93255 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results Note: a. Number of unskilled migrant workers by location/home region 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 Source: Authors’ results Note: a. Number of unskilled migrant workers by location/home region 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 Source: Authors’ results 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 Source: Authors’ results 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 Source: Authors’ results 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) Household Characteristics 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 Variable Food Education Housing Health Other .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 Variable Food Education Housing Health Other 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) Household Characteristics -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 Variable Food Education Housing Health Other 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 Household Characteristics 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   jmYˆ  m jm [7] Yˆ    x   V  m 0m m km m m 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: ˆ   E(e jm | z j  1, Xm )  0 jm  xmkjm   jmY [8] m jm m 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: ˆ E(ejm | z j  1, Xm )   0 jm  x m kjm   jmYm [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 ] m | z  1, X m ]   m   E[en | z  1, X m ]   n   E[e    en  E[em  | 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 , Y  ˆ ]  E[e | z  1, x , Y ]        m n m c m n  E[em ˆ  | z  1, x , Ym ]  E[em  | z  1, x m , Yc ] m      a  E[em n | z  1, x m , Yc ]   m   n   | z  1, x , Yc ]  E[e m       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: 1 1 QTE m  Fem (q | z  1, X m )  Fen (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 H_Qtile Counterfactual_Qtile Migrant OLS Counterfactual OLS Quantile OLS Figure 3a. OLS and Quantile Regression Estimates of Figure 3b. OLS and Quantile Regression Estimates 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_Qtile Counterfactual_Qtile Migrant H_OLS Counterfactual_OLS Quantile OLS 67 Figure 4a. OLS and Quantile Regression Estimates of Figure 4b. OLS and Quantile Regression Estimates Marginal Budget Shares (MBS), of Average Treatment Effect for the Migrant 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 H_Qtile Counterfactual_Qtile Migrant OLS Counterfactual OLS Quantile OLS zero Figure 5a. OLS and Quantile Regression Estimates of Figure 5b. OLS and Quantile Regression Estimates of 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_Qtile Counterfactual_Qtile Migrant H_OLS Counterfactual_OLS Quantile OLS zero Figure 6a. OLS and Quantile Regression Estimates of Figure 6b. OLS and Quantile Regression Estimates Marginal Budget Shares (MBS), of Average Treatment Effect for the Migrant 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 Migrant H_Qtile Counterfactual_Qtile Migrant H_OLS Counterfactual_OLS Quantile OLS zero 68 Figure 7a. OLS and Quantile Regression Estimates of Figure 7b. OLS and Quantile Regression Estimates of Marginal Budget Shares (MBS), Average Treatment Effect for the Migrant Households 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 Migrant H_Qtile Counterfactual_Qtile Migrant H_OLS Counterfactual_OLS Quantile OLS zero Figure 8a. OLS and Quantile Regression Estimates of Figure 8b. OLS and Quantile Regression Estimates of Marginal Budget Shares (MBS), Average Treatment Effect for the Migrant Households 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 Migrant H_Qtile Counterfactual_Qtile Migrant H_OLS Counterfactual_OLS Quantile OLS zero 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 Variables communication Housing furnishing goods [5] [6] [7] [8] coeff se coeff se coeff se coeff se 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] coeff se coeff se coeff se coeff se 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] coeff se coeff se coeff se coeff se 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 Variables communication Housing furnishing goods [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 Variables communication Housing furnishing goods [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: Cambridge University Press. 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) Number of household males over age 15 0.026*** 0.027*** 0.021*** 0.020*** 0.033*** 0.027*** 0.033*** 0.033*** 0.027*** 0.033*** (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 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 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 OLS OLS IV IV OLS IV IV OLS IV IV (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) Other controls Yes Yes Yes Yes Yes Yes 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 Sargan's test for overidentification (p-value) 0.534 0.427 0.751 0.674 0.503 0.429 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 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 OLS OLS IV IV OLS IV IV OLS IV IV (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) Number of household males over age 15 0.075*** 0.072*** 0.076*** 0.073*** 0.072*** 0.076*** 0.072*** 0.072*** 0.077*** 0.073*** (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) Other controls Yes Yes Yes Yes Yes Yes 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 Sargan's test for overidentification (p-value) 0.500 0.248 0.140 0.055 0.451 0.320 Robust standard errors in parentheses. For households with 6-18 children. 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 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) Migrant Households * Share of female 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) Household has international migrants, and they are all female --2007 -0.231 -11.578 0.218 (0.170) (9.299) (0.163) Migrant Households * Share of female 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) Household has international migrants, and they are all female --2007 -0.33 -1.849 0.282 (0.235) (8.064) (0.186) Migrant Households * Share of female 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 i su ( 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)  su i ( 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   j i j  , the within industry term W ( p)   j  j   , the interaction term I ( p)   j   j ( p)   j  i i  j ( p)   j   j ( p)   j  i i  , and the unemployment component U ( p )  su i ( p )  su i . 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: C ( p )   C  C   s ( p)  s    C , C   B, W , I , U  i i (2.3)  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  HL  HL (3.1) N p   Lp N Lp   Hp N Hp      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    1 F  (3.7) ln  Fbkjp   ln  Fkj   ln  kjp  w     D   Dbkjp   Dkj  IMMI  kjp  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    Dbkjp   Dkjp  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 ˆ , we follow OP (2008) and compute the assumed province p. From the fixed effect estimates I kj geographically invariant5 efficiency terms  Fkj and  Dkj by first normalising them to sum to one as follows: ˆ ) exp( I 1 (3.9) ˆFkj  kj , and ˆ  1  exp( Iˆ ) Dkj ˆ ) 1  exp( I kj kj The estimates from (3.9), together with ˆ IMMI from regression (3.8) are used to construct the ˆ as per equation (3.4). In the next section the composite labor input is composite labor input N kjp 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 ˆ )u (3.11) ln( wkjp )  I p  I kp  I kj  ln( N  EXP kjp kjp 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, ˆ ) . While using the (logarithm of) hours worked by immigrants in the k, j cell at time t, ln( Fkjt ) , as instruments for ln( N kjt 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 ˆ constructed in the previous education-experience groups. Note that the composite labor input N kjp 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 ˆ ’s are obtained from the twelve kj ˆ (4 schooling×3 experience) from regression (3.11) using the education×experience fixed effects I kj following normalisation: exp( I ˆ ) (3.12) ˆkj  kj  j 1 exp( Iˆkj ) 3  EXP , together with the  The estimated value of ˆ ’s are then used to construct a further kj level of CES composite labor input N ˆ using equation (3.3) for each schooling group kp 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    1 N  (3.14) ln  UPp   ln  UPp   ln  UPp  w     N   LPp   LPp  LL  LPp  A regression version of equation (3.14) is used to estimate the inverse elasticity parameter 1/  LL : w  1 Nˆ  (3.15a) ln  UPp   I  ln  UPp   uLp w  Lp   ˆ   LPp  LL  N LPp  where the province fixed effects ILp control for the geographical variation in ln UPp /  LPp  for this ˆ are obtained from the ˆ and N broad education category L, and the composite labor inputs NUPp LPp 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  ˆ   HSp   N 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. ˆ are used As before, the estimated region fixed effects for each broad education category I bp to compute the standardised relative productivity terms ˆ ’s and ˆ ’s for the relevant (l,m)-pair of lp mp detailed schooling groups in the broad education category b.8 Specifically: ˆ ) exp( I 1 (3.16) ˆlp  bp , and ˆ  1  exp( Iˆ ) mp ˆ ) 1  exp( I bp bp The estimates obtained from (3.16), together with the ˆ bb estimated for each broad ˆ and N schooling category b {L, H } are then used to construct the composite labor inputs N ˆ Lp 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    1 N  (3.18) ln  Hp   ln  Hp   ln  Hp  w     N   Lp   Lp  HL  Lp  Equation (3.18) serves as a basis for the following regression that is used to estimate the parameter 1/  HL : w  1 Nˆ  (3.19) ln  Hp  I  ln  Hp   up w  p  Nˆ   Lp  HL  Lp  ˆ for each province are computed in the previous ˆ and N where the composite labor inputs N Lp Hp 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)        wDbkjp ˆ HL 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1 e Lijt   ( Lijt )   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 first lag of the share of foreign workers in technicians & associate - 0.777*** - - - - - professional, QTt 1 first lag of the share of foreign workers in clerical & related - - 0.787*** - - - - occupations, QCt 1 first lag of the share of foreign workers in elementary occupations, - - - 0.633*** - - - QEt 1 first lag of the share of foreign workers in plant & machine operators and assemblers (directly - - - - 0.829*** - - employed), QDt 1 first lag of the share of foreign 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]* share of foreign workers in plant & machine operators and -0.002 - 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 Source: Regression results 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 Independent Variables Eqn. 1 Eqn. 2 -0.272 -0.213 log of capital-labor ratio, ln(K/L) [0.025]*** [0.020]*** 0.118 0.057 log of real value of fixed assets, lnK [0.017]*** [0.012]*** -0.311 -0.199 dummy variable for type of establishment, DT [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 share of foreign workers in clerical & related occupations, QC - [0.009] -0.018 share of foreign workers in elementary occupations, QE - [0.012] share of foreign workers in plant & machine operators and 0.004 - assemblers (directly employed), QD [0.001]*** share of foreign workers in plant & machine operators and -0.010 - assemblers (employed through labor contracts), QP [0.006] 1.364 -0.580 concerntration ratio, CR [0.256]*** [0.215]*** -0.069 -0.717 dummy variable for date of establishment, DY [0.081] [0.367]* 0.266 0.156 dummy variable for competitiveness, DC [0.034]*** [0.017]*** 0.717 0.754 AR(1) [0.006]*** [0.005]*** -0.0019 overall share of foreign workers, FL - [0.0007]*** -0.345 0.869 Constant [0.149]** [0.140]*** Dummy variables for industry, DI YES YES Observations 14046 14046 R-squared 0.5563 0.7091 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 Source: Regression results 150 Table 8: IV Estimation Results of the Total Wages and Salaries for All Workers (both Native and Foreign) Model Independent Variables Eqn. 1 Eqn. 2 -1.177 -1.056 log of capital-labor ratio, ln(K/L) [0.029]*** [0.018]*** 1.236 1.139 log of real value of fixed assets, lnK [0.021]*** [0.011]*** -0.203 0.045 dummy variable for type of establishment, DT [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 share of foreign workers in clerical & related occupations, QC - [0.216] 0.001 share of foreign workers in elementary occupations, QE - [0.006] share of foreign workers in plant & machine operators and -0.003 - assemblers (directly employed), QD [0.002]* share of foreign workers in plant & machine operators and -0.014 - assemblers (employed through labor contracts), QP [0.006]** 0.535 -0.375 concerntration ratio, CR [0.203]*** [0.162]** -0.907 -0.437 dummy variable for date of establishment, DY [0.216]*** [0.291] 0.106 0.036 dummy variable for competitiveness, DC [0.055]* [0.013]*** 0.744 0.792 AR(1) [0.005]*** [0.005]*** -0.0075 overall share of foreign workers, FL - [0.0006]*** 7.775 8.321 Constant [0.135]*** [0.091]*** Dummy variables for industry, DI YES YES Observations 14046 14046 R-squared 0.9706 0.9905 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 Source: Regression results 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 Local establishments Foreign establishments All establishments Year F NF Overall F NF Overall F NF Overall 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 Note: F – employ foreign labor, NF – did not employ foreign labor Source: DOS Table 11: Median of Real Total Wages and Salaries (for both Native and Foreign Workers) per Worker (RM), 2000-2006 Local establishments Foreign establishments All establishments Year F NF Overall F NF Overall F NF Overall 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 Note: F – employ foreign labor, NF – did not employ foreign labor Source: DOS 153 Table 12: Median of Unit Labor Cost, 2000-2006 Local establishments Foreign establishments All establishments Year F NF Overall F NF Overall F NF Overall 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 Note: F – employ foreign labor, NF – did not employ foreign labor Source: DOS Table 13: Median of Real Profit and Loss (RM thousand), 2000-2006 Local establishments Foreign establishments All establishments Year F NF Overall F NF Overall F NF Overall 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 Note: F – employ foreign labor, NF – did not employ foreign labor 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: Cambridge University Press. 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 Source: Ministry of Manpower 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 Source: Ministry of Manpower 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 Source: Philippine Overseas Employment Administration (POEA) 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. Successfully finished the 4-year nursing program; Failed the NLE once; Successfully passed NCLEX 2 14.44 15.44 and/or CGFNS; Successfully acquired 2-year hospital experience; Successfully employed abroad. Successfully finished the 4-year nursing program; Failed the NLE twice; Successfully passed NCLEX 3 12.82 14.10 and/or CGFNS; Successfully acquired 2-year hospital experience; Successfully employed abroad. Successfully finished the 4-year nursing program; Failed the NLE thrice; Successfully passed NCLEX 4 11.41 12.98 and/or CGFNS; Successfully acquired 2-year hospital experience; Successfully employed abroad. Successfully finished the 4-year nursing program; 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 Source: Philippine Overseas Employment Administration (POEA) 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 Source: Philippine Overseas Employment Administration (POEA) 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 Source: Department of Overseas Labor (DOLAB), Government of Vietnam 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 Source: Department of Overseas Labor (DOLAB), Government of Vietnam 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 Source: Department of Overseas Labor (DOLAB), Government of Vietnam 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 6 4 2 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 *Skilled/Unskilled 280 10 the total workforce Above 40% to 50% of *Skilled/Unskilled 450 15 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 Tier 2: above 25% to 40% of total Skilled 300 workforce Unskilled(1) 300 236 Sector Dependency Ceiling (DC) Worker Category Monthly Levy Rate ($) Tier 3: above 40% to 50% of total Skilled 450 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 $310 only applies to Non-Traditional Sources (NTS) or People’s Republic of China (PRC) workers in the Construction and Process industries, who have been exempted from the requirement of Man-Year Entitlements (MYEs). To be exempted from MYE, the foreign workers must have at least two years’ working experience in his respective industry and his employer has applied for a Work Permit without any Prior Approval support. S Pass Levies Sector Dependency Ceiling (DC) Worker Category Monthly Levy Rate ($) All sectors Basic Tier / Tier 1: Up to 20% of total Skilled 100 workforce(1) Tier 2: above 20% to 25% of total Skilled 120 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  ...  (1  r ) N  (1  r )    T0  1  ...  (1  r )     s   ph (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 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. 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 and domestic working sectors 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 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. 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 and domestic working sectors 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 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. *** 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