOU sets the total recruitment costs in Indonesia at RM3,070, it was found that
actual costs are much higher. An interview with the association for foreign maid recruiting agencies
reveals that total recruitment costs are between RM6,000-7,000, while SUHAKAM, the Human
Rights Commission of Malaysia (SUHAKAM) notes that the costs paid to foreign agencies could rise
as high as RM8,000-12,000 (Table 9).
Table 9. Estimated Recruitment Fees Paid by Workers by Country
Worker Fee
Nepalese security guards RM7,500-RM8,500
Indian restaurant workers from Chennai RM6,500-RM7,500
Bangladeshi service sector workers from Dhaka RM12,000-RM13,000
Filipino domestic workers from Manila RM6,000-RM8,000
Filipino service sector workers RM2,700-RM4,500
Indonesian domestic workers from Surabaya RM4,500-RM8,000
Indonesian domestic workers from Medan RM4,000-RM8,000
Thai restaurant workers from Patani & Yala Approx. RM2,500
Source: Informal interviews with foreign workers.
4. REFLECTIONS AND ASSESSMENT
The preceding sections have provided the historical context for foreign worker policy in
Malaysia and the present system of institutions and governance. In this section, specific issues are
distilled and elaborated to lay the groundwork for the policy recommendations.
Political and social implications
Malaysia’s foreign worker policy did not emerge as a pre-planned policy but more as one that
evolved as a de facto and ad hoc necessity. Its origins are to be found in attempts to solve a long-
standing problem: the existence of a large pool of illegal migrants in the country. This raises a key
point, namely, that there is a difference between foreign workers and foreign immigration.
Oftentimes, the issue of foreign workers is conflated with the more politically and socially explosive
251
issue of foreign immigration. While these two issues are closely related, they are not always the same.
Foreign workers do not always seek residency in the country where they work.
Malaysia is a multicultural country where ethnic and cultural divisions are still very evident.
For this reason, an influx of migrant workers can tilt delicate ethnic balances and create competition
and contention instead of cooperation and compromise. In Sabah, for example, economic migrants
from Indonesia and the Philippines have significantly reduced the indigenous Kadazan-Dusun-Murut
proportion of the population. This has created much dissatisfaction, with locals complaining not only
about the loss of political influence but also about losing out in competition for limited employment,
housing and business opportunities. In addition, there is the problem of the unregistered birth of
children of irregular migrants. Because they fear deportation, most foreign migrants do not register
the births of their children. Without documentation, these children grow up without access to basic
health and education and are forced to pursue dismal lives like itinerant trading or begging.
Foreign workers who are not seeking residency impose fewer social costs. They want nothing
more than to earn a decent wage and to repatriate their income in order to buy a house and perhaps
start a business in their home countries. Their presence in a country like Malaysia presents much less
of a social and political threat to local residents.
Foreign immigration is a highly controversial issue in most countries, including Malaysia. The
tension between negative public opinion about foreign workers and the economic need to import more
workers in labor critical fields forces some governments to adopt a strategy of non-transparency. In
Malaysia, this has, on occasion, produced uncertainty and confusion among stakeholders.
Security implications
While foreign workers may not have undermined law and order, they can still pose a potential
threat to the nation state. For instance, in the past there have been numerous incidents of unrest at the
immigration holding centres. A watershed incident for the government occurred in 2002 when
hundreds of foreign workers rioted at a textile plant and the police on hand were unable to stop the
rampage. Not surprisingly, three years later, an unpublished report by the Enforcement Division of the
Department of Immigration referred to irregular migrants as ‘Public Enemy No. 2’, next only to the
menace of drug addiction.
Many Malaysians believe that foreign workers are responsible for the rise in serious crimes
such as murder and rape in the country. However, the little evidence that exists does not support this
popular belief. A 2005 paper in the Journal of the Kuala Lumpur Police College examined statistics
between 1992 and 2002 and found that while foreigners on average commit about 3.8 crimes per
1,000, Malaysians commit 5.3 crimes per 1,000.
The more immediate security threats from foreigners appear to lie in non-traditional areas,
like the smuggling of alcohol, narcotics and firearms, human trafficking, and money laundering. For
example, the large pool of irregular workers in Malaysia has helped create a thriving forged document
industry that is capable of producing phony identity cards, birth certificates and driver licenses. Also,
human trafficking has taken a huge toll on Malaysia’s international reputation, as when the US placed
Malaysia in Tier 3 of its Trafficking in Persons Report 2009.
Economic implications
The government has responded to the fluctuating fortunes of the economy by adopting a
foreign worker policy that is flexible and pro-cyclical. In the 1998 Asian Financial Crisis, the intake
of foreign workers was frozen to protect the jobs of Malaysian workers and relaxed a year later when
recovery was in place. The Global Financial Crisis of 2008 saw similar efforts being taken, with a
252
further proposal to double the foreign worker levy. However, because of strenuous objections by
employers, this increase in the foreign worker levy never took place.
Malaysia’s policy on foreign workers, which is often referred to by government officials as a

“short term policy,” is really more a statement of hope than reality. When foreign workers account for
17% of the legal labor force and show few signs of diminishing over time, it seems that foreign
workers will represent an important part of the Malaysian work force in the future.
Given the number of foreign workers in Malaysia, it is not surprising that one of the major
concerns is the impact of these foreign workers on labor costs, productivity and inflation. For
instance, many economists have argued that foreign workers have depressed real wages, reduced the
incentive for capital deepening, discouraged technological upgrading and innovation, and prolonged
the existence of low productivity industries. For example, it has been noted that prior to the Asian
financial crisis, average annual real wage growth in Malaysia was 5.6% for export oriented industries
and 6.8% for domestic oriented industries; however, post crisis, these growth rates for wages fell to
1.9% and 1.4% respectively(Hunt 2009). According to some economists, growth rates for wages have
declined in Malaysia in recent years because foreign labor makes domestic labor supply more elastic
than would otherwise be the case and bids down the price of marginal units of labor.
During the past decade the government has made a number of statements about its intention to
reduce the country’s dependence on foreign labor. But these statements have not always been backed
up with consistent actions, in part, due to a need to adjust to changing global economic realities.
Globally, foreign direct investment (FDI) has been more difficult to attract since 2000 and relaxing
foreign worker restrictions may have been seen as one way to continue to draw FDI into Malaysia.
China also posed a competitive threat and the hiring of foreign workers may have been a way of
slowing the relocation of industries and the ensuing loss of jobs. However, one negative consequence
of this lack of government action has been a disregard for the longer-term development of local
workers in Malaysia. There has been a tendency to issue quotas based more on the immediate
demands of private employers than on the needs of local workers. Private employers have argued,
and the government itself has on occasion said, that foreign workers fill the vacuum for unskilled jobs
that no one else wants.
Management implications
The management of foreign workers exhibits many of the characteristics of the familiar agent-
principal problem in which the engagement of an agent to act on behalf of a principal raises problems
of moral hazard, conflicting interests and rent-seeking. The problems of management can therefore be
endemic and institutionalised. At the highest institutional level, the government may have an
increasing conflict of interest in the form of the revenue raised from foreign worker levies. Further on
down the institutional line, immigration officials, recruitment agents and outsourcing companies may
all benefit from a higher rather than a lower volume of foreign workers.
From the Malaysian experience, the principle would seem to be that transaction costs should
not significantly exceed the costs of compliance. For example, foreign recruiting agents often demand
high commissions and fees and make legal recruitment highly unattractive. Currently, it is much
cheaper for a prospective maid from Medan to enter Peninsular Malaysia via Klang, Selangor by boat
for RM99 as a tourist, overstay and work illegally, than to pay RM6,000 in recruitment fees to get a
job through a formal recruiting agent.
The foreign worker levy system used in Malaysia is also employed in Singapore, Hong Kong
and Taiwan. In Hong Kong’s case, however, the levy is set by legislation to be for training and
retraining. The purpose is not expressly to manage demand for foreign workers as seems to be the
case in Malaysia and Singapore. It is still unclear if the levy system does in fact constrain employers’
enthusiasm to hire foreign workers. In the first place, wage and non-wage cost differences between
foreign and domestic workers may be so large that the levy will not have any restraining effect at all.
253
In the second place, there is no foolproof way to ensure that employers do not pass down the costs of
the levy to the worker. There are reports that this is already a widespread practice. Under these
circumstances, the foreign worker levy may be of limited usefulness in regulating demand.
In Malaysia amnesty and repatriation exercises are conducted from time to time to try to
manage irregular foreign workers. However, because of frequent policy changes, these policy
initiatives are not very effective. For instance, irregular migrants who receive amnesty are required to
go home. At home they typically apply for a passport under a different name, re-enter Malaysia on a
tourist visa and rejoin the ranks of irregular migrants. Similarly, many irregular migrants who get
legalised during regularisation exercises revert back to illegal status after a year in order to avoid
paying the costs of the foreign worker levy.
The ineffectiveness of amnesty and repatriation exercises has led the government to take an
increasingly hard line on apprehending, detaining, and deporting illegal workers. Adopting a hard
line, however, has its costs in terms of adverse international publicity. Malaysia has only 17 detention
centres which can quickly become overcrowded. Rooting-out exercises can only be conducted when
there are vacancies in the holding centres, and deportations can be done quickly. There are also
significant costs involved. Malaysia also has a land border of 3,147 km with three countries and
common maritime borders with six countries. There are therefore considerable challenges in enforcing
cross-border movements.
Worker welfare implications
The ill treatment of foreign workers in Malaysia has received a great deal of negative national
and international publicity. This has been particularly true of severe cases of abuse, predominantly
with Indonesian maids, but also with workers from India.5 Exploitative and discriminatory practices
have also been highlighted, by the Human Rights Commission of Malaysia (SUHAKAM), the
MOHR, the Malaysian Bar Council, and various nongovernmental organisations. While the number
of reported cases of ill treatment of foreign workers is probably not significantly higher than other
migrant worker countries such as Hong Kong and the Middle East, the adverse publicity does not give
this impression.
The basic problem here is that unskilled foreign workers – especially those without much
education – do not have much bargaining power vis-a-vis employers. As a result, they are frequent
victims of exploitation and abuse. The recruitment process in many source countries is highly
exploitative. In some situations, foreign workers are recruited through several different middlemen
and arrive in Malaysia with a huge amount of debt. As noted earlier, in the case of maids in Malaysia,
up to half the agent’s fees are commonly passed back on to the worker through salary deductions.
Ultimately, the foreign worker may suffer the most, having to bear such costs in addition to the costs
already incurred from source country agents and other intermediaries.
The deduction of wages is common in the first few months of employment and is also
practiced in other countries in the region (Li, 2003). However, there have been reports where
deductions have been excessive or carried out in contravention of the employment contract. This can
be due to weak enforcement (for example, the large number and wide dispersion of households in the
case of domestic maids) and the fear that foreign workers will abscond, or to misunderstandings on
repayment of personal loans.
Foreign workers can be denied various statutory benefits either by specifically excluding such
provisions in their contracts or not honouring existing ones. Employers may impose a ‘no work, no
pay’ rule where deductions are made to a worker’s salary if they do not work, refuse to provide
mandatory paid leave (annual and sick leave) or pay overtime and award rates (Robertson, 2008).
5
Such as reported in the US Department of State Trafficking in Persons Report 2008–Malaysia which noted two incidents in
2008.
254
There have been numerous reports of contract switching or foreign workers being assigned to jobs
other than those agreed. Some domestic maids have also been prohibited from practicing their religion
or made to perform work contrary to their religious beliefs (e.g. Muslim domestic maids made to cook
and handle pork).
It is common to hear of cases where foreign workers have been forced to live in overcrowded
employer-provided accommodation. For example, there are reports that as many as 50 workers are
made to live in an overcrowded group house without electricity and running water. The inability of
legal foreign workers to switch jobs or employers creates a high dependence of these workers on their
employers and compromises their bargaining position with regard to the conditions of their work. The
Malaysian Trade Union Congress (2008) notes that, “When a worker seeks redress for unpaid wages
or raises other forms of labor dispute or abuse, the employer often retaliates by cancelling the work
permit. As a result, the migrant worker loses his or her status in the country and his or her right to
stay.”
In such situations, workers are only able to pursue their case with a 3-month special pass from
the Immigration Department at RM100 per month. This special pass only allows them to stay in the
country and not to work. Unable to work, and possessing little savings, most foreign workers cannot
afford this special pass because claims filed in the labor court often take months to settle.
The lack of bargaining power of foreign workers has resulted in a situation where legal
foreign workers face high costs and low rewards for remaining in a legal status. This, in turn, serves to
increase the number of irregular workers in the country, either through workers who enter the country
illegally or from legal workers who fall into an irregular status. While an irregular status may allow a
foreign worker to avoid the high costs of regularisation, it also puts that worker in a more vulnerable
position vis-a-vis errant employers and corrupt government officials.
5. THE WAY FORWARD
Formulation of a National Policy
Foreign worker policy in Malaysia needs to be made in a clearer and more informed manner.
Policymakers may want to consider adopting a formal and publicly announced ceiling on foreign
workers for each sector of the economy. The purpose of formulating such ceilings would be to
provide guidance to existing and prospective investors about the activities they should be considering
as well as the expected capital per employee. This should encourage (or at least not discourage) a
faster rate of adoption of technology and innovation. What the government would lose in terms of
perceived flexibility would be more than offset by the increase in efficiency gains in resource
allocation (i.e. corporations will be able to make better investment and human resources decisions).
It is also necessary to address the unequal costs of hiring foreign as compared to local
workers. While the Employment Act does not differentiate between local and foreign workers, many
factors outside the scope of the act continue to contribute to large cost differentials between local and
foreign workers. This is reflected not only in the difference in wage costs between local and foreign
workers, but also in terms of the non-wage costs, including social benefits, housing, personal accident
and medical benefits, leave and holidays, terms of payment, etc.
At present, the large difference in living standards between Malaysia and labor-sending
countries has created a situation of an elastic supply of foreign workers willing to work for very low
wages. By contrast, the supply curve for local workers is much less elastic and starts at a much higher
entry wage than that for foreign workers. As a result, local workers are excluded from many lower-
level jobs in the Malaysian economy.
In terms of wage costs, the Malaysian government is presently considering the
255
implementation of a minimum wage for Malaysians. It is very important to ensure that foreign
workers are included in these discussions. The exclusion of foreign workers from any minimum wage
scheme could result in a situation where the wage differential between local and foreign workers
would become so large that it would effectively end the hiring of local workers in certain sectors of
the economy. In terms of non-wage costs, it is clear that both local and foreign workers must be
accorded a minimum set of standard benefits and protections, including medical insurance and sick
leave, costs and procedures of hiring and firing, allowances for annual holidays and rest days, and
payment of wages. Many of these basic minimum benefits have been set out in various legislation (i.e.
Employment Act, Workmen’s Compensation Act, Industrial Relations Act, etc.). However, some
organisations such as the Malaysian Trades Union Congress note that foreign workers continue to
face problems when they are presented with substitute contracts which override their right to such
basic benefits and protections.
In the future it is likely that unskilled foreign workers will continue to comprise the bulk of
workers at the unskilled and semi-skilled levels; however, it is essential that policy does not actively
discriminate against local workers in such jobs. With an increase in costs for foreign workers, firms
would have a greater incentive to increase the productivity of all their workers, both local and foreign.
This could in turn lead to a virtuous cycle of upgrading, productivity increases, and economic
benefits. However, there is a danger that some unscrupulous employers may try to pass on any added
costs to foreign workers, as has been found in some cases. It will therefore be necessary to identify
the appropriate mechanisms for preventing employers from passing on such costs to foreign workers.
Taking a Differentiated Approach
This chapter has argued that it would be productive to take a differentiated approach in the
formulation of policy for foreign workers. One of the reasons why the present policy may appear to be
makeshift and truncated is that it attempts to achieve multiple goals simultaneously. Thus, attempts to
reduce the number of illegal foreign workers has led to depriving industry of legal workers, while
attempts to strictly control the use of legal foreign workers has inadvertently resulted in the creation
of more illegal workers. These conundrums are not easy to resolve with a broad policy brush and
without breaking down the problem into more manageable components.
Box 1 shows a simple policy matrix that could be helpful in making the necessary distinctions.
Malaysia’s policy should aim to utilise registered short-term foreign workers to its best advantage,
while actively discouraging long-term and undocumented migrants, in particular, those who are
chronic long-term overstayers.
In terms of legal long-term
Box 1. Policy Matrix for Foreign Workers residents (upper right hand corner
SHORT-TERM LONG-TERM of box), Malaysia encourages
RESIDENT RESIDENT foreign expatriates subject to their
meeting occupational and skill
LEGAL Manage requirements and having
-
(DOCUMENTED) Utilise employment. Foreign nationals
bringing in a certain amount of
ILLEGAL Regularise Prosecute/Penalise capital may also stay for long
(UNDOCUMENTED) Control Repatriate periods under the Malaysian My
Second Home (MM2H)
Source: Authors programme. Legal short-term
residents (upper left hand corner of
box) also pose little issue in terms of management, save for the debate about their numbers, utilisation
and impact on the domestic labor market. Since these workers are officially registered, they can be
either increased or reduced according to need.
For long term illegal migrants (lower right hand corner of box), the case is equally clear-cut:
they have been found guilty of violating the Immigration Act and are therefore liable to deportation.
256
To not prosecute these people would be to put Malaysia at serious social, political and security risk.
The case is less clear for illegal short-term foreign workers (lower left hand corner of box).
These workers have no intention of remaining in Malaysia, but they have fallen into an illegal status
due to technicalities such as failure to renew a visa or pay the foreign worker levy. Quite often, this
can be due to business failure or attempts by employers to exploit their workers. At present, these
illegal short-term workers are apprehended as irregular workers and deported. While these workers
may have technically violated the terms of their stay in Malaysia, it might be useful to consider
granting them a grace period of 6 to 9 months to regularize their status or to find new employment.
At present, the policy is one of very low tolerance. A more enlightened approach would be to allow
these illegal short-term migrants to continue working and repatriating earnings legally rather than
force them to join the longer-term underground workforce.
Greater Institutional Co-ordination
The management of foreign workers involves understanding the potential moral hazards,
conflicts of interest and opportunities for rent-seeking behaviour in the whole process of foreign
worker migration. The challenge is to minimize perverse incentives so that the maximum benefits of
policy can be derived at the minimum social welfare cost.
The present Cabinet Committee on Foreign Workers and Illegal Immigrants should continue
to be the highest level body of decision making on foreign worker policy. The Cabinet Committee
should establish the ceilings of legal foreign workers in the various sectors, their skills and
occupations, and preferred source countries. It should also set the broad policy and strategies for the
detection and repatriation of illegal immigrants. In order for policy decisions to be taken in an
informed manner, a National Advisory Council on Foreign Workers should be established comprising
government agencies, the private sector, academics and nongovernment organisations. This Advisory
Council could meet twice a year to raise and discuss migrant-related issues. Alternatively, an Annual
Dialogue could be held between Cabinet ministers and stakeholders that emulate the dialogues that
are now being conducted by the Ministries of Finance and International Trade and Industry on other
topics.
Once decisions on foreign workers have been made, the implementation machinery must
ensure that these decisions are effectively carried out. Partial execution of policy decisions can
introduce doubts and uncertainties and open opportunities for more rent-seeking behaviour. As
highlighted earlier, there is need for better co-ordination among the government ministries,
departments and enforcement agencies on the implementation of policy.
The decision on the number of foreign workers needed, their occupation and skill sets and,
most importantly, the oversight and regulation of recruitment agencies and employers should be
vertically integrated under the MOHR. The MOHR should also be made the primary ministry
responsible for all matters of foreign worker welfare. The MOHA, and the DOI under it, should
continue to be responsible for visa issuance, border control and for the management of illegal
immigrants. It should also maintain its key role in declaring amnesties and repatriation.
Strengthening Management Practices
Government-to-government MOUs should be upgraded and governments of source countries

should be urged to adopt best practice standards in terms of ensuring good governance of their
recruitment agencies. Processes should be as transparent and competitive as possible in order to
reduce recruitment costs. The latter is one of the leading factors responsible for foreign worker debt.
Employment contracts which contain clauses that override labor legislation of the host country must
no longer be tolerated. The labor standards that are stipulated under Malaysian labor law must be
made applicable to all workers, both foreign and domestic. For special cases such as domestic maids,
separate legal amendments should be introduced.
257
In order to reduce the possibilities for worker exploitation, private employers should be
required to obtain foreign workers directly from source countries. This would mean that they have a
vested interest in finding the best candidates and that they are legally liable for the recruitment and
employment of foreign workers. This would mean that the Malaysian government would have to stop
sanctioning private outsourcing companies, a decision that appears to have been made but not yet
implemented.
The Malaysian experience seems to suggest that the foreign worker levy has only limited
effectiveness in regulating industry demand for foreign workers. It would appear that the higher the
levy, the greater the incentive to victimise the foreign worker through unlawful deductions.6
Improving Foreign Worker Welfare
It is important to take steps to improve the welfare of foreign workers in Malaysia. One key,
but controversial, issue here would be to allow foreign workers to bring their spouses and children
with them. Taking this step would put Malaysia in line with international conventions and basic
human rights. However, the presence of family members might increase the temptation of foreign
workers to settle in Malaysia after the expiration of their contracts.
Exploitation of foreign workers in Malaysia can also be reduced by:
 Introducing a minimum wage for both foreign and domestic workers. In order for this to be
effective, daily rated work will have to be banned and the practice of salary deductions will
have to be monitored and strictly enforced.
 Requiring foreign workers to pay recruitment costs upfront so that there are no reasons for
employers to make salary deductions. Employers should also be prohibited from keeping
workers’ passports and the Foreign Worker Card should be accepted by all government
agencies.
 Reformulating the practices and procedures of law enforcement agencies in conducting

checks on foreign workers. These checks must be done in a way that is civil, just and humane.
 Making amendments to the Immigration Act 1959/63 so as to enable foreign workers

involved in legal cases in Malaysia to remain in the country at no extra cost, and to work in
the country until their case is decided
.
 Phasing out outsourcing agencies and replacing them with government sponsored agencies.
Reducing Illegal Migrants
Currently, Malaysia imposes stringent penalties on employers who hire illegal foreign workers,
including fines, jailing and/or caning. Such penalties may have had the unintended effect of pushing
irregular foreign labor further underground. Fewer records are kept of such irregular workers by
employers. Being undocumented, they are in a more vulnerable position and more likely to be
exploited.
Employers in violation of offenses against the Employment or Immigration Acts should

certainly be punished. However, it is not clear whether jailing and caning sentences represent a good
6
The Malaysian government has recently declared that the levy for foreign workers should be borne by the
employer and not by the foreign worker, beginning April 1, 2010. It is too early to gauge the compliance with
this directive and what steps are taken to enforce it.
258
deterrent. This is especially true in the case of large corporations, where it is quite difficult to
identify who is legally responsible. Moreover, fines should be set a level that serve as an effective
deterrent. Fines which are set too low are not effective as employers may be more prepared to risk
being caught.
In addition, it is necessary to strengthen procedures and mechanisms to deal with irregular

migrants.
 A large number of irregular migrants have entered Malaysia by abusing the visa on arrival
(VoA). It is therefore recommended that the procedures for visas and entry be tightened up in
order to minimise the abuse of the VoA. Additionally, the government has announced its
intention to set up a biometric identification system for foreign workers, and the use of this
system should be expanded to complement enforcement activities.
 Amnesty exercises have proved ineffective in reducing the number of irregular workers
because there is no clause preventing them from coming back into the country. In many
cases, irregular workers are sent back just so they could go through the due processes to re-
enter Malaysia again as legal workers. It is therefore proposed that repeat offenders of the
amnesty programmes be barred from re-entering Malaysia for a specific period of time, e.g.
10 years, as is practiced in Japan.
 Employers and recruiting agents who disregard the law must be prosecuted to the fullest
extent of the law. One of the chief complaints of source countries is that Malaysian
government is seen as acting only in cases involving irregular migrants, while simultaneously
turning a blind eye to the illegal actions of employers. Under the present situation, irregular
migrants who are apprehended are detained until they can face trial, before being sent back.
Often, this results in high costs both to Malaysia and the irregular worker, given that the
worker may be detained for a lengthy period of time. It is recommended that irregular
workers who are apprehended be repatriated swiftly with a minimal period of detention, and
that the cost of repatriation be borne by their employer.
 Malaysia faces a major problem regarding the unregistered children of irregular migrants.
Many of these children have been in Malaysia a long time and are effectively stateless. It is
recommended that the Malaysian government either accelerate consultation with source
countries to repatriate these children, or consider the option of naturalizing them in Malaysia.
259
REFERENCES
A.A. Abdul Rashid. 1996. Foreign Workers in the Construction Industry, paper presented at the
Workshop on Foreign Labor in Malaysia, Institute of Higher Studies, University of Malaya, Kuala
Lumpur.
Abubakar, Syarisa Yanti. 2002. Migrant Labor in Malaysia: Impact and Implications of the Asian
Financial Crisis. EADN Regional Project on the Social Impact of the Asian Financial Crisis working
paper.
Amar Singh Sidhu. 2005. The Rise of Crime in Malaysia: An academic and statistical analysis. In
Journal of the Kuala Lumpur Royal Malaysia Police College, No. 4, 2005, pp. 1-28.
Athukorala, Prema-Chandra and Chris Manning. 1999. Structural Change and International Migration
in East Asia: Adjusting to Labor Scarcity. Singapore: Kyodo Printing. 256 pp.
Azizah Kassim, 2001, Integration of foreign Workers and Illegal Employment in

Malaysia. In International Migration in Asia: Trends and Policies. Organisation of
Economic Co-Operation and Development (OECD), Paris.
Azizah Kassim, 2003, Illegal Immigrants in Sabah: Their Metamorphosis and

Survival Strategies. In Miyazaki Koji (ed.) Dynamics of Border Societies in Asia,
Research Institute for Languages and Cultures of Asia and Africa (ILCAA) Tokyo
University of Foreign Studies, pp. 47-79.
Azizah Kassim, 2005, ‘Security and Social Implications of Cross-National Migration

in Malaysia’. In: Mohamad Jawhar Hassan, (ed.) 2005. Pacifying the Pacific. ISIS:
Kuala Lumpur.
Azizah Kassim, 2005a, ‘Cross-border Movement of Foreign Workers in Malaysia: A

Comparative Analysis.’ In: Master Builders Journal, 3rd Quarter, Kuala Lumpur,
pp.78-91.
Azizah Kassim, 2008, Managing Strangers in Our Midst: A Critical Analysis of

Malaysia’s Foreign Workers’ Policy. Paper presented at the MSC6 (6th Annual
Conference of the Malaysian Social Science Association) Kuching, Sarawak, August.
Azizah Kassim, 2009, The Refugees in Sabah: State Responses, Public Stereotypes
and the Dilemma over their future. In Journal of Southeast Asian Studies, Kyoto
University, Japan, Vol. 47, No.1, and June 2009:50-86.
Bahrin T.S & Rachagan, 1984, The Status of Displaced Filipinos in Sabah: Some
Policy Considerations and Their Long Term Implications. In Lim Joo-Jock and Vani
S. (eds.) Singapore Armed Separatism in Southeast Asia Institute of Southeast Asian
Studies, pp. 190-213.
Brillo, Bing Balthazar C. 2008. ‘Path Dependence, Increasing Returns, and Philippine
Labor Migration Policy’. Crossroads, Vol. 8, No. 1, pp. 24-61.
Chew Sang Hai. 2004. Foreign Construction Workers: Why We Need Them? In
Azizah Kassim. (ed.) Public Responses to Foreign Workers in Sabah, Kota Kinabalu;
Universiti Malaysia Sabah, pp. 47-54.
Dairiam, Grace. 2006. ‘Country Case Studies on Bilateral Labor Agreements:
260
Malaysia’. Paper presented at the 12th Tokyo Workshop on International Migration
and Labor Market in Asia, 17th February, 2006.
Ford, Michelle, 2006, The Regulation of Indonesian Migration to Malaysia. In

Amarjit Kaur and Ian Metcalfe (eds.) Mobility, Labor Migration and Border
Controls in Asia, Palgrave Macmillian, New York, pp. 228-247.
Hunt, Laalitha. 2009. ‘Significant drop in real wages in Malaysia’. The Star Online,
January15,2009.http://biz.thestar.com.my/news/story.asp?file=/2009/1/15/business/322966&sec=busi
ness, accessed June 20, 2010.
Kanapathy, Vijayakumari. 2008. ‘Managing Cross-border Labor Mobility in

Malaysia: Two Decades of Policy Experiments.’ PECC-ABAC Conference on
International Labor Mobility, Seoul, Korea. http://www.pecc.org/resources/doc_details/690-
managing-cross-border-labor-mobility-in-malaysia-two-decades-of-policy-experiments, accessed July
22, 2009.
Kaur, Amarjit and Ian Metcalfe, (eds.) 2006. Mobility, Labor Migration and Border
Controls in Asia. Great Britain: Antony Rowe. 322 pp.
Li, Simon. 2003. ‘Fact Sheet: Hiring of Foreign Domestic Helpers in Selected
Jurisdictions’. The Research and Library Services Division of the Legislative Council
Secretariat, Hong Kong. http://www.legco.gov.hk/yr02-03/english/sec/library/0203fs07e.pdf,
accessed June 15, 2010.
Manning, Chris and Pradip Bhatnaga (2006). ‘Coping with Cross-Border Labor
Flows Within Southeast Asia’ in Amarjit Kaur and Ian Metcalfe (eds), Mobility,
Labor Migration and Border Controls in Asia (Antony Rowe Ltd: Great Britain).
Mohd. Rizal Omar, 2004, Hubungan pendatang asing dengan penduduk tempatan:
Satu kajian kes di Kampung Kundasang (Interactions between immigrants and the
local population: A case study in Kundasang). Graduation Exercise, School of Social
Sciences, Universiti Malaysia Sabah, Kota Kinabalu.
Mohd Yaakub Hj Johari & Ramlan Goddos, 2003, Migrant workers in Sabah:
Characteristics, Processes, Prospects and Challenges. In Mohd Yaakub Hj Johari &
Chong Shu Yaw (eds.), Reinventing Sabah: Global Challenges and Policy Responses,
Institute for Development Studies, Sabah & Konrad Adenauer Foundation, Germany.
MTUC. 2005. ‘Concluding Resolution’. MTUC Conference on Migrant Workers, April

18-19, 2005, Petaling Jaya, Malaysia. http://www.mtuc.org.my/migran workersworkshop05.htm.
Ramasamy, P., 2004, International Migration and Conflict: Foreign Labor in

Malaysia. In Aris Ananta and Evi Nurvidya Arifin (eds.) International Migration in
Southeast Asia, Institute of Southeast Asia Studies, Singapore, pp. 273-295.
Ramlah Daud, 2005, Pekebun kecil warga asing di Kampung Kembara Jaya,Daerah
Kinabatangan, Sabah (Foreign smallholders in Kampung Kembara Jaya,
Kinabatangan District, Sabah), Graduation Exercise, School of Social Sciences,
Universiti Malaysia Sabah.
Robertson, Philip S., Jr. 2008. ‘Migrant Workers in Malaysia – Issues, Concerns and
Points for Action’. Commissioned by the Fair Labor Association, October 2008.
Sairah Jool, (1997/98), Pendatang Indonesia dalam kegiatan penanaman sayur-
261
sayuran di daerah Kota Belud, Sabah (Indonesian immigrants in vegetable growing
activities in Kota Belud, Sabah). Graduation Exercise, Department of Southeast Asian
Studies, Faculty of Arts and Social Sciences, University of Malaya, Kuala Lumpur.
Suresh Narayanan and Yew-Wah Lai. 2005. ‘The Causes and Consequences of
Immigrant Labor in the Construction Sector in Malaysia’. International Migration,
Vol. 43 Issue 5, pp. 31-57.
United States of America Department of State. 2008. Trafficking in Persons Report

June 2008. US Department of State Publication, Office of the Secretary for
Democracy and Global Affairs and Bureau of Public Affairs, June 2008.
United States of America Department of State. 2009. Trafficking in Persons Report

June 2009. US Department of State Publication, Office of the Secretary for
Democracy and Global Affairs and Bureau of Public Affairs, June 2009.
Wong, Diana. 2006. ‘The Recruitment of Foreign Labor in Malaysia: From

Migration System to Guest Worker Regime’. In: Kaur, Amarjit and Ian Metcalfe,
(eds.) 2006. Mobility, Labor Migration and Border Controls in Asia. Great Britain:
Antony Rowe, pp. 213-227.
Official/Government Documents
Buku Panduan Dasar, Prosedur Dan Syarat-syarat Penggajian Pekerja Asing Di

Malaysia. Jabatan Tenaga Kerja Semenanjung Malaysia, Kementerina Sumber
Manusia, Putrajaya, 2006.
Akta Imigresen 1959/63 (Akta 155) & Peraturan-Peraturan Dan Akta Pasport 1966
(Akta 150) Hingga Mei 2003. International Law Book Services, Wisma ILBS, Bandar
Damansara Perdana, Selangor.
Anti-Trafficking in Persons Act 2007 (Act 670). Available at

http://www.niew.gov.my/niew/en/download/doc_details/367-anti-trafficking-in-persons-act-2007,
Accessed 28 August 2010.
Laporan Program Khas Pengampunan Tahun 2004-2005, Jabatan Imigresen Malaysia

, 1st March 2005.
Government of Malaysia, Economic Report (various years), Kuala Lumpur. MOF

(Ministry of Finance),
Government of Malaysia, Fourth Malaysia Plan, 1981-1985. Kuala Lumpur. National

Printing Department.
Government of Malaysia, Ninth Malaysia Plan, 2006-2010. Kuala Lumpur. National

Printing Department.
262
Chapter 12: A Cost-Benefit Analysis of the Legal Status of Migrant
Workers in Thailand
CHARAMPORN HOLUMYONG
and SUREEPORN PUNPUING
Institute for Population and Social Research (IPSR), Mahidol University
ABSTRACT: Thailand as a labor-receiving country with large porous borders with labor-
sending countries and as a result about 53 percent of the migrant workers in Thailand in
2009 was estimated to be unregistered. While data on the number of unregistered migrants in
Thailand are often spotty and unreliable, it seems likely that the number of unregistered
migrants has increased in recent years. The challenge for the Government is thus to design
policies to manage migrant flows by encouraging migrant workers to register. Drawing up
memorandums of understanding (MOUs) with labor-sending countries is a key policy tool in
this regard. This paper discusses the costs-benefits of Thailand MOU policies. It identifies
different aspects of MOU designs – the ease and costs of migration and remittances
processes, the duration of migrant worker contracts – that affect the success of MOUs and
makes recommendations on these aspects.
1. OVERVIEW OF INTERNATIONAL MIGRATION IN THAILAND
Thailand is relatively unique among the countries of East Asia in that it is both a
labor-sending and a labor-receiving country. As a labor-sending country, Thailand currently
has about 810,000 international migrants working in such countries as Taiwan, South Korea,
Singapore and Malaysia (World Bank, 2011). Many of these migrant workers are skilled and
semi-skilled. As a labor-receiving country, Thailand hosts about 2.8 million migrant workers
(Table 1). Most of these workers are unskilled and come from 3 neighboring countries:
Cambodia, the Lao People’s Democratic Republic (Lao PDR) and Myanmar.
The focus of this paper is on Thailand as a labor-receiving country and so it is

important to note that many of the migrant workers in Thailand are unregistered and
undocumented. According to Table 1, in 2009 about 53 percent of the migrant workers in
Thailand were unregistered. While data on the number of unregistered migrants in Thailand
are often spotty and unreliable, it seems likely that the number of unregistered migrants has
increased in recent years.
Disparities in income levels and economic development are a crucial pull factor
drawing both registered and unregistered migrants to come to work in Thailand. While
poverty and low incomes are probably the main motivating factors for migrants from
Cambodia and Lao PDR, political and ethnic conflict are probably the leading cause of
migration from Myanmar.
263
Table 1: Estimated number of foreign workers in Thailand, 1996- 2009
Year Registered Non Registered Total
1996 293,652 700,000 406,348
1997 293,652 424,037 717,689
1998 90,911 870,556 961,467
1999 99,974 886,915 986,889
2000 99,956 563,820 663,776
2001 568,249 281,751 850,000
2002 409,339 558,910 968,249
2003 288,780 711,220 1,000,000
2004 849,552 149,848 999,400
2005 705,293 807,294 1,512,587
2006 668,576 1,104,773 1,773,349
2007 546,272 1,253,728 1,800,000
2008 501,570 1,298,430 1,800,000
2009 1,310,690 1,492,260 2,802,950
Sources: Martin (2007), Rattanarut (2009)
Migrant workers from these three neighboring countries make a large contribution to
the Thai economy. For example, Sussangkarn (1996) has estimated that migrants increased
Thai GDP by half of one percent in 1995. As Thailand has emerged as a major migration hub
in South East Asia1 (Sciortino and Punpuing 2009), migrant workers have helped fill jobs that
are either unattractive or undesirable to Thai workers. Migrants currently make up
approximately 5 per cent of the Thai labor force. According to Sciortino et al (2009), their
contributions to the nation reached an estimated USD 2 billion in 20052 (Martin, 2007 cited in
Sciortino et al., 2009). Besides filling jobs that are unattractive to Thai workers, migrant
workers sustain the competitive capacity of labor-intensive industries by expanding the
supply of cheap workers and lowering the costs of production. This enhances the ability of
Thai firms and enterprises to compete in the global market. Recently, more than 300,000
employers in Thailand have placed their inquiries for low skilled migrant workers from
Myanmar, Cambodia and Lao People’s Democratic Republic (PDR).
The demand and supply of labor in Thailand and its neighboring countries are partly
determined by demographic factors. In 2010, Thailand’s population was about 66 million,
while Myanmar, Cambodia and Lao People’s Democratic Republic (PDR) had populations of
53, 15 and 7 million respectively (Table 2). Lao PDR population is about half of the
Cambodian population, and eight times less than that of Myanmar. However, it is estimated
that in 2025, Lao PDR’s population will almost double. During 1995-2025, Cambodia,
Myanmar and Thailand’s populations increased only by 69%, 40% and 20% respectively
(U.S. Census Bureau, International Data Base 2010).
Table 2 Population in Lao PDR, Cambodia, Myanmar and Thailand (in thousands)
1995-2025
Country 1995 2005 2010 2015 2025

Lao PDR 4,846 6,216 6,994 7,811 9,456
Cambodia 11,228 13,527 14,753 16,148 18,967
Myanmar 43,994 50,572 53,414 56,320 61,748
Thailand 58,883 64,251 66,405 68,210 70,755
Source: U.S. Census Bureau, International Data Base 2010
1
Thailand together with Malaysia, Singapore, and Brunei is a major migration hub in South East Asia. [Asis, M.
(2004) ‘Not Here for Good?; International Migration Realities and Prospects in Asia’ The Japanese Journal of
Population 2(1)]
2
The migrants’ contribution was approximated as 1.25 percent of Thai GDP in 2005.
264
In the future it is anticipated that the flow of migrant workers into Thailand will
continue to grow. During the past decade the number of registered migrants from Myanmar
doubled, and the number from Laos and Cambodia increased by 2.2 and 4.6 times,
respectively. On the demand side, Thailand in the coming decade will probably face a decline
in its rates of labor force participation. Without the presence of foreign workers, labor
shortages could develop in several key sectors of the economy.
Migrants in Thailand
In Thailand migrants are concentrated in Bangkok and its vicinities, the southern part
of Thailand and in border provinces around the Thai-Myanmar border (figure 1). The top 10
provinces for migrant workers are mostly large cities such as Bangkok, Chiang Mai and
Phuket. Bangkok alone hosts almost one fifth of all registered migrant workers in Thailand.
Border towns also provide crucial gateways for migrant workers into the country. Ranong,
Tak, Kanchanaburi, Ratchaburi and Trat Provinces represent high migrant-populated areas.
Sources:1. Thailand map- http://gemselect.com

2.Calculated from a database of the Office of Foreign Workers
Administration, MOL (2009)
Figure 1 Top 20 provinces maintaining high number of registered migrant workers
Migrants in Thailand are concentrated in particular occupations and sectors,

specifically agriculture, construction, fishing and fish processing, as well as domestic works
(Table 3). For example, more than half of all workers in domestic works are migrants.
Similarly, while about 15% of migrant workers are in fishing and fish processing sectors, less
than 2% of Thai workers work in this sector. However, proportions of Thai workers in
265
agriculture remain higher than that of the migrants, at 25% and 40% respectively (Martin,
2007).
Table 3 GMS migrant population by selected occupations and sectors, 2002- 2011
Year Laborers Fish Construction Domestic
Fisheries Processing Helpers
Agriculture
2002 153,269 842,217 N/A 65,052
2003 113,710 66,523 N/A 52,685
2004 N/A 60,566 73,281 182,604 124,365 126,297
2005 N/A 61,202 75,117 182,673 124,790 126,369
2006 583,580 23,708 80,743 127,028 106,614 84,996
2007 484,723 15,809 68,290 103,255 86,041 61,549
2008 447,637 9,836 58,890 92,200 76,206 53,933
2009 1,184,592 56,578 136,973 221,703 220,236 129,790
2010 844,329 28,918 101,849 171,857 148,211 87,926
2011 362,130 14,190 46,040 103,120 52,911 32,773
Source: Database of the Office of Foreign Workers Administration, MOL (2002-2011)
2. MIGRATION MANAGEMENT POLICIES
This section will focus on migration management policies relating to the flow of
migrant workers from Lao PDR, Cambodia, and Myanmar. Managing migration flows from
these three countries is of great policy significance for the Thai government. In addition to the
broad framework of the Alien Employment Act B.E. 2551, crucial efforts of the government
effort to manage migration inflows include registration decrees, bilateral agreements,
National Verification (NV) and Memorandum of Understanding (MOU) on Cooperation for
the Employment of Workers.
The Alien Employment Act B.E. 2551 was enacted as a long-term measure to assist
in the management of labor migrants from neighboring countries. Regularization systems,
together with NV, have been implemented in the general belief that migrants from the three
neighboring countries are temporary. The bilateral MOUs have been elaborated to fulfill
demand for low skilled labor under the official employment systems of both sending and
receiving nations.
Migration Registration
Since the early 1990s, Thailand has faced an influx of migrants into the country,
particularly from Myanmar. On the one hand, most of these migrants could probably not
obtain a visa or border pass that would allow them to legally work in Thailand. However, in
special circumstances, a visa can be exempted for visitors from selected countries; these
visitors can obtain a border pass that will enable them to work in the country.
Anxious to address employers’ demands for low-wage workers, the Thai cabinet
made a series of decisions to classify migrant workers and to ensure their entry either into
worker registration systems or temporarily displaced persons camps. Different classifications
of undocumented persons in Thailand include displaced persons, undocumented migrants,
refugees from threats of war, students/intellectuals, visitors who overstayed their Thai visas
and undocumented migrant workers (Soe Soe 2002).
Cabinet decisions allowed employers to hire foreign workers by rapidly expanding

the size and scope of allowed work through the worker registration system. The first cabinet
decision in 1988 permitted employers in 10 provinces along the Thai-Myanmar borders to
266
register workers. In 1993, the Cabinet allowed registration of workers in fishing and fish
processing in 22 coastal provinces. 1996 was the first time cabinet decisions established a
system of 2-year work permits for foreign workers in 39 provinces and 8 industries, and later
expanded to cover 43 provinces and 11 industries. In 1998, health fees were included in
registration fees with the expanded scope of the foreign workers to cover 54 provinces and 47
job types. (Huguet and Punpuing 2005; Archavanitkul and Vajanasara 2008).
In July 2003, Thailand’s National Security Council adopted a resolution on the

management of irregular migration concerning registration and work permit concepts. This
resolution contained six main approaches, including: i) accepting the use of irregular migrant
workers in some sectors but limiting the overall number by considering demand by sectors;
ii) preparing individual records and identification cards for migrant workers; iii) employing
only migrant workers and not their family members; iv) ensuring that proper wages are paid;
v) implementing effective repatriation measures and vi) developing the economy of regions
on the Thai border in order to reduce the volume of migration into urban areas.
In order to get a work permit from the MOL, irregular migrants and their dependents
need to register with the Ministry of Interior (MOI) at their place of residence. The MOI
records all persons living in Thailand while the MOL issues work permits to only migrant
workers with an employer’s guarantee. In the process of work permit applications, the
migrant workers must obtain a medical check-up at a designated hospital or clinics. The
migrant workers are then eligible for medical treatment as if they were a Thai national. In
2004, there were 1,284,920 migrant workers (including migrants of all ages) registered with
the MOI but only 831,275 workers had applied for work permits from the MOL.
In 2006, the cabinet allowed re-registration of the migrant population which led the
number of migrant workers registered with the MOI to increase to 1,523,289 (of which 55%
are males and 45% are females) (Archavanitkul & Vajanasara 2008; MOL, 2009).
In 2008, the Thai government approved the Working of Aliens Act B.E 2551 by
incorporating lessons from the 2004 registration. However, the registration system remains
based on a renewal of work permits, which results in declining numbers of registered
migrants over time. For example, in 2007 the total number of registered migrants was
546,272 and by 2009 it was only 382,541. However, in 2009 the Thai government allowed
new migrants to register, which resulted in 1,094,984 registered migrants of which 939,940
(85.8%) were migrants from Myanmar, 84,166 (7.7%) from Lao PDR and 70,778 (6.5%)
from Cambodia (MOL, 2009).
Although the Thai government has attempted to improve the process of migration
registration, there are still a large number of unregistered migrants in the country. Factors
that lead migrants to remain unregistered include a lack of information about the process of
registration and an inability to travel and register when employers refuse to participate.
Another reason is that registered migrants still do not receive full protection in terms of
working conditions, wage rates and basic human rights. Finally, because of the high costs of
registration, many families cannot afford to register all their adult family members (Punpuing
et al., 2005).
In principle, employers must pay for the work permit fees. However, in practice,
many employers refuse to do this. Many employers also keep the original copy of their
employee’s work permit or passport in order to prevent their migrant workers from moving
away. Originally migrant workers were not allowed to change their employers and travel out
of their current workplace district. Such restrictions resulted in all types of worker
exploitation (Punpuing et al., 2006). However, since 2008, work permit holders have been
allowed to change their employers if they agree to pay for a new work permit.
267
Even after migrant workers go through the timely process of applying for and
receiving a work permit, that permit is only valid for one year. Martin (2004) has
recommended that work permits should be issued for two years with the fees varying
according to duration of employment (for example, a lower fee for seasonal work in
agriculture or for workers who do not want to be away from home for more than one year). A
2-year work permit could provide greater job security for both employers and employees
(Martin, 2004).
Memorandum of Understanding (MOU)
The Thai government signed MOUs on Cooperation for the Employment of Workers
with Cambodia, Lao PDR and Myanmar to develop the system of labor import (IOM, 2007)3.
The MOUs allow migrants of these countries to legally migrate and work in Thailand for a
maximum of four year4 employment terms. The migrants who complete four years of
employment must then take a three-year break before re-applying for employment in
Thailand.
The main objectives of the Memorandum of Understanding (MOU) are: i) to ensure a

proper procedure of employment of migrant workers; ii) to ensure effective repatriation of
migrant workers; iii) to increase protection of migrant workers and iv) to prevent flow of
undocumented cross-border migrants, trafficking, and employment of undocumented migrant
workers.
While the MOUs with Lao PDR and Cambodia have been effective since 2004, the
MOU with Myanmar was not implemented until late 2009. By September 2009, there were a
total of 26,720 imported workers (11,877 from Lao PDR and 14,843 from Cambodia).
While the MOU system is a potentially useful means for managing migration from
neighboring countries, many loopholes still exist. One essential concern is the time duration
of the work permit. It is important that the MOUs specify that contract workers can stay for a
maximum of 4 years, with the condition of returning after a three years break. However, the
current registration and work permit system does not put any limits on the maximum duration
of stay. Previous studies found a significant proportion of registered migrant workers have
been in Thailand for 5 years or longer. The period that migrants are entitled to work in
Thailand affects their net gain from migration. Hence the shorter time period of the MOU
relative to that of the registering system could discourage migrants from participating in the
MOU schemes.
To date, migrant participation under the MOU system has been quite limited. The
requested number of migrants made by employers has been hard to meet. For example, in
2006 employers requested 51,000 Laos and 17,500 Cambodian workers. However, only 7%
of the requested Lao PDR migrants and 3% of the requested Cambodian migrants actually
arrived. Moreover, up to 10% of migrant workers who migrated under the MOU system left
Thailand before their 2-year terms expired (Vasuprasat, 2007).
Time consuming processes and high costs have limited the progress of the MOU. To
import migrants under the MOU system, employers must process their requests 2 to 3 months
3
i. Thailand –Lao PDR- MOUs on Employment Cooperation, signed by Thai Minister of Labor and Lao Minister
of Labor and Social Welfare on 18 October 2002.
ii. Thailand-Cambodia- MOUs on Cooperation in the Employment workers, signed by Thai Minister of Labor
and Cambodian Minister of Social Affairs, Labor, Vocational Training and Youth Rehabilitation on 31 May
2003.
iii. Thailand – Myanmar- MOUs on Cooperation in the Employment of Workers, signed by Thai Minister of
Foreign Affairs and Myanmar Minister of Foreign Affairs on June 2003.
4
The work permit is valid for 2 years and can be renew once for another 2-year round.
268
in advance. Furthermore, importing migrant workers under the MOU system is expensive.
Migrant workers pay approximately 20,000 Baht or USD 588 to migrate under the MOU
system. With these high costs, employers usually advance the payment to the employment
agents and these amounts are later deducted from the migrants’ salary. However, the high cost
to import migrants under the MOU has been a burden to employers. For example, a factory
requesting 100 migrants will have to advance 2 million Bath or USD 58,824 to import
migrant workers.
Nationality Verification
In recent years, the Thai government has proclaimed its desire to eliminate irregular
migration. To achieve this goal, in 2003 the government established a Nationality Verification
(NV) process for Cambodian, Laotian and Myanmar workers. This scheme aims at
documenting the large number of irregular migrants who already reside and work in Thailand.
By November 2009, 125,156 migrant workers had their nationality verified through
the NV system. Of these 62,020 were Cambodia migrants, 58,430 were Lao migrants and
4,706 were migrants from Myanmar. While these numbers are encouraging, they almost
certainly represent only a small proportion of the undocumented migrants currently residing
in Thailand. In particular, the number of migrants from Myanmar registering with the NV
system seems especially small. Important reasons for the limited ability of the NV system to
reach more undocumented migrants include: i) rumors regarding an unofficial tax to be
collected from the relatives of migrants in the origin community; and ii) the possibility that
migrants wishing to register might be arrested by the Myanmar government. While there has
been no evidence of unofficial tax and arrest, Thai and Myanmar governments have launched
various campaigns to dispel these rumors.
The procedures of NV are complex, time consuming and costly. The workers were
originally required to register for the NV process by February 28, 2010.5. Upon the
completion of the request for NV, migrants must complete their biographical information in
the application forms received by their employers either from MOL or Provincial
Employment Offices (PEO). Later on, the Department of Employment submits NV
application forms to origin countries’ authorities (International Organization for Migration
(IOM) 2009). Once their identities are verified, migrant workers are granted temporary
passports. They will then be entitled to apply for a visa and a 2-year work permit. Going
through all steps of the NV process can take months. The time required to complete the NV
process is especially long for migrants from Myanmar. While the NV operational centers for
Lao and Cambodian migrant workers are inside Thai territory, migrant workers from
Myanmar must cross the border to report themselves to the Myanmar Temporary Passport
Issuance Offices. The centers are located three major border towns of Tachileik and
Myawaddy.
Migrant workers also have to pay high costs for NV. Costs incurred from all steps
(including food, transportation and procedural fees) are estimated to total 9,000 Baht or USD
2656. Later on, migrants must pay another 3,980 Baht or USD 117 for a work permit, medical
check-up, and health insurance (Naewna, 2010).
Nationality verification has also been viewed as a sensitive issue. Human rights
activists have become concerned about its complications and unreasonably high costs that
5
The deadline to register to enter the NV process was February 28, 2010 which later was extended to be March 2,
2010. Thai government has made clear (in April 2010) that there will be no more round of NV registering process
after March 2, 2010.
6
The costs of NV are estimated by, Research Director for Labor Development, Human Resources and Social
Development Program at the Thailand Development Research Institute (TDRI), 2009.
269
could lead to human rights violations and human trafficking (Pichai, 2009). For example,
some groups of migrants – such as Rohingya, Shan and migrant workers born in Thai territory
-- cannot verify their nationality. Their future status remains unclear.
Provincial Policy: Porous Border Management
In Thailand the demand for low skilled migrant labor is especially high in the border
regions. In these border areas there is a continuing need for low skilled labor in such sectors
as agriculture, fishing, animal husbandry and services.
The Alien Employment Act B.E. 2551 (2008) permits migrant workers from
neighboring countries to work within the border areas and the contiguous areas of Thailand.
In these areas the provincial government is given the authority to manage migration and the
levy system is used to control the number of migrant workers. Levies vary according to
locations, sectors, and types of work, ranging from 200-600 Baht, approximately USD 6 – 18,
per migrant worker.
The management of cross-border migrant workers varies across provinces. In general,

immigration officers monitor border crossings at the checkpoints while voluntary ranger
forces and border patrol policemen manage migrant workers along the border. For
undocumented migrants, there are many times when migration management procedures
become based on a so-called ‘promise system,’ whereby employers and migrants promise to
return to the border checkpoint at the due time in order to maintain their credits for future
migration. However, Holumyong and Sciortino (2009) found that migrant workers who
planned to move to other provinces or those who wished to permanently stay in Thailand do
not usually come through these checkpoints but are rather smuggled across borders.7
Curbing illegal migration in the border regions is a challenging task. Comprehensive

policies are needed to maximize the advantages of migration and the net benefits for the Thai
economy while simultaneously strengthening the protection of legal migrants. In the past
year, the Ministry of Labor (MOL) has established seven strategies for implementing its
migration policy. These strategies include: managing alien worker procedure, determining
alien worker employment standards, strengthening the border, stopping new irregular
migrants, managing deportations, implementing public relation and follow up. These
strategies have as their ultimate goal ‘promoting the employment of documented migrant
workers while eradicating irregular migration’.
3. THE APPLIED COST-BENEFIT ANALYSIS MODEL
On the basis of the preceding, it is clear that the Thai Government has had a difficult
time managing the inflow of migrant workers. The ad-hoc, short-term registration policies of
the Government have created uncertainty in the labor market and this has led to large
fluctuations in the number of registered migrant workers over time. Because of its
complexity and high cost, the NV scheme has had little impact on the flow of migrant
workers into Thailand. While the MOU system will probably be used to supply most migrant
workers in the future, the small number of migrants currently participating in this system
points to various problems in this program.
The Thai Government has not yet developed a comprehensive policy capable of
managing the migration needs of the 2.8 million migrant workers living within its borders.
Since devising such a comprehensive policy for managing migration must perforce consider
the decision-making process of the migrant, this section seeks to analyze how individuals
7
Interviewed information provided by Aranyaprathet Immigration Office and migrant workers at the
border in Srakeaw, and Trat provinces.
270
make migration decisions using an applied cost and benefit model. In this model the net costs
and gains to migrant workers under three types of migration status -- the MOU system,
registered migrant workers and irregular migrant workers – are measured and compared.
Theories and models used in the analysis
In economic terms, migration is an investment decision. It is caused by unequal wage

rates and job opportunities between two areas of a country or across two (or more) countries
of the world. From the standpoint of the migrant, it is expected that the returns from
migration will be higher than the costs of migration, and that the migrant possesses adequate
information about the costs and benefits of moving (Todaro 1976; Todaro and Harris 1976).
Spear (1971) developed a formal economic model based on these insights.

Specifically, Spear (1971) developed a migration model employing Sjasstad’s (1962) concept
showing that a migrants’ decision to migrate was determined by the difference in earnings
between origin and destination areas, the cost of moving, the discount rate, and the expected
number of earning years. His model can be written as follows;
Mij = (Ydj – Yi) - T

N (1+r)j
Where:
Mij = migration from area i to j
Ydj = earnings in qth year at destination
Yi = earnings in qth year at origin
T = cost of moving
N = Number of years in which earnings are expected
r = discount rate on future earnings
Our model for analyzing the migration status of an individual is based on Spear’s
migration model. In our model the choices of migration status are limited to three: (a) being
irregular migrants; (b) being registered migrants; and (c) being migrants under the MOU. Our
model includes both monetary costs and returns and non-monetary factors. The monetary
costs and returns include any measurable monetary factors. Hence net annual income is based
on the estimated annual wage rate. Net incurred costs include the total costs of migrating to
Thailand as well as the monetary costs for maintaining legal status (such as official fees,
document costs, agency fees, and health insurance). The non-monetary factors related to
migration include those social benefits and psychic factors associated with being a
documented or undocumented migrant worker.
Three variables – health benefits, employment benefits and social benefits – are
included in the model in the form of dummy variables. Each migrant worker who receives a
benefit is assigned as 1, and those who do not receive that benefit are be assigned as 0. Social
benefit scores for each migrant worker range from 0 to 3, with a higher score indicating
higher social benefits.
Psychic factors are also included in the model. These psychic factors are divided into
5 categories, including the probability of obtaining a job, employment security, wage, over
time employment, and safety and freedom from harassment8. Since 5 categories are
8
Psychic cost and benefit scores are calculated based on migrants’ perspective on each migration
status- being irregular migrants, registered migrants, and migrants under the MOUs. Advantage of each
category based on migrants’ perspective will add one score on the psychic benefits of that certain
migration status. On the other hand, disadvantage of each category will add one score on the psychic
cost. For example, if a migrant worker agrees that being registered increase the probability of finding
employment, one score will be granted in the psychic benefit part of registered migrants. Since 5
271
considered, psychic benefit and psychic cost scores for each migration status range from 0 to
5. However, the psychic costs indicate the negative impact on the psychic factors. Combining
psychic benefit and cost, psychic factor scores range from -5 to 5. Positive numbers indicate
that psychic benefits outweigh the psychic costs while negative numbers indicate that psychic
costs outweigh the psychic benefits.
The full cost-benefit model used in our analysis, including monetary costs and
returns, non-monetary factors, and the probability of obtaining employment, can be expressed
as model A.
Model A
Net gain of each migration status is calculated in the form;

 Y YN   T TN  
MOU  ρ  Y0  1  ...  N 
 T0  1  ...     s   ph
 (1  r ) (1  r )   (1  r ) (1  r ) N  
 Y YN   T TN  
REG  ρ  Y0  1  ...  N 
 T0  1  ...     s   ph
 (1  r ) (1  r )   (1  r ) (1  r ) N  
 Y YN   T TN  
NON  ρ  Y0  1  ...  N 
 T0  1  ...     s   ph
 (1  r ) (1  r )   (1  r ) (1  r ) N  
Where
MOU = The difference of the returns and incurred costs of
migration through the MOUs.
REG = The difference of the returns and incurred costs of
migration of registered migrants.
NON = The difference of the returns and incurred costs of
migration of irregular migrants.
YN = Net income in year N
TN = Net incurred cost in year N
r = Discount rate
N = Number of years expected to be employed in
Thailand
ρ = The probability of obtaining employment.
s = Net social benefits
 ph = Difference of the cost and benefit psychic factors
Employment duration
In the model the number of years expected to be employed in Thailand varies by an

individual’s migration status. For most migrants, we assume that the number of years
expected to be employed in Thailand is 4 years9. For MOU migrants, a worker receives a two-
year work permit which can be renewed one time for another 2 years. In our model four years
categories are considered, psychic benefit score of each migration status will range from 0 to 5. If a
migrant worker agrees that being irregular reduce the probability of finding employment, one score will
be granted in the psychic cost part of irregular migrants. Since 5 categories are considered, psychic cost
score of each migration status will range from 0 to 5. The psychic costs indicate the negative impact on
the psychic factors.
9
One-year and two-year employment durations are also calculated and included in the analysis for
comparison.
272
is also the expected term of employment for registered and irregular migrant workers (see
Punpuing et al., 2006).
Source of data
Our model is based on data collected from a cost-benefit analysis survey in Thailand.
This survey was conducted in Nakorn Pathom, Rayong and Ubon-Rachathani Provinces
during October 2009 to April 2010. As shown in Table 4, the sample includes a total of 111
Laos and Cambodian migrant workers in the construction and domestic sectors.
Table 4. Number of migrant workers by migration status included in the survey
# of migrants Construction Domestic work Total

workers
Migration status
MOUs 9 16 25
Registered 32 22 54
Irregular 12 20 32
Total 53 58 111
Ethnicity
Lao 4 41 45
Cambodian 49 17 66
Total 53 58 111
Source: Thailand cost-benefit analysis survey, 2010.
The survey included two very different sectors of the economy. The construction
sector in Thailand includes some of the higher-paying jobs open to migrant workers. By
contrast, the domestic sector includes some of the lowest paying jobs for migrants. This
sector is not protected by working conditions laws or minimum wage provisions. In the
survey both migrant workers and employers were interviewed in order to pinpoint the key
differences between the regulatory, immigration and institutional environments affecting
formal and informal migration. The survey questionnaire included a variety of questions
related to the monetary costs and benefits, social benefits and psychic factors related to each
migration status.
The cost-benefit analysis on migration status
Our analytical strategy uses Model A to compare net gains from migration among the
three types of migration status: (a) irregular migrants (NON); (b) registered migrants (REG);
and (c) MOU migrants (MOU). Our analysis shows the difference in net gains among the
three migration states of Laotian migrant workers and Cambodian migrant workers in
construction and domestic working sectors. Tables 5 and 6 show monetary benefits and costs
for the various groups of workers. In Table 5 monetary benefits are reported in five groups -
monetary benefits in year 1(By1), monetary benefits in year 2(By2), monetary benefits in year
273
3(By3), monetary benefits in year 4(By4), and total monetary benefits (Btotal money). The
monetary benefit for each migrant group is the average net annual income calculated from the
survey data10subjected to the discount rate. Net income in each year includes the estimated
average annual wage rate of Laotian and Cambodian migrant workers in each sector. In Table
6 monetary costs are reported in five groups - monetary costs in year 1(Cy1), monetary costs in
year 2(Cy2), monetary costs in year 3(Cy3), monetary costs in year 4(Cy4), and total monetary
costs (Ctotal money). The monetary cost for each migrant group is the average net cost calculated
from the survey data11subjected to the discount rate. Net incurred cost includes official fees to
migrate to Thailand and to maintain their migration status in Thailand. Table 5 shows that
monetary benefits are highest for migrants under the MOU in all groups, except for
Cambodian migrants in the domestic work sector. Table 6, however, shows that monetary
costs are highest under the MOU for all groups. Costs incurred by MOU migrants are
approximately twice as high as those incurred by registered migrants, and roughly 2-9 times
higher than those incurred by irregular migrants.
10
Due to survey field limitation, Laotian migrant workers under the MOUs system in the construction sector are
not included in this survey. Their monetary benefit is then calculated based on the ratio of the monetary benefit of
Cambodian migrant workers under the MOUs system in the construction sector to that of Cambodian registered
migrant workers in the construction sector.
11
Due to survey field limitation, Laotian migrant workers under the MOUs system in the construction sector are
not included in the survey. Their monetary cost is then calculated based on the cost information received from the
labor-importing agency and Office of Foreign Workers Administration.
274
Table 5. Monetary benefits of Laos and Cambodian migrant workers in construction and domestic working sectors
By1 By2 By3 By4 B total money
MOU 70,334 68,955 67,603 66,278 273,171
REG* 69,390 61,226 53,356 45,771 229,744
Lao migrant Construction NON* 44,820 43,941 43,080 42,235
workers sector * 174,076
MOU 61,980 60,765 59,573 58,405 240,723
Domestic REG* 60,488 53,371 46,511 39,899 200,269
work sector NON* 53,964 52,906 51,869 50,851
* 209,590
MOU 78,027 76,497 74,997 73,526 303,046
REG* 76,979 67,923 59,192 50,777 254,871
Cambodian Construction NON* 62,591 61,363 60,160 58,980
migrant workers sector * 243,094
MOU 60,600 59,412 58,247 57,105 235,363
Domestic REG* 57,980 51,159 44,583 38,242 191,967
work sector NON* 61,992 60,776 59,585 58,416
* 240,770
Source: Thailand cost-benefit analysis survey.
Note: discount rate (r) = 0.02
* the probability of obtaining employment (ρ) of a registered migrant workers start at 1 and diminishes by 0.1 yearly.
** the probability of obtaining employment (ρ) of an irregular migrant workers equal to 0.9.
275
Table 6. Monetary costs of Laos and Cambodian migrant workers in construction
and domestic working sectors
Cy1 Cy2 Cy3 Cy4 C total money
MOU 20,000 2,110 4,375 2,028 28,513
REG* 4,688 3,575 3,116 2,673 14,051
Lao Construction NON 2,050 3,529 3,460 3,392
migrant ** 12,432
workers MOU 20,000 2,769 5,021 2,661 30,451
Domestic REG* 4,962 4,391 3,826 3,282 16,461
work NON 1,431 1,000 980 961
** 4,373
MOU 20,000 2,027 4,295 1,949 28,271
REG* 6,169 4,158 3,623 3,108 17,058
Cambodian Construction NON 2,423 1,859 1,822 1,787
migrant ** 7,891
workers MOU 20,000 2,122 4,387 2,039 28,548
Domestic REG* 6,602 4,687 4,085 3,504 18,877
work NON 2,200 447 438 430
** 3,515
* the probability of obtaining employment (ρ) of a registered migrant workers start at 1 and
diminishes by 0.1 yearly.
The results of the net benefit computation are shown in Table 7. This table divides net
benefits into 3 groups - net monetary benefits, net social benefits and net psychic benefits. βmoney
provides a measure of net monetary gain. βs provides a measure of net social benefits gain, and βph
provides a measure of net psychic benefit gain.
Table 7: . Net gains of Laos and Cambodian migrant workers in construction

βmoney βs βph
MOU 244,658 2.2 3.67
Constructio REG* 215,693 2.25 3.67
Lao n NON*
migrant * 161,644 1.5 -1.33
workers MOU 210,272 2.2 3.66
Domestic REG* 183,808 2.49 3.29
work NON*
* 205,217 2.07 -1.95
MOU 274,776 2.11 2.94
Constructio REG* 237,813 1.98 2.27
Cambodian n NON*
migrant * 235,203 1.56 -1.06
workers MOU 206,816 2 2.24
Domestic REG* 173,090 1.1 1.65
work NON*
* 237,255 1 -1.88
276
Table 7 shows that net monetary benefits (βmoney ) are highest for migrants under the MOU in
all groups, except for Cambodian migrants in the domestic work sector. On the other hand, net
monetary benefits in the domestic work sector are the lowest for registered migrants. Nevertheless,
net monetary benefits of irregular migrants are not much different comparing to those of registered
migrants and migrant under the MOU.
In Table 7 net social benefits (βs ) include health benefits, employment benefits and benefits
related to basic human needs. According to the table, irregular migrant workers gain the smallest
social benefits of any group. The social benefits received by migrant workers under the MOU system
and the registered migrant program are roughly the same, except for Cambodian migrant workers in
the domestic work sector.164.
In Table 7 net psychic benefit gain (βph ) shows the average difference in the psychic costs
and benefits of migrant workers in each sector. According to the table, net psychic benefit gains are
positive for all groups of registered migrant workers and migrant workers under the MOU system.
However, net psychic benefit gains are negative for all groups of irregular migrants. These results
suggest that the psychic costs of being an irregular migrant are higher than the psychic benefits.
Among documented migrant workers, migrant workers under the MOU system rank the first and
registered migrant workers rank the second in terms of net psychic benefit gains. On the basis of these
results, we can conclude that the higher the level of legal migration status, the higher the gain of net
psychic benefits.
In Table 7 net social and psychic benefits are quite high for migrants under the MOU and
registered migrants. At the same time, net psychic benefits are negative for irregular migrants. Even
though their net monetary benefits are not much different when compared to registered and MOU
migrants, irregular migrants live with the fear of deportation and a high risk of exploitation.
The impacts of time-consuming MOUs process
As discussed above, importing migrant workers under the MOUs is a time-consuming process
that takes about 3 months165. For migrant workers, the time required to process MOUs reduces the net
gains of MOU migrants. During the 3-months of processing, the income of these migrants would be
zero. Table 8 shows the net gains and net benefits of the three types of migrant workers when the time
taken to process MOUs is taken into account.
Table 8 shows that the time required to process MOUs significantly reduces the net benefits
received by MOU migrants in the first year. This reduction in first year benefits makes the net gain of
Laotian migrants under the MOUs in construction smaller than that of registered migrants and very
close to that of irregular migrants. The impact of the processing time of MOUs is even worse for the
group of Laotian migrant workers in the domestic sector and Cambodian migrant workers. The net
gains of migrant workers under the MOUs are smaller than those of registered and irregular migrants.
164
Small size of sampling of Cambodian migrant workers in the domestic work sector in the survey could be the reason of
this large gap.
165
Information received from in-depth interviewed government authorities and MOUs agencies.
277
Table 8. Net gains of Laos and Cambodian migrant workers in construction
βmoney βs βph
MO 176,365**
U * 2.2 3.67
Lao REG
migrant * 215,693 2.25 3.67
workers Construction NON
** 161,644 1.5 -1.33
MO 150,091**
U * 2.2 3.66
REG
Domestic * 183,808 2.49 3.29
work NON
** 205,217 2.07 -1.95
MO 199,014**
U * 2.11 2.94
Cambodian REG
migrant * 237,813 1.98 2.27
workers Construction NON
** 235,203 1.56 -1.06
MO 147,975**
U * 2 2.24
REG
Domestic * 173,090 1.1 1.65
work NON
** 237,255 1 -1.88
*** Income are calculated under the time-consuming process (3-month processing period
The findings of Table 8 are important, because they serve to explain the small number of
migrant workers currently participating in MOUs. Migrants in the MOU program spend a lot of time
waiting to be processed and they lose a lot of net benefits during this waiting time. Many migrants
cannot afford this waiting time and so they tend not to participate in this program.
4 WHAT LESSONS CAN BE LEARNED FROM THE COST-BENEFIT MODEL?
The results of the cost-benefit model suggest the following lessons:
1. The time-consuming MOU process reduces net benefits and net gains enjoyed by migrant
workers under the MOU system. The timing process makes the monetary gains of being in the
MOUs system smaller than those of other migration statuses. In the future Thai migration
management policy should avoid such time consuming processes in order to more effectively
stimulate the labor market.
2. The net gains of migrating under the MOU system are larger in relative terms under a longer
time horizon. When considering their choice of migration status, migrant workers may tend to
use shorter time frames to consider their possible net gains under each migration status. The
278
shorter time frame would magnify the effect of the costs of fees and legal documents and
shrink the earning advantages of being documented.
3. The initial cost of migrating under the MOUs is high. Reducing the initial high costs of the
MOU system might encourage greater migrant participation in this system over the long run.
4. Effective repatriation measures could enlarge the benefits of migrating under the MOU
system. The repatriation funds paid to migrant workers while completing their employment
term could encourage them to remain working for their whole contract period. The
repatriation funds paid to migrants will increase the net gains and benefits of migrant workers
under the MOU system.
5. Effective basic right and benefits should be provided. The rights and benefits offered to
migrant workers could increase non-monetary benefits and net gains of migrant workers
under the MOU system.
In recent years MOUs have served as one legal channel for migrants wishing to move to
Thailand. This is a new step in migration management policy that corresponds to the labor needs of
the country. Even though the MOUs still need to be refined and improved, the scheme represents one
useful means for meeting the labor needs of both migrants and employers. As long as the MOUs can
offer the means for improving the lives of migrants, there is hope for using this scheme to manage
migration in Thailand. While many procedures still need to be adjusted, with the collaboration of all
parties including the government, NGOs, employers, and migrants, MOUs in particular and Thai
migration policies in general may now be headed towards what represents a more long-term strategy
for managing migration in Thailand.
279
REFERENCES
The Alien Employment Act B.E. 2551. 2008. The Government Gazette, 125(37).
Archavanitkul, A and Vajanasara, K. 2008. Employment of Migrant Workers Under the
Working of Aliens Act 2008 and the List of Occupation Allowed to Foreigners. International
Organization for Migration (IOM), Bangkok.
Asis, M. 2004. Not Here for Good?; International Migration Realities and Prospects in Asia.
The Japanese Journal of Population, 2(1) from http://www.ipss.go.jp/webj-
ad/WebJournal.files/population/2004_3/maruja/ 2004mar.pdf. retrieved 15st September 2008.
Awatsaya P., Khaing Maar Kyaw Zaw, Caouette, T, and Punpuing, S. 2008. Migrant
Domestic Workers: From Burma to Thailand. In Transborder Issues in the Greater Mekong
Sub-region edited by Thaweesit, S, Vail P. and Sciortino, R. Mekong Sub-region Social
Research Center (MSSRC). Ubon Ratchathani University, Thailand: 305-333.
Bryant, J., & Rukumnuaykit., P. 2007. Labor Migration in the Greater Mekong Sub-region :
Does immigration to Thailand reduce the wages of Thai workers ? Bangkok, Thailand:
World Bank.
Calvó-Armengol, A., & Jackson, M. O. 2004. The Effects of Social Networks on

Employment and Inequality The American Economic Review, 94(3), 426-454.
Chalamwong, Y., & Prugsamatz, R. 2009. The Economic Role of Migration Labor Migration
in Thailand: Recent Trends and Implications for Development. Thailand Development
Research Institute Quarterly Review, 24(3), 3-9.
Davanzo, J. 1981. ‘Microeconomic Approaches to Studying Migration Decision’, in

Migration Decision Making, edited by DeJong, G. and Gardner, R. New York: Peragamon:
90-129.
Findley, S.E. 1993. The Third World City: Development Policy and Issues. In Third World
Cities: Problems, Policies, and Prospects, edited by Kasarda, I.D. and Parnell, A.M. Sage
Publications, Newbury Park: 1-31.
Fah, H. K. 2010. Thai labor ministry confirms 'no new round of workers registration'. Shan
Herald,fromhttp://www.shanland.org/index.php?option=com_ content&view=article&id=3008:thai-
labor-ministry-confirms-no-new-round-of-workers-registration&catid=93:general&Itemid=291,
retrieved 1st May 2010.
Goldscheider, C. 1987. Migration and Social Structure. Sociological Forum, 2 (4): 674-696.
Holumyong, C. & Sciortino, R. 2009. Background Paper: Migration from GMS Countries to
Thai Border Provinces Bangkok, Thailand: The World Bank Group.
Hugo, G. 1994. Migration as a Survival Strategy: The Family Dimensions of Migration.

Paper presented at the United Nations Expert Group Meeting on population Distribution and
Migration, Santa Cruz, Bolivia, 18-22 January.
Huguet, J. W., & Punpuing, S. 2005. International Migration in Thailand. Bangkok,

Thailand: International Organization for Migration (IOM).
ILO/Japan-MOL. 2010. Background Paper. Paper presented at the ILO/Japan-MOL National
Seminar on Agenda for Labor Migration Policy in Thailand.
280
Ioannides, Y. M., & Soetevent, A. R. 2006. Wages and Employment in a Random Social
Network with Arbitrary Degree Distribution The American Economic Review, 96(2), 270-
274.
International Organization for Migration (IOM). 2010. Migrant Information Note April
2010; Nationality Verification Regulations: February - April 2010 updates.
International Organization for Migration (IOM). 2009. Migrant Information Note August
2009; Procedures for nationality verification of Myanmar/ Burma nationals in Thailand.
International Organization for Migration (IOM). 2007. Memorandum of Understanding.

Migrant News (Quarterly Newsletter).
Institute for Population and Social Research (IPSR). 2008. The Evaluation of the HIV/AIDS
Prevention Program among Migrant Workers in the Border Provinces in Thailand Under
Prevention of HIV/AIDS among Migrant Workers in Thailand (PHAMIT) supported by
Global fund.
IRIN. 2009. Thailand: Burmese Migrant Children Missing out on Education. IRIN, from
http://www.irinnews.org/report.aspx?ReportID=84844, retrieved 1st May, 2010.
Lee, E.S. 1966. A Theory of Migration. University of Pennsylvania, Demography, 3(11):

47-57.
Martin, P. 2007. The Contribution of Migrant Workers to Thailand: Towards Policy

Development. Bangkok: International Labor Organization.
Massey, D.S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A, Taylor, J.E. 1994. An
Evaluation of International Migration Theory: The North American Case, Population and
Development Review, 20 (4): 699-752.
Naewna. 2010. UN Focuses on Burmese Migrants Deported Plans of Thai Government after
the Deadline of Nationality Verification (in Thai) (UN-Jap-Ta-Thai-Plukdun-Rangngan-
Burma-Tsung-Mod-Vaela-Yuun-Kor-Pisoot-Sunchart)Naewna,from
http://www.naewna.com/news.asp?ID=200756, retrieved 1st May 2010.
Office of Foreign Workers Administration, Ministry of Labor. 2009. Database of Migrant
Workers from Cambodia, Lao PDR, and Myanmar, retrieved 9th December, 2009.
Population Reference Bureau (PRB). 2009. Literacy Rate, Ages 15-24,
http://www.prb.org/Datafinder/Topic/Bar.aspx?sort=v&order= d&variable=8, retrieved 20th August
2009.
Population Reference Bureau (PRB). 2009. Secondary School Enrollment (as% of school-age
enrollment) http://www.prb.org/Datafinder/Topic/Bar.aspx? sort=v&order= d&variable=60,
retrieved 20th August 2009.
Pichai, U. 2009. UN Rapporteur urged to probe nationality verification Mizzima, from

http://www.mizzima.com/news/regional/2788-un-rapporteur-urged-to-probe-nationality-
verification.html, retrieved 18th September 2009.
Punpuing, S., Awatsaya P., Khaing Maar Kyaw Zaw, and Caouette, T. 2005. Migrant
Domestic Workers: From Burma to Thailand. Publication no. .Institute for Population and
Social Research, Mahidol University (in Thai).
281
Punpuing, S., Jampaklay, A., Kittisuksathit, S., Prohmmo, A., & Pearson, E. 2006. The
Mekong Callenge: Underpaid, Overworked and Overlooked (Vol. 2). Bangkok: International
Labor Organization (ILO).
Ravenstein, E.G. 1889. The Laws of Migration. Journal of the Royal Statistical Society, 52
(2): 241-301.
Rattanarut, N. 2009. Migration Policy in Thailand, from

http://www.cdri.org.kh/dan/conf78oct09/d1/1.%20Mr%20Nara%20Rattanarut%20-%20Thailand.pdf,
retrieved 18th September 2009.
Sciortino, R and Punpuing,S. 2009. International Migration in Thailand, 2009.

International Organization for Migration (IOM), Bangkok office, Thailand.
Sjaastad, L. A. 1962. The Cost and Returns of Human Migration. The Journal of Political
Economy, 70(5): 80-93.
Skeldon, R. 1997. Migration and Development: A Global Perspective, London. Longman.
Soe Soe. 2002. Migration Report: Burma: Identifying the Issues and Needs of Migration from
Burma into Thailand. Joint Research Project of the Federation of Trade Unions Burma and
Asian Migrant Center.
Speare, A. 1971. A Cost-Benefit Model of Rural to Urban Migration in Taiwan. Population

Studies, 25(1), 117-130.
Stark, O. 1991. The Migration of Labor, London: Basil Blackwell (Chapters 1-3).
Thongyou, M. 2009. Transnational Labor Migration: A Synthesis Study of Migration

Policies and Consequences of Migration in Thailand. Bangkok: Thailand Research Fund.
Todaro, M.P. 1976. Internal Migration in Developing Countries. Geneva: ILO. (Chapter 3).
Todaro, M. P. and Harris J. 1976. Urban Job Expansion, Induced Migration and Rising
Unemployment: a Formulation and Simplified Empirical Test for LDCs," Journal of
Development Economics, 3(3): 211-225.
U.S. Census Bureau, Population Division: International Data Base. 2010.

http://www.census.gov/ipc/www/idb/informationGateway.php, retrieved 20th August 2010.
Xue, L. 2007. Social Capital and Employment Entry of Recent Immigrants to Canada-
Evidence from the Longitudinal Survey of Immigrants to Canada (LSIC) Wave 1 and Wave 2
(Preliminary Findings). Paper presented at the 9th National Metropolis Conference. from
http://ceris.metropolis.net/ 9thMetropolisConference/WorkshopPresentations/B1_Xue.pdf.
Zelinsky, W. 1971. The Hypothesis of the Mobility Transition. The Geographical Review,
61 (2): 219-249.
282
CONTRIBUTORS
Late Mahani Zainal Abidin, formerly Chief Executive Officer, Institute of Strategic and International
Studies, Malaysia
Richard H. Adams, Jr , Consultant, World Bank
Ahmad Ahsan, Lead Economist, World Bank
Angel Aguiar, Research Economist, Purdue University
Syud Amer Ahmed, Economist, World Bank
Michael Alba, President, Far Eastern University, Philipines
Mahani Zainal Abidin, formerly Chief Executive Officer,Institute of Strategic and International
Studies, Malaysia
Mitzie I. P. Conchada, Assistant Professor, School of Economics, De La Salle University, Philippines
Alfredo Cuecuech, Economist and Faculty Member, El colegio de Tlaxcala
Arie Damayanti, Deputy Director, Graduate Program Economic Faculty, University of Indonesia
Charamporn Holumyong, Institute for Population and Social Research, Mahidol University
Ifa Isfandiarni, Research Associate, University of Indonesia
Azizah Kassim, Principal Research Fellow, Institute of Strategic and International Studies, Malaysia
Ari Kuncoro, Professor of Economics, University of Indonesia
Dilaka Lathapipat, Human Development Economist, World Bank
Liew Chei Siang, Faculty Member, Institute of Malaysian and International Studies, Faculty of
Economics and Management
Daniel Mont, Principal Research Associate, Instutional Research Information Service, University
College of London
Sureeporn Punpuing, Institute for Population and Social Research, Mahidol University
Nguyen Huyen Le, Institute of Labor Science
Trang Nguyen, Senior Economist, World bank
Ririn Purnamasari, Economist, World Bank
W John P. R. Rivera, Assistant Professor, School of Economics, De La Salle University, Philippines
Tham Siew Yean, Faculty Member, Institute of Malaysian and International Studies, Faculty of
Economics and Management
283
Terence Too, Institute of Strategic and International Studies, Malaysia
Tereso S. Tullao, Jr., Full Professor, School of Economics, De La Salle University, Philippines
Terrie Walmsley, Associate Professor, Purdue University
Yap Mui Teng , Senior Research Fellow, Institute of Policy Studies, Lee Kuan Yew School of Public
Policy, Singapore
284
World Bank Group
1818 H Street NW,
Washington DC 20433

Managing International Migration For Development in East Asia

Enviado por

Dados do documento

Título original

Direitos autorais

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Managing International Migration For Development in East Asia

Enviado por

Direitos autorais:

Public Disclosure Authorized

Managing International Migration

Richard H. Adams, Jr.

Office of the Chief Economist

Richard H. Adams, Jr.

Office of the Chief Economist

The World Bank

Rights and Permissions

TABLE OF CONTENTS .........................................................................................................................i

OVERVIEW ......................................................................................................................................... iii

PART 1. MIGRATION AND ITS IMPACTS

Chapter 2: Remittances, Household Investment and Poverty in Indonesia

Chapter 3: More or Less Consumption?: The Effect of Remittances on Filipino

Chapter 4: Impacts of International Migration and Remittances on Child Outcomes

Chapter 5: The Effects of Immigration on the Thai Wage Structure

PART 2: MANAGING MIGRATION

Chapter 7: Indonesia’s Regulatory, Institutional and Governance Structure for

Chapter 10: Singapore’s System for Managing Foreign Manpower

International migration, defined as the movement of people across international borders, is

OVERVIEW OF THE CHAPTERS

PART 1: MIGRATION, DEMOGRAPHY, POVERTY AND INVESTMENT

As suggested by Walmsley et al, one of the most important development effects of

PART 1: MIGRATION, LABOR PRODUCTIVITY AND WAGES

PART 2: MANAGING MIGRATION IN LABOR SENDING COUNTRIES

PART 2: MANAGING MIGRATION IN LABOR RECEIVING COUNTRIES

The applicability of this theoretical framework is supported by broader empirical analyses.

Economic Impacts of Demographic Change

Table 2: Percentage Change in Net Migration* per Annum

a) the accumulation of capital over time;

Table 3: Distribution of unskilled foreign workers across sectors in ASEAN countries

Demographic Changes and Real Wages

Figure 2: Forecasted Annual Growth Rates in Skilled Labor by Region

Source: Authors’ results

Source: Authors’ results

Source: Authors’ results

Figure 5: Cumulative Percentage Changes in Real Wages of Unskilled Workers due to

Source: Authors’ results

Total Forecasts Liberal Migration

Hong Kong Singapore

Yearly Growth Rates

Malaysia Rest E. Asia

Yearly Growth Rates

Source: Authors’ results

Hong Kong Thailand

0 Endogenous 0.2 Endogenous

Yearly Growth Rates

1 Endogenous 1.5 Forecasted Unskilled

Yearly Growth Rates

Source: Authors’ results

Malaysia / Indonesia South Korea / Rest of East Asia

Malaysia / Philippines Malaysia / Thailand Singapore / Malaysia

Source: Authors’ results

The Macroeconomic Impact of the Liberalization of Migration

Source: Authors’ results

The Sectoral Impact of the Liberalization of Migration

IMF (2010) "Balance of Payments Statistics Online" http://www.imfstatistics.org/, Retrieved on July 19

and Poverty in Indonesia

RICHARD H. ADAMS, Jr.*

2000 2000 2000 2007 2007 2007

Expenditure Description No Remittances No Remittances

Cs /EXP = βs + as /EXP + γs (log EXP) (1)

Estimating the Econometric Model with Selection Controls

Table 3. Conditional Logit Model on the Household Probability of Receiving Remittances,

Dependent variable: Does household receive international remittances in 2007

**Significant at the 0.05 level.

Variable Food Education Housing Health Other

0.328 0.323 0.385 0.361 (3.7)* 0.355 0.343 (3.8)*

0.261 0.258 0.218 0.225 (-26.7)* 0.208 0.217 (-16.3)*

0.010 0.014 0.050 0.010 (4.4)* 0.028 0.010 (2.0)

ATT (Percent difference) -29.17% -27.78%* -27.27%*

Food, beverage and 0.454* 0.433* 0.262*** -0.171 -39.5

Fuel, light and water 0.056* 0.062* 0.092*** 0.030 49.5

Education and medical 0.061* 0.056 0.088*** 0.033 58.9

Durable goods 0.078* 0.085* 0.080*** -0.005 -5.4

Transportation and 0.073* 0.083* 0.092*** 0.009 11.4

Housing 0.115* 0.119* 0.228*** 0.109 91.8

Home improvements and 0.038* 0.040* 0.024*** -0.017 -41.4

Other consumption items 0.125* 0.124* 0.134*** 0.011 8.6