WPS7057 Policy Research Working Paper 7057 Evidence on Policies to Increase the Development Impacts of International Migration David McKenzie Dean Yang Development Research Group Finance and Private Sector Development Team October 2014 Policy Research Working Paper 7057 Abstract International migration offers individuals and their families agreements for countries whose workers have few other the potential to experience immediate and large gains in migration options, developing new savings and remittance their incomes, and offers a large number of other posi- products that allow migrants more control over how their tive benefits to the sending communities and countries. money is used, and some efforts to provide financial edu- However, there are also concerns about potential costs cation to migrants and their families. Suggestive evidence of migration, including concerns about trafficking and together with theory offers support for a number of other human rights, a desire for remittances to be used more policies, such as lowering the cost of remittances, reducing effectively, and concerns about externalities from skilled passport costs, offering dual citizenship, and removing exit workers being lost. As a result there is increasing interest barriers to migration. Research offers reasons to be cautious in policies which can enhance the development benefits of about some policies, such as enforcing strong rights for international migration and mitigate these potential costs. migrants like high minimum wages. Nevertheless, the paper This paper provides a critical review of recent research on finds the evidence base to be weak for many policies, with the effectiveness of these policies at three stages of the no reliable research on the impact of most return migra- migration process: pre-departure, during migration, and tion programs, nor for whether countries should be trying directed toward possible return. The existing evidence base to induce communal remitting through matching funds. suggests some areas of policy success: bilateral migration This paper is a product of the Finance and Private Sector Development Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at dmckenzie@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Evidence on Policies to Increase the Development Impacts of International Migration # David McKenzie, World Bank Dean Yang, University of Michigan Keywords: Migration policy, Remittances, Return Migration, Impact Evaluation JEL codes: O15, F22 # We gratefully acknowledge funding for this paper from the World Bank’s Global Knowledge Partnership on Migration and Development (KNOMAD) program. We appreciate helpful feedback from the IOM and an anonymous external reviewer. All views expressed in this paper are those of the authors and need not represent those of the World Bank or the University of Michigan. 1. Introduction International migration offers individuals the potential to experience immediate and large gains in their income, with these gains far exceeding those from any other development policy intervention (Clemens et al, 2009; McKenzie et al, 2010). These gains are typically shared with household members accompanying them, and through remittances, with family members (and others) remaining in their home country. Both the number of migrants and the remittances they have sent have grown rapidly over the past two decades. The number of people living outside their country of birth totaled 232 million in 2013, up from 154 million in 1990, 1 while recorded remittances sent to developing countries were US$321 billion in 2010, compared to only $49 billion in 1990. 2 In addition to the direct effects of higher incomes, research has found migration to provide a number of other positive benefits to the sending country, including facilitation of trade, technology transfer and foreign direct investment, providing incentives for greater investment in globally marketable skills, and transmission of democratic norms. However, these positive impacts of migration are also accompanied by concerns over a number of potential costs. These include concerns about human trafficking and abuse of migrant rights; concerns about the abilities of households receiving large, temporary flows of remittances to save appropriately; concerns that the benefits of migration do not spread far beyond the immediate household; and concerns that developing countries are losing the positive externalities of high skilled workers. There is increasing interest among development institutions and developing country governments in identifying policies that can enhance the development impacts of international migration, by facilitating more of the benefits and mitigating some of these potential costs. This has led to the deployment of a wide range of migration policies that occur throughout the different stages of the migration process (pre-departure, while the migrant is abroad, and upon possible return), and which cover a wide range of different areas (financial sector, social protection, employment policies, etc.). 1 http://esa.un.org/unmigration/documents/The_number_of_international_migrants.pdf [accessed May 5, 2014] 2 Amounts are expressed in constant 2011 US dollars, and sourced from the World Development Indicators, as reported in Clemens and McKenzie (2014). Clemens and McKenzie (2014) note that much of this increase in remittances likely reflects changes in measurement, rather than genuine growth. 2 However, this substantial policy interest and growing policy efforts stand in stark contrast to the limited empirical evidence that can help guide policy. The vast majority of existing research on migration and development is focused on measuring and understanding the determinants of migration, and the impacts of migration and remittances on development, rather than on the effects of policies designed to change these impacts. 3 This has begun to change in the last few years, with a number of innovative studies explicitly testing the impacts of migration policies relative to a counterfactual of some alternative policy that could be implemented. In this paper we review this new literature and assess the state of evidence on the impacts of different policies intended to improve the development impacts of international migration. These development impacts can include impacts for the migrants themselves, their families, and/or the countries from which they migrate. This means we will not consider policies whose primary goal is the well-being of firms and workers in migrant-receiving developed countries. As a result, we exclude many areas of immigration policy in developed countries. We focus on international migration, and only discuss work on internal migration policies to the extent they are useful contrasts or are informative about international migration policy. Rather than just summarizing the results in the literature, we will also offer our assessment of the rigor and reliability of these results, in order to also draw out areas where there is considerable policy effort without reliable evidence as to its effects. We begin with a discussion of policies that occur at the pre-migration stage. These include efforts to facilitate more migration, and pre-departure training sessions to provide migrants with more information and skills. We then turn to policies directed toward migrants and their families while migrants are abroad. These include rights policies, efforts to expand financial access, policies to make it easier and cheaper for migrants to remit, policies to channel remittances toward purposes with broader development impacts, policies to encourage communal remittances and the development of home-town associations, and integration policies. Finally we examine policies directed toward returning migrants. These include efforts to remove regulatory, bureaucratic and informational barriers inhibiting return migration, policies intended to change the incentives to return, and policies designed to make return migrants more productive and 3 See for example the recent handbook chapter by Hanson (2010), and also Yang (2011) for a review of recent research on the economics of migrants’ remittances. This is true even of much of the recent experimental and quasi- experimental literature reviewed in McKenzie and Yang (2012). 3 reintegration easier. We find the evidence base to be weak for many of these policies, especially those directed toward return migration, and discuss possible reasons for this lack of research before concluding. 2. Policies pre-migration Much of international migration takes place without explicit policy actions by developing country governments. Destination countries set their immigration policies, and individuals who are eligible to move under these policies decide whether or not to do so, while others migrate illegally. Yet survey results show that many more people around the world express a desire to move to another country than currently do so (Esipova et al, 2011), while others remain misinformed of the potential gains from moving (McKenzie et al, 2013). Bureaucratic, financial, informational, and other constraints prevent many individuals who would benefit from moving from doing so. As a result, one key avenue for migration policy to enhance development impacts is through facilitating more people to migrate, and ensuring they are well-informed when doing so. The main policy levers used to do this are through explicit policy efforts to facilitate migration, and through pre-departure orientations or training programs designed to ensure migrants are well-informed about conditions abroad. Table 1 summarizes these types of policies and the evidence for each, which we discuss in detail below. 2.1 Policy efforts to facilitate (or hinder) migration Beam et al. (2013) note that there are two categories of policy actions that developing country governments have taken to facilitate international migration. The first consists of bilateral actions, which involve cooperation with governments or employers in destination countries, leading to formal agreements to allow labor migration of specified numbers and types of workers. A leading example of this has been the Philippines, which has signed at least 49 bilateral migration agreements with 25 destination countries. Large migration flows have occurred through some of these migration agreements, and given the importance of legal barriers in regulating migration, it seems likely that such agreements have allowed more migration from the signatory countries. However, we are unaware of research which has identified how much of 4 this migration would have happened in the absence of such signed agreements, and hence lack evidence on the causal effect of such policies. One exception to this comes from bilateral migration policies negotiated between New Zealand and several Pacific Island nations (including Samoa, Tonga, and Vanuatu) to allow seasonal migration to New Zealand through the Recognized Seasonal Employer (RSE) program. Gibson and McKenzie (2014) examine the impacts of this program on households in Tonga and Vanuatu. They find that individuals participating in the program lack almost any other opportunity to migrate internationally, so that the program generates new migration, and that this migration results in large increases in income and consumption for the participating households. The use a matched difference-in-difference strategy to identify these program impacts, arguing that the newness of the program and limited availability of places in the program enables comparable households to be found to those who were selected to participate in the program. The second set of actions countries can take are unilateral actions, which do not require the cooperation of other governments. These are most evident when used to hinder migration. 4 For example, a number of countries restrict the rights of women to migrate (e.g. Gabon, Libya, Saudi Arabia), and others require all citizens to get government permission to travel abroad (e.g. Cuba, the Islamic Republic of Iran, the Democratic People’s Republic of Korea). McKenzie (2007) shows in a linear regression that countries with such restrictions have 5 to 6 percent less migrants per capita than countries with similar income, population, and governance levels which do not have these restrictions. Countries also can affect the ease of migration by imposing high passport fees and cumbersome procedures for obtaining a passport. McKenzie (2007) shows the cost of a passport exceeds 10 percent of per capita income in some countries, and that higher passport costs are associated with less migration. This association continues to hold after controlling for population, per capita income, and government effectiveness. Governments can also take more positive unilateral actions to facilitate migration. Beam et al. (2013) conducted a large-scale randomized experiment in the Philippine province of Sorsogon testing the impact of unilaterally facilitating international labor migration. Prime-age adults from randomly-sampled households across the province were surveyed, and then randomly assigned to 4 Developing countries which are destinations for immigrants also pursue a number of policies to restrict immigration, which we do not discuss here. 5 either a control group or to one of many treatment groups. Households assigned to treatment groups received one or more of several possible interventions intended to facilitate international migration. A subset of treated households received the full set of interventions (the “most intensive” treatment). The interventions targeted the following possible reasons people might not migrate internationally: 1) information barriers (information about job search, migrating abroad, financing migration, and passport processing); 2) frictions in job search (assistance in enrolling in an online job-finding website set up by the project to lower search costs and facilitate matching between recruiters and workers); and 3) documentation barriers (assistance and a full subsidy for passport application). Overall, the authors find no evidence that any of the individual interventions or combinations of interventions led to increases in international migration. The most intensive treatment led to increases in job search and doubled the rate of international job offers, but had no identifiable effect on international labor migration. Even the highest overseas job-search rate induced (22%) fell far short of the share initially expressing interest in migrating (34%). The authors conclude that unilateral migration facilitation will at most induce a trickle, not a flood, of additional emigration. A related study was conducted by Beam (2013), also in Sorsogon province, Philippines. Within a sample of prime-age adults, the study randomly assigned participants to a control group or one of a set of treatment groups. The treatments involved offering information on wages and qualifications for typical overseas jobs, or provision of an incentive (a restaurant voucher) conditional on attending a “job fair” (an event at which job-seekers connect with placement agencies for overseas jobs). Study participants in the treatment groups received the information treatment, the job fair incentive treatment, or both. The study found that the information treatments did improve knowledge about overseas wages and job qualifications. Neither information treatments nor the job fair incentive treatment lead to increases in search activity for overseas jobs. However, the job fair incentive treatment had unexpected impacts: it led to increased search for jobs in the national capital, Manila, and also to higher rates of formal employment 10 months after the job fair. Along with the previous study, this work suggests limits to the ability of unilateral facilitation policies to foster more migration. 6 2.2. Pre-departure orientation or training programs A number of programs are designed to provide potential migrants with some type of orientation or training prior to departure. These include programs aimed at easing the transition to life abroad, programs providing technical skills to enhance their productivity abroad, programs designed to prevent human trafficking and abuse, and programs designed to teach financial literacy. (We discuss pre-departure financial literacy programs separately in the next sub- section.) Pre-departure orientations have been implemented in a variety of contexts. Perhaps the most prominent is the Philippine government’s Pre-Departure Orientation Seminar (PDOS), a one-day session which has been required since 1983 for labor migrants departing for new jobs overseas and is administered by the government as well as by non-government organizations (NGOs) under contract with the government. The stated aims are to prevent abuse of migrant workers, protect their labor rights, and educate workers on laws, culture, and customs of destination countries. Additional topics covered include health and safety, financial literacy, and travel procedures and tips. Starting in 2009, the PDOS was expanded to include 4-6 days of training for migrant household service workers on job skills, first aid, language, culture, and stress management (known as the Comprehensive Pre-Departure Education Program, or CPDEP) (International Labor Organization 2013). Other important examples of pre-departure training programs are the wide range of programs run by the International Organization for Migration (IOM). The IOM has been conducting training for migrants since the 1950s, with 352,000 migrants involved in their training programs over the decade 2001-2010. Of this number, 86 percent were resettlement migrants (refugees and humanitarian entrants), and the remainder were labor migrants, asylum seekers, marriage migrants, immigrant visa applicants, and family members of trafficked persons (International Organization for Migration 2011). Impact evaluation of pre-departure orientation training has been minimal thus far. A large number of evaluation reports have been conducted, but to our knowledge these are primarily “process” evaluations that involve audits of activities via examination of internal documents, site visits, and in-depth interviews with key internal and external stakeholders. An example of a 7 process evaluation of this sort is Norad (2010), which focuses on the IOM’s anti-trafficking programs. Another example is Asis and Aguinas (2012), which provides a high-level assessment of shortcomings and potential improvements in the pre-departure programs of the Philippines, Indonesia, and Nepal. Of the more limited number of impact evaluations, none to our knowledge use credible research designs that can allow identification of causal effects of the programs on migrant welfare or other outcomes. For example, IOM (2006) seeks to assess the impacts of an anti-trafficking informational program in Cambodia on the knowledge of anti-trafficking messaging among individuals remaining in Cambodia (non-migrants). It uses survey data on program participants and compares responses to surveys of non-participants, finding that participants have better knowledge about information relevant for anti-trafficking. The study does not describe how study participants and control group respondents were selected, so it is not possible to assess whether the two populations can be credibly compared so as to establish the causal impact of the program. The study also does not estimate impacts on migrants themselves, but only on knowledge of non-migrants. Given the popularity of these programs, it is important going forward to conduct prospective, randomized evaluations of pre-departure trainings, or to find opportunities to use natural experimental designs so that causal effects can be established using observational data. It will also be important to measure impacts on the migrants themselves (rather than, for example, simply measuring information of non-migrants remaining behind). One ongoing study (Omar Mahmoud et al 2013) is doing precisely that, by randomly assigning departing labor migrants from the Philippines to different types of modified pre-departure orientation seminars, and tracking impacts on migrants over time. 2.3 Financial literacy programs for migrants and their families (whether before or after departure) Owing to the large wage gains possible when individuals migrate from developing to developed countries, and the large amounts sent home by migrants in the form of remittances, migrant- origin households are very often faced with managing amounts of money substantially larger than the household budgets of those without migrant members. This has raised concerns among policy-makers and non-government organizations that financial decision-making in migrant households may be suboptimal, particularly in households whose members migrated relatively 8 recently. Motivated by such concerns, a common type of intervention provides financial literacy or financial education training to migrant workers and their families remaining behind in the home country. Financial literacy programs are commonly conducted as part of pre-departure orientation training programs (discussed in the previous section), such as the Pre-Departure Orientation Seminar (PDOS) mandated by the Philippine government for labor migrants, or the migration training programs run by the International Organization for Migration. Important evidence on the impact of providing financial education to migrant workers and their families is provided by Doi et al. (2014), who implemented a randomized controlled trial among Indonesian women about to depart for overseas work as household servants (maids), and these women’s families. Study participants were randomly assigned to either a control group (that received no training) or to one of three treatment groups. In the three treatment groups, financial literacy training was provided prior to the migrant’s departure for overseas, for either: 1) the migrant alone, 2) a left-behind family member alone, or 3) both the migrant and the family member. The training was conducted over sessions lasting either 1 day (for the migrants) or 2 days (for family members), and covered financial planning and management, savings, debt management, sending and receiving remittances, and migrant insurance. The study examined impacts on families remaining behind in Indonesia, finding that each type of treatment did increase financial knowledge (for example, on knowledge of financial terminology). In addition, training both the migrant and family led to increases in savings in the origin household, but the other two treatments (migrant only and family only) did not have a similar savings impact. None of the treatments had substantial impacts on remittances received. The study highlights the complementarities from training both migrants and family members. Financial literacy training also occurs frequently at destination, targeting the migrant. Seshan and Yang (2014) examine such a program for married male migrant workers in Doha, Qatar who were from Kerala, India and whose wives remained behind in India. Participants were randomly assigned to either a control group or a treatment group. The treatment group was invited to attend a one-time motivational session on personal financial management that stressed the importance of savings and of making joint financial decisions with spouses remaining behind in India. The session lasted 3 hours and was led by a popular motivational speaker. Impacts were measured via follow-up surveys of both migrants in Kerala and their wives in India. The treatment led both 9 migrants and their wives to be more likely to report that they made joint financial decisions with their spouses, and wives also became more likely to seek out additional financial education themselves. The study also found evidence of heterogeneity in treatment effects on financial outcomes. Among migrants with lower savings at baseline (prior to treatment) the treatment led to higher total (Qatar plus India) household savings and higher remittances sent by migrants to wives. Gibson et al. (2014) implemented a randomized controlled trial of a financial literacy intervention among migrants in Australia and New Zealand. The study coincided with the introduction of a new online tool for comparing remittance transaction fees across providers, and the introduction of a new remittance method in New Zealand (provision to migrants of a second ATM card which could be sent to family members back home to allow withdrawals from the migrant’s New Zealand bank account). The study sample consisted of migrants from Tonga, East Asia, and Sri Lanka. Study participants were randomly assigned either to a control group that received no training and a treatment group that was invited to a financial education session that focused on helping migrants compare among different remittance-sending methods. The authors find that the treatment did lead to increases in financial knowledge and in seeking information about remittance methods for the Tongan and East Asian participants. (There were no effects on knowledge or behavior in the Sri Lankan sample, which the authors ascribe to that sample already having high education and high experience with remittances.) The authors also find no effects of the treatment on remittance frequency or total amounts of remittances sent. This absence of a treatment effect on remittance outcomes is attributed by the authors to be in part due to barriers to the use of alternative remittance methods for remittance receivers. While additional studies of the impact of financial education in migrant populations would be useful, these studies suggest the diversity of possible impacts of different types of interventions in different populations. A particular financial education intervention (such as the one conducted among Kerala migrants in Qatar) could have different impacts in different subpopulations, suggesting that differentiated interventions targeted at the particular needs of subpopulations would be worth exploring. Both the Seshan and Yang (2014) and Gibson et al (2014) studies examined the impact of post-departure training of migrants alone, while the Doi et al (2014) study examined pre-departure training of migrants and families. Future studies could therefore 10 examine the impact of financial literacy interventions among family members after the migrants have departed for overseas. 3. Policies while migrants are abroad The widest range of policies designed to enhance development impacts of migration occur once migration has taken place. Table 2 summarizes the range of policy instruments and the evidence for their effectiveness. A first set of policies involves ensuring adequate rights for migrants, with debate as to which rights should be protected. A second broad area of policy concerns financial access and remittances, with policies to ensure migrants and their families have access to appropriate financial products, efforts to lower the cost of sending remittances, and policies to get migrants to contribute more toward activities with broader development outcomes. A related area is to get migrants to form home-town associations to build communal public goods. Finally, policies directed toward migrants who intend on staying abroad can affect their ability to integrate successfully into their new countries. 3.1 Rights policies for migrants The United Nations International Convention on the Protection of the Rights of All Migrant Workers and Members of Their Families (CMW) of 1990 stipulates a comprehensive set of civil, political, economic, and social rights for migrants. It is based on the principal of equal treatment of migrants and nationals, including rights not to have identity documents confiscated, rights to equal treatment with nationals before the courts, the right to form associations and trade unions, and the right to equal treatment with regard to remuneration. However, this convention is the least ratified of all the major international human rights treaties, and by 2012 no major migrant receiving country had ratified it (Ruhs, 2013). The question which then arises is whether it would be good for development for more destination countries to grant all these rights? In almost all cases we would expect migrants experiencing more rights to be directly better off as a result. However, a number of these rights impose costs on the host country government, or host country employers. As a consequence, we should expect to see greater rights for migrants resulting in either lower wages, or in fewer migrants being hired. Ruhs (2013) documents this pattern across countries, finding that countries which provide migrants with more rights in terms of access to retirement benefits, unemployment benefits, and access to public education have less 11 open migration programs, providing fewer opportunities for individuals to migrate. McKenzie et al. (2013) look in more detail at a particular type of economic right: the right to a minimum wage or to a wage equal to that prevailing for natives. They find evidence from Filipino migrant worker contracts consistent with the idea that binding minimum wages restrict the number of individuals who can migrate, and make migrant jobs more vulnerable to economic shocks at destination than would be the case if wages could adjust. They then carry out a difference-in- differences analysis of the effect of a minimum wage change for Filipina domestic workers. They find that doubling the minimum wage that these workers must be paid did result in higher wages for the workers who did migrate, but also resulted in a 55 to 57 percent reduction in the number of workers going to the destinations where this new law was binding. An example of a law change granting more rights to migrant workers comes from Naidu et al. (2014), who examine a reform in the United Arab Emirates which gave migrant workers the right to change employers after their contract ends without having to receive a letter of no objection from their previous employer. Using variation in the end date of worker contracts around the reform, they find that workers whose contract ended post-reform experienced a 10 percent increase in real earnings relative to workers whose contracts ended before the reform. The right to be able to change employers and move from one job to another clearly conveys benefits on migrant workers. However, the authors were not able to examine whether employers adjusted to this reform by changing their recruiting or wage-setting for new workers- it is possible that employers might be less willing to hire new workers, or will offer them lower initial wages, if there is a greater possibility of them leaving for a different employer. Existing evidence therefore suggests that there is a trade-off between the rights of migrants to be paid equally to native workers in the destination countries, and the opportunities for poor people from developing countries to dramatically increase their incomes through migration: requiring higher minimum wages does limit the number of people who get to migrate. In contrast, a different set of rights are more basic ones that involve freedom from abuse and exploitation, and freedom to leave. This involves migrants being able to retain their passports and depart the country at any time they like, being paid the amount they are contracted to be paid, being protected from being beaten or sexually abused by their employer, and being able to report any abuse that does occur and see employers who violate these rights face the consequences of doing 12 so. Failure to provide migrants with these rights directly reduces the development impact of their migration (lowering the incomes migrants earn and their well-being) and also makes it difficult to use revealed preference as an argument for positive migration benefits since violations of these rights typically are violations of the conditions assumed by migrants in making their migration decisions. Moreover, while enforcement of these rights will slightly increase the cost of hiring a worker for employers, this works in the right way for development impacts – it should be more costly for more abusive employers, leading lower quality employers who provide little in the way of development benefits to exit the market, while having minimal impact on other employers. While we are unaware of studies which examine the impact of reforms in these more basic rights, there seems little economic rationale not to provide these types of rights to workers. 3.2 Financial access for migrants and their families Recent decades have seen a dramatic increase in development efforts involving provision of financial services to the poor. Initiated by the micro-lending revolution, such efforts have expanded to include a range of financial services beyond credit, including savings, insurance, and inter-household transfer mechanisms or remittances. Given this background, it is therefore natural that interventions providing access to financial services have also been explored in the context of transnational households (international migrants and their families remaining behind). The motivation behind provision of financial services for transnational household members is similar to motivations behind providing such services more generally: they give households essential tools that can help them achieve their financial and life objectives. Credit and savings provide lump-sums that can be invested in activities (e.g., education, health, small enterprises) that yield longer-term returns, and provide tools that help households cope with unexpected shocks. Transfer or remittance services allow households to give and receive assistance from distant members of their social network. Transnational households do of course have special characteristics that have implications for the emphases and design of financial access interventions. Reliable remittance services in particular are crucial for allowing migrants to support family members left behind in the origin country. The fact that high proportions of transnational households already use remittance services means that it can be practically feasible for financial access interventions to be linked in some way to 13 remittance services. An example is offers of savings accounts into which migrants can remit in the home country. In very practical terms, offers of other financial services can occur when migrants are visiting a branch location of a financial institution to make a remittance transaction. Private financial institutions are making great efforts to increase financial access for migrants and their families back home. There has been a dramatic expansion of money transmission services in the past few decades: market leaders Western Union and MoneyGram have expanded substantially, and there has been substantial entry into the international money transfer market of other standalone money transfer organizations as well as banks, credit unions, and other types of financial institutions.5 International organizations such as the Inter-American Development Bank (in particular its Multilateral Investment Fund, known by its Spanish acronym FOMIN) and the World Bank have been actively involved, via funding and technical assistance, in facilitating the development of financial services for transnational households, in both host and origin countries of migrants. An often-heard policy objective is to seek to raise savings levels in transnational households, motivated by the aggregate benefits of higher savings levels as well as benefits at the household level (in terms of asset accumulation and self-insurance via buffer stocks in face of shocks). For example, Hall (2010) provides an overview of a decade’s worth of projects funded by FOMIN to expand financial access for remittances and related financial services. One example (out of dozens of projects profiled) is an initiative with the Asociación Mexicana de Uniones de Crédito del Sector Social [Mexican Association of Social Sector Credit Unions] (AMUCSS), to help AMUCSS strengthen provision of remittances, savings, credit, and other financial services in rural Mexico. The project included assistance with marketing remittance services to migrants in the US, financial literacy training for migrants and left-behind family members, design of new financial products, and cross-sales of other financial products to remittance clients. Two recent randomized field experiments have provided evidence among migrants in the US of the effectiveness of different approaches to promoting savings in transnational households. A randomized controlled trial among migrants from El Salvador by Ashraf et al (forthcoming) tested ways to stimulate savings in El Salvador. The study was particularly interested in whether 5 For overviews, see Orozco (2004), Orozco et al (2010), and DeParle (2007), among others. 14 migrants sought to control or influence how much was saved by family members in the home country, so the treatments offered varied in the degree to which migrants could monitor and control family members’ savings in account at a bank in El Salvador that partnered with the researchers on the research project. Salvadoran migrants in Washington, DC, were randomly assigned to either a comparison group (that experienced no savings intervention) or to one of three treatment groups. In Treatment 1, migrants were offered the opportunity to open a new account at the partner bank in the name of someone in El Salvador, into which the migrant could remit funds. This account allowed the migrant to deposit but not to withdraw, nor to observe withdrawals. Treatment 2 offered the migrant the opportunity to open an account at the partner bank to be held jointly by the migrant and someone in El Salvador. This new joint account allowed joint observability of account balances as well as joint withdrawals (both the migrant and the El Salvador person were given an ATM card for the account). Finally, in Treatment 3 migrants were offered, in addition to the accounts offered in Treatments 1 and 2, the option to open an account in the migrant’s name only. Treatment 3 therefore offered migrants the greatest degree of monitoring and control over savings in El Salvador. Treatment 2 offered some control (the joint account only), but less than Treatment 3. Treatment 1 offered essentially no direct control by the migrant, since it only offered migrants the option of remitting into accounts owned by another person, which they could not directly monitor. The results of the study provide evidence that migrants do value and take advantage of opportunities to exert control over savings in their home country. Migrants were much more likely to open savings accounts at the partner bank in El Salvador, and accumulated more savings at the partner bank, if they were assigned to the treatment condition offering the greatest degree of monitoring and control (Treatment 3). Migrants desired savings that are jointly held with family members, as well as savings only for themselves: there were substantial increases in savings in both the joint accounts shared between migrants and someone in El Salvador (offered in Treatments 2 and 3), and in the accounts for migrants alone (offered only in Treatment 3). The increase in savings in the new accounts offered was likely to be a true increase in savings, since there was no evidence that the saved funds were simply shifted over from other types of savings 15 (either from other accounts at the partner bank, or from other types of savings reported in a follow-up survey). Strikingly, the impact of Treatment 1 (where accounts were offered only in the name of someone in El Salvador) on savings was much smaller in magnitude and not statistically significantly different from zero. This result is also important, as it reveals that the frequently-made policy recommendation to foster savings in migrants’ home countries by encouraging migrants to remit directly into savings accounts of remittance recipients would be much less effective, compared to interventions that also improved and encouraged migrant monitoring and control over home- country savings. A related experiment was conducted among Mexican migrants in Texas by Chin et al (2011). That study was interested in understanding the impact of facilitating migrant access to savings accounts in the host country (in this case, the U.S.), rather than in the origin country. The study drew a sample of Mexican migrants and randomized them into either a control group (receiving no intervention) or a treatment group that was given assistance in obtaining a matricula consular identity card from the Mexican Consulate, whose primary benefit was that it could be used as identification when opening a bank account in the U.S. The study found that migrants in the treatment group were more likely to open U.S. savings accounts, accumulated more savings in the U.S., and remitted less to Mexico. The study also found heterogeneity in treatment effects: for migrants who at baseline (prior to treatment) reported that they had no control over how their remittances were used by recipients in Mexico, the treatment had larger positive impacts on U.S. bank account take-up and on savings, and also led migrants to shift more savings from Mexico to the U.S. These findings are consistent with those of Ashraf et al (forthcoming), also establishing that migrants have preferences over how the remittances they send home are used. In the Chin et al (2011) case, migrants respond to increased access to savings accounts in the U.S. by saving more in the U.S. instead of in Mexico (presumably because savings in Mexico must be done through intermediaries). In sum, existing studies have documented the positive impacts of interventions facilitating savings access for transnational households, with account offers in both the host and origin countries. There is also consistent evidence that migrants have preferences over how remittance recipients in the home country use remittances, in particular how much of remittances are saved. 16 A policy lesson is that if an objective is to raise savings levels in the origin country, interventions among transnational households that seek to expand access to origin-country formal savings facilities should give priority to savings services that offer migrants some ability to monitor or control savings. If the policy aim includes raising savings by migrants in the host country, then facilitating access to host-country savings facilities can help achieve this objective. Future research on financial access in transnational households would do well to explore the impacts of financial services other than savings, such as credit and insurance, which to our knowledge have not been studied. We turn to discussion of remittance-related financial services in the next sub-section. 3.3 Policies to lower remittance costs and to induce migrants to remit more Motivated by the positive development impacts of remittances (see Yang 2011 for an overview), several efforts are underway to seek to reduce barriers to remittance flows. In 2009, the G8 Heads of State Summit agreed to an objective of reducing the average cost of sending remittances from 10% to 5% in five years (the so-called “5X5 objective”) via policies such as improved information, transparency, and promotion of competition in the money transmission market (G8 2009). For example, FOMIN has supported initiatives to create regulatory and legal frameworks for remittance transfers in Latin America and the Caribbean, with the objective of promoting the development of remittance markets, increasing access and lowering costs for migrant remittance senders and their recipients in the home country (Hall 2010). An important approach that has been attempted in a variety of contexts is to improve migrants’ ability to compare remittance transaction fees. Migrants who are able to access lower-fee money transmitters may send more remittances in response. In addition, improved ability to comparison- shop may increase competition across money transmitters and lead them to lower their prices. One example is Remittance Prices Worldwide, a website (http://remittanceprices.worldbank.org) maintained by the World Bank that provides remittance fee prices across 226 country corridors covering 32 major remittance sending countries and 89 receiving countries. Governments and institutions like the World Bank have also set up remittance-price-comparison website targeted at specific markets, such as Mexico (www.remesamex.gob.mx), the UK (www.sendmoneyhome.org), Italy (http://www.mandasoldiacasa.it), and in Australia and New 17 Zealand (www.sendmoneyhomepacific.org), with the latter being the subject of the Gibson, McKenzie, and Zia (2014) remittances financial literacy experiment. Although remittance costs have fallen following these policy efforts, it is unclear how much of the reduction in costs is due explicitly to the policy actions per se, rather than arising from technological advances and competition that might have lowered costs anyway. Several recent randomized field experiments offer insights on how migrants are likely to react to reductions in remittance costs. The existing evidence suggests that response to fee reductions can be quite large. Aycinena et al. (2010) implemented a randomized field experiment that estimates the causal impact of remittance transaction fees on remittances. In partnership with a money transfer operator, Salvadoran migrants in Washington D.C. were randomly assigned differently-sized discounts on remittance transaction fees. The remittance fee reductions led to higher transaction frequency by remitters: each $1 fee reduction led to an additional 0.11 transactions per month. There was no change in the dollar amount remitted per transaction. These remittance responses to price reductions are large in magnitude: a $1 reduction in the remittance transaction fee leads to average fee savings per month of only $0.47, but the corresponding increase in average remittances sent per month is an order of magnitude larger, $25. Given that the impacts found in the Aycinena et al (2010) were surprisingly large, a related study, Ambler et al. (2014b), aimed to confirm these results and to make some improvements in research design. The Ambler et al (2014b) study also examined the impacts of offering migrants in the Washington, DC metro area discounts on remittance fees, this time among migrants from either Guatemala or El Salvador. Migrants were randomly chosen to be offered $3 discounts on remittance fees (off a base price of $8) for remittances sent through the partner organization. The discounts were valid for 10 weeks, for as many transactions as the respondent chose to carry out. In contrast to the previous study, in this case the discount was limited to remittances sent to one specific remittance recipient (the one the migrant had just remitted to at the time of enrollment in the study). This was done to limit the possibility that migrants would allow others to use their discount to send to other recipients, which would lead to a spurious upward bias in the estimated treatment effect on remittances. 18 Ambler et al (2014b) find that these discounts lead to large increases in the number of transactions and total amount remitted (measured using administrative data from the partner money transfer company). These effects of the discounts persist up to 20 weeks after migrants are no longer eligible to receive them. These are real increases in remittances: follow-up survey responses help rule out that migrants are sending remittances on behalf of others or shifting from other remittance channels. An important question that arises from the findings of both these studies is why migrants send so much more in total remittances in response to relatively small changes in remittance fees. Ambler et al (2014b) suggest that the pattern of results could be generated if remittance recipients in the home country have reference dependent preferences regarding their expected level of remittances, which evolve slowly as remittance levels change, and if migrants are partially naïve about the extent to which recipients’ preferences exhibit such reference dependency. With this characterization of preferences, migrants might respond to the remittance fee discounts by sending more remittance transactions during the discount period, intertemporally substituting for remittances later (post-discount) remittances. But once migrants increase their remittance frequency during the discount period, recipients in the home country raise their reference point for remittance receipts (now expecting higher total amounts per time period). Migrants, however, do not fully anticipate this change in recipients’ reference points. Once the discount ends, migrants find that recipients expect them to maintain the level of remittances (in total dollars per time period) that they experienced during the discount period. Migrants therefore do not immediately return to their previous level of remittances, but only do so in a gradual fashion. This can explain the persistence of higher remittance levels for some time after the end of the discount. Other non-experimental work comes to similar conclusions that reductions in remittance fees would raise remittances. Gibson et al. (2006) provide evidence on remittance responses to fee reductions from survey hypotheticals. Tongan migrants in New Zealand who they surveyed say that they would send substantially higher remittances in response to reductions in the fixed cost component of the remittance fee. The estimated elasticity of remittances to changes in the fixed cost component of the remittance fee is -0.22. Freund and Spatafora (2006) use cross-country 19 data to show that remittance fees are negatively correlated with total remittance flows at the country level. Additional research has found that alleviating information asymmetries within transnational households (between migrants and families back home) could also raise remittances sent by migrants. Improved information on the conditions of families back home could lead migrants to send more. In addition, if families know more about the conditions of migrants overseas, and in particular if migrants have incentives to understate their economic situation so as to dampen requests for assistance, improving families’ information on migrants’ situations could lead to higher remittances. Ambler (2013) implements a lab-in-the-field experiment aimed at revealing the consequences of asymmetric information within transnational households. Study participants are migrants from El Salvador who are living and working in the US (in Washington DC), and a close relative in El Salvador. A key question in the study is whether migrants send more remittances home out of their earnings when potential remittance recipients in the home country have better information about migrant earnings. Specifically, migrants are asked to specify how much of a $600 windfall they would like to share with their El Salvador relative, and are randomly assigned to a treatment where the relative is told that the migrant got this windfall and had to decide how much to share (otherwise, the relative is told nothing). Strikingly, migrants do share significantly more with relatives when relatives are informed about this windfall. This result suggests (extrapolating outside the experimental context) that migrants may share more with their families back home when families have better information on migrants’ income realizations. A complementary set of findings comes from Batista and Narciso (2013), who implemented a randomized study among 1,500 immigrants in Ireland to estimate the impact of facilitating communication between migrants and their contacts abroad on the extent and value of remittance flows. In the experiment, migrants were randomly assigned international telephone calling credit to their home country for a given time period. The authors find a sizable, positive impact of the intervention on the value of migrant remittances sent to individuals who have been receiving remittances from the migrant already, but no impact on other individuals. 20 On the policy front, the evidence suggests that reforms that reduce migrant remittance fees can have larger impacts on remittance flows than might have been expected. Such reforms include increases in competition in money transmission markets or improvements in information for migrants on the relative costs of different money transmission services. These results also suggest policies which improve information flows between migrants and their families back home (e.g., improved telecommunications) may promote greater remittance sending. 3.4 Policies to channel remittances toward purposes with broader development impacts The recent experimental studies on facilitating savings access for migrants, Ashraf et al (forthcoming) and Chin et al (2011) reveal that migrants value control over savings, both in the origin country and in the host country. These findings raise the possibility that migrants may have preferences for other uses of remittances in their origin countries, aside from savings. Such preferences have potential development implications if migrants have stronger desires to see remittances used for purposes that have longer-term and broader development impacts. For example, migrants may have stronger desires that remittances be used for productive investment, housing, education, or other purposes that enhance households’ longer-run income prospects and/or generate positive spillovers to other households. Migrants may also prefer that some of their remitted funds be used to directly fund public or investment goods for the broader community. It is also possible that migrants might send more remittances in total if a reliable way could be found to channel their funds to these various purposes. A large variety of initiatives are ongoing to channel migrant remittances toward longer-run household investments (enterprises, housing, human capital investments) as well as public goods at the community level. One example is an initiative in collaboration with a Brazilian state- owned bank, Caixa Economica Federal, and the Serviço Brasileiro de Apoio à Micro e Pequena Empresas [Brazilian Service of Support to the Micro and Small Enterprise] (SEBRAE) to catalyze small enterprise investment in Brazil by Brazilians in the U.S. The project provided web-based entrepreneurship training for migrants, and a suite of financial products (remittance, credit, and savings) to support establishment and development of the new businesses in Brazil. This and other projects supported by FOMIN are surveyed in Hall (2010). Programs that are 21 more oriented toward promoting investments by home-town associations (HTAs) are covered in the next section. Evidence on the demand for and impact of mechanisms that help migrants make such investments in the home country is somewhat limited at present, but a couple of studies provide initial insights. Both studies are focused on use of remittances for education, and seek to enhance migrant ability to channel funds toward the education of particular individuals (of their choice) in their home country. Ambler et al (2014a) implemented a randomized controlled trial to test migrant demand for a mechanism that allowed them to channel remittances toward educational expenditures for a particular individual of their choice in the home country. The study population was migrants from El Salvador intercepted at Salvadoran Consular locations in Washington, DC. Migrants in the study were randomly assigned to either a control group (that received no intervention) or to one of three treatment groups. Respondents in each treatment group were offered a new remittance product, named “EduRemesa”, that allowed migrants to target remittances toward the education of a specific student they selected in El Salvador. Migrants would allocate funds to an EduRemesa in certain amounts (ranging from US$300 to US$800), and the selected student in El Salvador would receive an ATM card. The total amount of the EduRemesa funds would be deposited over 10 equal monthly increments over the course of the school year. The EduRemesa did not actually control use of the funds for education (the students simply withdrew cash which could be freely spent on anything), but sponsored students were told that the funds were intended to support their education. The three treatment groups differed in the level of subsidy provided for the EduRemesa. In the “no match” treatment group, migrants had to provide the full amount of EduRemesa funds. This treatment group was included to establish migrant demand for a remittance mechanism that enhanced targeting of remittances for education of a particular student. The second treatment group provided a 1:1 match for the EduRemesa: the research project contributed $1 for each $1 the migrant contributed. The third treatment provided a 3:1 match. The match treatments were offered to assess the extent to which migrant demand for the EduRemesa mechanism could be enhanced with matching funds. 22 The study found no demand at all (exactly zero take-up) for the EduRemesa among migrants in the “no match” treatment group, a small amount of take-up (7%) in the 1:1 match treatment, and 19% take-up in the 3:1 match treatment. The 3:1 match treatment led to statistically significant increases in total educational expenditures on the target student, a reduction in that student’s labor supply, and an increase in the likelihood the target student attended private school. The increase in target student educational expenditures did not appear to come at the expense of reduced expenditures on other students in the household. All told, the EduRemesa study finds no evidence that migrants have an unsubsidized or “pure” demand for control over the use of remittances for educational purposes. But migrants do appear interested in channeling remittances toward educational expenditures of specific students when given matching funds to do so, and when this occurs there are positive impacts on beneficiary students. In a complementary study, De Arcangelis et al (2014) partnered with a Philippine bank (Bank of the Philippine Islands) with branches in Rome to design and pilot-test a new remittance product, called EduPay. This product allowed migrants overseas to channel tuition payments for particular students directly to those students’ educational institutions in the Philippines, avoiding the need to send tuition payments via family members or others in the Philippines who might not be completely trusted to make such payments reliably. This pilot study consisted of two components: 1) the pilot implementation of new remittance product, to provide proof of concept and estimate the level of demand, and 2) a survey to better understand how enhanced control over education expenditures in the home country might affect the volume of remittances sent home by migrants. Proof of concept was demonstrated by successfully implementing a total of 178 EduPay payments for 55 students in the Philippines. In a relatively population-representative sample (not screened for interest or suitability in any way), 6.6% of those approached went as far as to fill out an authorization form allowing our study to contact the Philippine student and school and to make payment arrangements, although bureaucratic delays from some of the schools acted as a constraint on some transactions being made. In another sample screened for suitability and who were willing to respond to an extensive research survey, 43.9% filled out the authorization form. 23 Experimental responses indicated that migrants are willing to remit more to beneficiaries in the Philippines when their transfers can be “labeled” as intended for educational expenses. The impact of allowing labeling was to increase transfers by 15.3%. On top of this “labeling”, the impact of actually channeling funds to educational institutions is relatively modest (only a 2.2% increase on top of the labeling). These results indicate that a remittance product that simply allowed senders to attach a label to remittances as intended for education could have nearly as much impact on remittance sending as a product that actually channeled payments to schools. Note that this result is inconsistent with the finding of Ambler et al (2014a), discussed above, that Salvadoran migrants have zero demand for the unsubsidized EduRemesa educational remittance product. The inconsistency in results across these studies mean that it is important to investigate the relative impacts of education-labeled vs. education-channeled remittance products in follow-up work, to ascertain whether the experimental responses found by De Arcangelis et al (2014) hold up in a real-world setting. 3.5 Policies to encourage communal remitting and home-town associations A home-town association (HTA) consists of a group of migrants coming from the same community in the migrant-origin country. Often these associations have been spontaneously created by migrants as a way of socializing with one another, providing services for new migrant arrivals, and potentially acting to work together to help implement social projects back in their home community. Beauchemin and Schoumaker (2009) use event history models to show that, controlling for other observables, villages in Burkina Faso with a migrant association, are four times as likely to have a health center, 2.8 times as likely to have a primary school, and 2.6 times as likely to have a road, suggesting that these HTAs help build local infrastructure. Chauvet et al. (2013) examine the impact of Malian HTAs in France on the provision of local public goods in Mali using panel data analysis. This way they can use a difference-in-differences strategy to compare changes in infrastructure in villages which have an HTA start activities to changes in villages without HTAs. They find evidence that Malian HTAs have helped improve schools, health centers and water amenities in Mali. A challenge facing both studies is the non-random formation of HTAs and their activities, which methods such as difference-in-differences only partially deal with. 24 Nevertheless, these results do suggest that HTAs can have positive effects on local infrastructure provision, helping alleviate concerns that collective remittances may simply substitute for public finance, or end up being spent on projects that have more limited development impacts for their home communities. This then raises the question of whether policy should try to encourage the formation of such associations, and/or try to encourage such associations to engage in more of such activities. One simple thing governments can do is not stand in the way of such associations: Chauvet et al. (2013) note that French law prohibited foreigners from gathering in associations, and the rise in Malian HTAs occurred following a change in this law. But a number of migrant-sending countries also implement policies to actively encourage HTAs, and to try to get them to send more collective remittances. The most famous example of such an approach is the Mexican tres por uno (3 x 1) program. This program has a long history, beginning in the state of Zacatecas in the 1970s, where some municipalities agreed to match the contributions of migrant associations toward public works (1 for 1). In 1992, the state government added to the match so that each dollar sent by migrants was matched by a dollar each from the municipality and state (2 x 1), and in 1999, federal funding was also added to the match to make it three dollars matched for each dollar contributed by the association (Garcia Zamora, 2007). The program became a nationwide program in 2002, administered by the Ministry of Social Development (SEDESOL), and by 2010 had a program budget of $US1.7 billion (Duquette-Rury, 2014). The money is used to finance local development projects such as electrification, water, road paving and maintenance, education and health projects, and town beautification (Aparicio and Mesguer, 2012). Despite its long history, until very recently the program had not been subject to any rigorous evaluation. One key challenge is the non-random selection of municipalities to participate in the program. Aparicio and Mesguer (2012) examine the correlates of participation, and find that participation is unsurprisingly higher in high migration communities. However, since very poor municipalities have little migration (McKenzie and Rapoport, 2007), the result is that the program can be somewhat regressive, directing state and federal funding toward somewhat wealthier communities. Moreover, the program was launched nationwide by President Vicente Fox of the PAN party, and they find that municipalities and states with greater PAN electoral support were also more likely to participate. 25 Duquette-Rury (2014) provides the first evaluation of the program that tries to control for selective participation. She estimates the impact of participating in 3x1 over the 2002-2008 period on changes in public goods infrastructure between 2000 and 2010. To attempt to deal with selection, she estimates a Heckman selection equation to model participation in the program, with the exclusion restriction being that having a PAN incumbent in the 2000 electoral cycle, conditional on vote shares in later elections, predicts participation in the program but has no independent effect on public goods provision. Under this assumption, she finds 3x1 program expenditures to significantly and positive affect household access to sanitation, water and drainage in participating rural villages. However, she also finds that households receive less family remittances as collective remittances to their municipalities increase. Taken together, this evidence does support the idea that home town associations can increase the provision of local infrastructure. A number of qualitative studies have expressed concerns about the extent to which these projects are sustainable, with funding not often being provided for maintenance (Torres and Kuznetsov, 2006). But the larger concern from the point of view of policy is twofold: first, it is unclear empirically the extent to which the matching funds lead to more of such projects being undertaken, versus crowding out funding that the associations would otherwise provide on their own. In particular, evidence from the charitable giving literature has found that 2 for 1 and 3 for 1 matches lead to no increase in giving relative to 1 for 1 matches (Karlan and List, 2007). Second, given the sometimes regressive nature of such programs, it is unclear whether public funding devoted to this program is better for development than other uses this social funding could be put toward – for example, in Mexico SEDESOL also runs the Progresa/Oportunidades program, and it is far from clear that additional dollars going toward 3x1 have greater impact than using this funding instead for cash transfers to poor households. Since a number of other countries, including Haiti, Somalia, the Philippines, Peru, and Colombia (Duquette-Rury, 2014) are looking to implement their own co-financing programs with HTAs, additional evaluations to answer these questions are important. In addition to promoting collective remittances, several countries also try to encourage their migrants abroad to invest in projects in the origin country. For example, the Philippines Embassy in Qatar brings agricultural projects to pitch as potential investments to migrants there. Most of these programs seem relatively small in scale, and it is unclear whether the government has a 26 comparative advantage in identifying investment projects. A notable example is the Mi Comunidad (My community) project that was launched in 1996, by the Mexican state of Guanajuato. The idea was to use investments by migrants to start small maquiladora garment manufacturing plants in migrants’ communities of origin. The state provided technical assistance, worker training, and assistance to get the projects working. Torres and Kuznetsov (2006) looked at the program around 2000, and viewed it as highly promising, noting it had set up 21 plants and generated 500 permanent jobs in the home communities. But Iskander (2005) notes that the program was a dramatic failure, with only four maquilas surviving after five years, and those that remain were struggling. She notes several issues: the very physical isolation and lack of opportunity that was a spur for migration also isolated plants from production and supply chains, and workers, once trained, often migrated for better prospects in larger cities or abroad. 3.6 Integration policies for migrants A number of destination countries offer public programs designed to facilitate the economic and social integration of immigrants. In some cases these are voluntary, while in others they are compulsory for certain groups of migrants such as asylum-seekers or migrants receiving welfare benefits. These programs have been used for some time in the Nordic countries, but have also been introduced recently in several other countries such as Germany (Rinne, 2013). Such programs can include language training, assistance finding jobs, and also information on the culture and norms of the country. These programs can improve development outcomes to the extent they allow migrants to access better jobs, earn higher incomes, and/or have better mental health and subjective well-being through easier assimilation. There is a large literature which shows associations between earnings and being proficient in the language of the destination country (e.g. Chiswick and Miller, 1995), although some evidence suggests the returns to knowing the native language can be relatively low for low-skill occupations (Berman et al., 2003). Despite this, there is very little rigorous evidence that shows that stand-alone language training for migrants is successful in increasing language proficiency and raising employment outcomes. A before-after study of migrants in Canada by Weiermair (1976) found gradual improvement in immigrant wages after undertaking a language course, but it is unclear how much of these wage gains would have occurred anyway. Hayfron (2001) uses an instrumental variables approach to try to measure the impact of language training in Norway, 27 finding positive impacts on proficiency, but no impact on earnings. However, the instruments used for language proficiency (whether or not they have a Norwegian wife, and their mother tongue) seem likely to also have independent effects on labor market outcomes. More rigorous evidence comes from an evaluation of an immigrant integration program in Finland (Sarvimäki and Hämäläinen, 2012). They take advantage of a discontinuity induced by a law change, whereby participation in the integration program was only obligatory for those who entered the population register less than two years before the reform was launched. Using regression-discontinuity analysis to compare outcomes for migrants on either side of the reform deadline, they estimate that the integration plan increased employment and annual earnings threefold, and halved social benefits. They note their estimate is a local treatment effect, applicable to individuals on unemployment benefits who only took courses when required to. They suggest that one of the main impacts of the reform was to provide more resources for language training, and to allow immigrants to retain unemployment benefits while taking a language course. The training course is relatively expensive (13,000 euros), but the benefits in terms of higher earnings for those induced to take the courses appear to be larger than this. However, they note that even with this program, the average person induced to take courses as a result of the reform was still out of work for half of his sixth year in Finland. A related set of policies aims to better connect immigrants to jobs. Rinne (2013) and Butschek and Walter (2013) provide overviews of evaluations of active labor market policies intended to help immigrants, noting that few of these programs are explicitly targeted at immigrants. Joona and Nekby (2012) use a randomized experiment in Sweden to examine whether more intensive counselling and coaching by public employment caseworkers improves the employment of immigrants relative to those just participating in the standard introduction programs for new immigrants. They find intensive coaching results in a six percent higher probability of being employed, but based on this, calculate that the costs of the program exceed the benefits. 4. Policies directed toward returning migrants Given the large expected income gains most international migrants face when comparing working abroad to working in their home country, staying abroad may be the privately optimal option for many migrants. Nevertheless many migrants do return. This is most obvious for 28 temporary worker programs which require workers to return at the end of a specified period, and migrant-receiving countries use a number of policies to try to ensure that these workers do return. 6 However, return rates can also be quite high for migrants not participating in temporary programs: for example, Dumont and Spielvogel (2008) report exit rates within the first five years of residence ranging between 19 percent in the United States and 60 percent in Ireland, while Gibson and McKenzie (2012) show high return rates for the highest academic achievers from five countries. Such return can be motivated by preferences to be closer to family or to other non-income aspects of their home country, by relative changes in the circumstances of the home and destination countries that change the relative attractiveness of the two, by particular migrants not doing as well in finding work as they expected ex ante, and by target savers having achieved some savings goal abroad that enables them to then carry out some investment they desire in their home country. 7 Many migrants are thus choosing whether to return or stay abroad, and presumably making what they feel is the best choice for them. There are three broad types of policies that then interact with these decisions. The first are policies that seek to remove regulatory, bureaucratic, and informational barriers that prevent individuals who might want to return from doing so, or from being as productive as they could be upon return. The second are policies that seek to get people to return who would otherwise choose not to, by changing the financial and other incentives facing their return decision. The third are policies that focus on making return migrants more productive and reintegration easier for migrants when they do return. Table 3 summarizes these types of policies, and we discuss each in turn, although note that in practice these categories are not mutually exclusive, with a number of policies having components of each of these types. Furthermore, we note that the evidence base is extremely weak for almost all of these policies. 6 These policies can include restricting the ability of the migrant to bring his or her family members, penalties on employers if workers overstay, using the potential of future work spells as incentives, implicit competition between sending communities, etc. See Gibson and McKenzie (2014) for discussion of this in the New Zealand seasonal worker case, and Dumont and Spielvogel (2008) for more OECD evidence. These policies are not intended to directly improve development outcomes (although they may of course still affect them), and so we do not discuss them in this paper. 7 Yang (2006) shows that return migration of overseas Filipino workers is more likely when migrant host countries experience more negative economic (exchange rate) shocks. 29 4.1 Policies to remove the regulatory, bureaucratic, and informational barriers that inhibit return migration There are a number of regulatory and bureaucratic barriers that can make it more difficult and costly for migrants to return to their home countries. One important example is citizenship or residency rights. Migrants from countries without dual citizenship may have had to give up their home country citizenship in order to gain citizenship abroad. If they wish to return later to their home countries, they may face difficulties in being allowed to move permanently back, as well as the disincentive of having to close off the option of migrating again by potentially giving up their newly acquired foreign citizenship. Individuals with only legal residency abroad may be even more reluctant to move back than those with foreign citizenship, since returning home can result in violating continuous residence requirements required for them to retain permanent residence status. For example, permanent residents in the United States can lose their permanent residence status if they remain outside the U.S. for more than one year. 8 Le Blang (2011) uses cross-sectional data on migrants in Spain and Germany, and finds an association whereby migrants from countries which offer dual citizenship send more remittances and express higher intents to return. Dual citizenship also can confer benefits on migrants if they do not return. Mazzolari (2009) uses a difference-in-differences approach to compare immigrants in the U.S. from five Latin American countries which granted dual citizenship rights in the 1990s to migrants from other Latin American countries, and finds the migrants granted dual citizenship rights increased employment and wage earnings, and reduced their use of welfare benefits relative to the comparison group. The number of countries which allow for dual citizenship has grown dramatically, from only 26 in the mid-1970s to 84 countries in 2006 (Le Blang, 2011). Dual citizenship makes it easier for migrants to return. However, many migrants marry citizens of other countries, and so another constraint they face are limits to the ability of their non-citizen spouse and children to live, work, and attend schools in the migrant’s country of origin. Malaysia gives permanent residence status to foreign spouses and children as part of its efforts to facilitate return migration (Lowell, 2001). It is likely such efforts help, and they are certainly unlikely to reduce development impact, but we are unaware of any evaluation evidence. 8 http://www.uscis.gov/green-card/after-green-card-granted/maintaining-permanent-residence [accessed March 26, 2014]. 30 An additional bureaucratic constraint to return migration for many migrants concerns the portability of their social benefits, particularly retirement benefits. There are two issues here. The first is that migrants who work in multiple countries may not accrue enough years of work to become fully vested in the pension systems of either their home or their destination countries. The second is whether they retain eligibility to receive pension payments if they return to their home countries. OECD countries have many bilateral agreements, but coverage is very limited for migrants who move from one developing country to another (Avato et al, 2010). Pension portability can directly improve the well-being of return migrants, but we are unaware of literature which quantifies this, or which looks at its impact on the rate of return migration. In addition to regulatory and bureaucratic barriers, informational barriers may prevent some migrants from returning. Information may be lacking about specific job opportunities at home, or about changes in conditions in the home country. A number of governments attempt to reduce job search frictions by making it easier for domestic firms to locate emigrant workers who may be interested in returning, and vice versa. For example, Jamaica’s returning residents program has a databank of migrants abroad that prospective employers can use (Thomas-Hope, 2004), Bulgaria runs an annual job fair to try to initiate direct contact between Bulgarian emigrants and leading companies in Bulgaria (TFMI, 2012), and Moldova has held job fairs in Italy to provide information about job opportunities back in Moldova. 9 It is unclear how successful any of these efforts have been in terms of increasing return migration or improving the jobs that return migrants get. A more specialized information barrier is that facing refugees, who may not have migrated for work reasons, and who may wish to return to their home communities providing that peace and reconciliation efforts, or disaster recovery, has made the home community safe to return to. They may lack information as to the conditions on the ground, making them cautious about returning. The UNHCR and IOM programs for the reintegration of refugees organize “go-and-see” visits which are intended to overcome this constraint 10. There does not appear to be evidence as to whether this leads to more return and better outcomes for these refugees. 9 http://www.legal-in.eu/en/archive/100-job-fair-regarding-moldovan-labour-market-in-italy [accessed 26 March, 2014]. 10 http://www.unhcr.org/pages/49c3646cfe.html [accessed 26 March, 2014]. 31 4.2. Policies intended to change the financial and other incentives to return Even in the absence of regulatory, bureaucratic, and informational barriers, many migrants may not choose to return to their home countries. There are two main cases where policy efforts try to change these private decisions. The first case concerns asylum-seekers and some refugees and unemployed migrants, with two potential policy motivations – a desire of migrant receiving countries to avoid the costs and expenses of either forcefully deporting individuals or of having individuals dependent on welfare benefits, along with an idea that these are vulnerable groups who may be financially constrained from returning and setting up viable livelihoods. The second, quite distinct case, concerns high-skilled emigrants, where the idea is that their return will have positive externalities for their home countries that they fail to take account of in their private decisions. We discuss these two cases. Assisted voluntary return (AVR) programs offer rejected asylum seekers assistance in returning to their country of origin. 11 The typical program involves paid transportation and logistics facilitation to enable travel to the home country, with some programs also offering grants to set up small businesses or access to training programs once they arrive. Destination country governments find these programs to be considerably less costly (between one-tenth (Black et al, 2011) and one-quarter (Strand et al., 2008) of the cost) than forced return of rejected asylum applicants (who may otherwise go through appeal processes and involve costly deportation procedures). The counterfactual here is not clear, since in the absence of the program, migrants may end up being forcefully deported, but could also potentially gain the right to remain in the destination country through an appeal process. The limited available evidence suggests that these programs are often not very popular. Strand et al., (2008) discuss the case of Afghan nationals in Norway, where close to 2000 had received a final rejection to their asylum application, but only 69 Afghan adults chose to return through this program over a two year period, with more than 206 forcibly returned over the same time frame. The situation is a bit different for refugees and unemployed migrants, who have the option of remaining in the destination country. A number of destination countries have voluntary return programs or “pay-to-go” schemes intended to entice refugees to return to their countries of origin 11 The IOM now refers to these programs as AVRR (Assisted voluntary return and reintegration). We follow the terminology used in the literature we cite, which allows for these programs not to include a reintegration component. 32 after conflict has ended, and to entice unemployed migrants to return to their home countries rather than receive welfare assistance. These programs have a long history, with France’s Aide au Retour program launched in 1977 to target unemployed migrants in France and offer 10,000 French Francs to return to their country of origin for good (Plewa, 2012). The programs typically provide airfare, some reintegration assistance, and a lump sum resettlement amount which can be reasonably sizeable (e.g. Denmark offered Iraqi immigrants up to USD 9,000 per adult and USD 10,500 per child, Dumont and Spielvogel, 2008). Black et al. (2011) have identified 128 such programs, noting a resurgence in interest during the recent global economic crisis. As an example, they note Spain’s Voluntary Return Plan targeted unemployed foreign nationals, and was launched in November 2008. However, by April 2010, only 11,400 immigrants had agreed to leave Spain through this program, which was a tiny fraction of the unemployed immigrant population, and was mostly made up of Latin Americans, despite the target of the policy likely to be Moroccans. The low take-up rates for these voluntary return programs suggest that most migrants do not think that participating in these programs will improve their well-being. It remains unclear the extent to which such programs largely subsidize the returns of those who would have returned anyway. Concerns about “brain drain” and a desire to benefit from the skills and knowledge that migrants have gained abroad motivate a range of policies designed to foster the temporary or permanent return of high-skilled migrants. Temporary return programs typically aim to link high-skilled emigrants to opportunities to help their home countries, relying on a combination of altruism and coverage of the costs of participating. A prominent example is the UNDP’s TOKTEN (Transfer of Knowledge Through Expatriate Nationals) program, launched in 1977. 12 The program relies on volunteers who return to their country of origin for a period of between two weeks and three months to share their expertise. They are paid travel costs and a small allowance, but not professional fees. A total of around 5,000 people have participated in this program in nearly 50 developing countries over a 20 year period (Dumont and Spielvogel, 2008). The number of consultants involved in any particular country can be small – Sri Lanka received 43 consultants over the 8-year period 1996 to 2004, going to a mixture of universities, government ministries, and NGOs. A qualitative assessment of the performance in Sri Lanka by Wanigaratne (2006) 12 http://www.ilo.org/dyn/migpractice/migmain.showPractice?p_lang=en&p_practice_id=26 [accessed March 26, 2014]. 33 revealed incidences where universities appear to have benefited from this exchange, but also cases where consultants made technical recommendations that were not suitable for local conditions, and noted that the sporadic nature of the engagement made sustained impacts doubtful. Such programs are also offered by some developing countries themselves rather than through international agencies – Thailand’s Reverse Brain Drain project aims to facilitate technical linkages between Thai institutions and migrants abroad, while Argentina’s RAICES program supports short-term returns of scientists to Argentina (Lowell, 2001; Jonkers, 2008). A related example comes from the Temporary Return of Qualified Nationals to Afghanistan (TRQN) project studied by Kuschminder (2013), which brought 59 highly-skilled Dutch-Afghans to work with a variety of public and private institutions in Afghanistan for three months. Again participants are officially volunteers, receiving living, travel, accommodation, and travel allowances within Afghanistan that total up to USD 2000 per month. Kuschminder gives as examples participants who helped teach new computer skills, helped design new curricula for university courses, and a participant who was a trained engineer who taught a specialty course on electrical engineering to workers from private firms. Again participants and those they worked with complained the time period was too short in many cases, and it is unclear how much these migrants were able to add value beyond what would have occurred in their absence. This second point is perhaps less of an issue in post-conflict societies with severe skill shortages, but there is little evidence to date to measure actual impacts. Other programs seek to encourage high-skilled migrants to permanently return to their countries of origin, or at least to return for a period of several years. Countries provide a broader range of incentives in an effort to do this. These have included features such as tax exceptions, interest- free or low interest loans, temporary salary supplements to facilitate career entry, assistance with housing, with schooling for their children, and with employment for their spouse (Lowell, 2001; Jonkers, 2008; TFMI, 2012). An example is Malaysia’s Returning Expert Program 13, which provides a flat tax of 15 percent on employment income for 5 years, the ability to import 2 cars tax-free, and which gives permanent residence status to a foreign spouse and children within 6 months. Sometimes these programs are targeted to recruit skilled individuals back to the public 13 http://www.talentcorp.com.my/malaysians-abroad/returning-expert-programme// [accessed March 26, 2014]. 34 sector. An example is the Phillippines’ “Return of Talent” program, which over two phases recruited 60 highly qualified Filipinos abroad to fill vacant positions in the public sector (Tornea, 2003). Perhaps the best-known of these programs is the Return of the Qualified African Nationals (RQAN) program managed by the IOM, which has since evolved to become the Migration for Development in Africa (MIDA) program. The RQAN provided airline tickets for the return migrant and their dependents, shipment of their possessions, purchase of professional equipment needed for their work, settling-in expenses, and assistance finding positions in their home countries (Lowell, 2001). It placed around 2,000 highly skilled persons in 41 African countries between 1974 and 1990 (World Bank, 2006), so the effective placement in any given African country in a particular year was low on average. Existing “evaluations” of this program appear to consist largely of surveys which directly ask those participating whether they feel they are contributing to development in their home country, and whether they felt the program was important in their decision to return (e.g. Pires et al, 1996). The main purpose of such programs is to generate externalities for others in the developing countries of origin. However, the numbers of people involved in these programs have often been relatively small, which coupled with qualitative evidence on the types of activities these participants have been involved in, suggest that the aggregate externalities are likely rather limited. In many cases the programs may just end up subsidizing the return of individuals who were likely to return anyway (potentially speeding up this return), and as Dumont and Spielvogel (2008) note, these policies can have adverse effects in terms of feeding resentment among non- migrants 14, or even potentially encouraging more people to emigrate in order to get these benefits upon return. 4.3 Policies intended to make return migrants more productive and reintegration easier A final set of return migration policies are intended to make it easier for migrants returning to be productive, earn more, and re-adjust to life in their home countries. One of the most comprehensive of these types of programs is the Overseas Foreign Worker (OFW) reintegration 14 For example, an evaluation of the return for qualified nationals program in Bosnia-Herzegovina noted that grants to help returnees set up businesses were criticized by businessmen who had stayed in the country (Sandgren, 2001). 35 program provided by the Philippines. It attempts to address both the economic and social needs of returning workers. Training programs and training are offered for those who would like to start up small businesses, while a psycho-social component includes services like family counselling, stress debriefing, and community organizing programs intended to help the migrant fit back into life in the Philippines (Tornea, 2003). As a second example, Poland offers returning migrants dedicated websites, and a ‘return migrant handbook’ with information on programs to help them find work and deal with the logistics of resettlement (TFMI, 2012). Reintegration programs that involve training and/or credit to start small enterprises are also widely used in programs that help refugees resettle. Strand et al. (2008) provides qualitative evidence on the success of this approach for Afghans returning from Norway. They note that in a few cases moderately successful businesses were set up, but the majority appeared to exist only on paper, with participants using the business as a mechanism to convert start-up business grants into cash, and most respondents stating an intention to re-migrate. While such programs can sound intuitively appealing, and potentially overcome financial or skill constraints that limit the ability of migrants to work productively, there is no existing evidence as to their success. Moreover, there are at least three concerns with such programs. The first is that not everybody wants to be an entrepreneur, and many return migrants will have been working in wage jobs previously, with no experience in running a business. Second, existing evaluations of training programs have had at best mixed results, even amongst those individuals interested in starting businesses (McKenzie and Woodruff, 2014) and there is no reason to expect return migrants to be particularly good at running businesses. Third, it is unclear why such programs should be targeted explicitly at return migrants, rather than being part of a portfolio of training and work assistance options offered to all individuals in a given region. Finally policy makers can help make return migrants more productive by facilitating the recognition of the qualifications and skills gained abroad. For example, Argentina’s RAICES program offers the translation and accreditation of qualifications formally earned abroad (Jonkers, 2008). The Bologna Process aims to formalize recognition of higher education qualifications within Greater Europe, but many migrants from developing countries can still experience difficulties getting overseas qualifications recognized within their home countries. A second issue is to how to get recognition for on-the-job skills learned abroad, and for qualified 36 migrants to be able to return to employment systems that principally reward years of service – Gibson and McKenzie (2012) find this to be an issue for highly skilled migrants returning to public sector jobs in developing countries. However, again there does not appear to be research which demonstrates the impact of policy efforts to better recognize skills earned abroad. 4.4. Why is the evidence base so bad on return migration policies? Table 3 shows that almost all policies directed toward returning migrants have no rigorous evidence as to their effects. The following example from IOM (2005, p.iii) is representative of the state of most evidence on return migration policies “This evaluation report summarizes the findings, conclusions and recommendations of a multistakeholder team using participatory approaches to an evaluation of IOM’s reintegration projects … It reflects the application of a “learning-focused” process, aimed more at identifying lessons learnt than at assessing actual performance or impact levels.”. That is, the standard approach, if an evaluation is done at all, is to attempt to examine the process of the policy, with no reference to a counter-factual of what would have happened without the policy. We can speculate at several reasons why very little in the way of rigorous evaluations have been done for these types of policies. A first reason is that the size of many of these programs is very small – a program taking 20 or 30 high-skilled workers over a couple of years is going to find it very hard to have sufficient statistical power to detect any impact of the program unless the impacts are massive. A second, related, reason is that many of these programs have not been very popular. As a result, there may seldom be a natural comparison group of people who wanted to participate in the program but who were not able to be accommodated. A third reason is that some of the intended outcomes (such as spillovers from high-skilled emigrants coming back) can be difficult to measure. A final reason may concern the types of organizations carrying out these policies. The International Organization for Migration (IOM) and national government migration agencies are the main implementers of many of these policies, and these agencies have not typically funded or had staff trained in rigorous evaluations. However, we believe none of these obstacles is insurmountable, and, especially given the dire state of existing knowledge, there is ample scope for future work to provide more credible evidence on these types of policies. 37 Conclusion Currently, strong evidence is lacking on the impacts of many of the wide range of policies intended to enhance the development impacts of international migration. Nevertheless, in the last few years a growing literature has begun to demonstrate the impacts of some of these programs. The evidence to date largely comes from a few studies of specific cases, and so more research is needed to examine how generalizable these results are. Nevertheless, the preliminary evidence suggests some areas of policy success: bilateral migration agreements for seasonal migration with countries whose workers have few other migration options; developing new savings products for migrants which allow them some control over how this money is used; and some efforts to provide financial education to migrants and their families. In some other areas there is suggestive evidence, which combined with theory, offers a rationale for policy action: efforts to lower the cost of remittances, reduce the cost of a passport, provide dual citizenship, and remove exit barriers to migration. Finally, existing research suggests reasons to be cautious about some other types of policies: enforcing strong rights for migrants like high minimum wages does make some migrants better off, but at a cost of reducing the opportunity to migrate for others; integration programs that provide language training and job search assistance for migrants can have positive effects, but at relatively high costs. Nevertheless, there are many types of programs for which the only existing evidence is largely case study or process evaluation at best, with no consideration of a counterfactual. This is particularly the case for a large range of return migration programs, but is also true of popular policies like matching funds to encourage communal remittances. Moreover, while one can think of reasons to justify these programs (e.g., subsidies to get high-skilled migrants to return might be justified in terms of the externalities they bring), there are also reasons to be apprehensive about the true impacts of these programs (e.g., subsidies might just end up funding the return of people who would return anyway, or could even deter return by signaling that the home country is so undesirable that it requires subsidies to get someone to live there). There is therefore a strong need for research to provide better evidence on many of these migration policies in order to ensure that they really can enhance the development impacts of international migration. 38 References Ambler, Kate (2013), “Don't Tell on Me: Experimental Evidence of Asymmetric Information in Transnational Households,” working paper, IFPRI. Ambler, Kate, Diego Aycinena, and Dean Yang (2014a), “Channeling Remittances to Education: A Field Experiment Among Migrants from El Salvador,” American Economic Journal: Applied Economics, forthcoming. Ambler, Kate, Diego Aycinena, and Dean Yang (2014b) “Remittance Responses to Temporary Discounts: A Field Experiment among Central American Migrants,” working paper, University of Michigan, 2014. Aparicio, Francisco Javier and Covadonga Meseguer (2012) “Collective Remittances and the State: The 3x1 Program in Mexican Municipalities." World Development 40 (1):206-222. Asis, Maruja and Dovelyn Rannveig Agunias (2012), Strengthening Pre-Departure Orientation Programs in Indonesia, Nepal, and the Philippines, International Organization for Migration and Migration Policy Institute. Ashraf, Nava, Diego Aycinena, Claudia Martinez A., and Dean Yang (forthcoming) “Savings in Transnational Households: A Field Experiment among Migrants from El Salvador,” Review of Economics and Statistics. Avato, Johanna, Johannes Koettl and Rachel Sabates-Wheeler (2010) “Social Security Regimes, Global Estimates, and Good Practices: The Status of Social Protection for International Migrants”, World Development, 38(4): 455-66. Aycinena, Diego, Claudia Martinez and Dean Yang (2010) “The Impact of Remittance Fees on Remittance Flows: Evidence from a Field Experiment Among Salvadoran Migrants,” working paper, U. Michigan. Batista, Catia and Gaia Narciso (2013) “Migrant Remittances and Information Flows: Evidence from a Field Experiment”, IZA Discussion Paper no. 7839. Beam, Emily, David McKenzie and Dean Yang (2013) “Unilateral Facilitation Does Not Raise International Labor Migration from the Philippines”, World Bank Policy Research Working Paper no. 6689. Beam, Emily (2014) “Incomplete Information in Job Search: Evidence from a Field Experiment in the Philippines,” working paper, National University of Singapore. 39 Beauchemin, Cris and Bruno Schoumaker (2009) “Are migrant associations actors in local development? A national event-history analysis in rural Burkina Faso" World Development 37 (12):1897-1913. Beck, Thorsten and Maria Soledad Martinez-Peria (2011) “What Explains the Price of Remittances? An Examination Across 119 Country Corridors”, World Bank Economic Review 25(1): 105-31. Berman, Eli, Kevin Lang, Erez Siniver (2003) “Language-skill complementarity: returns to immigrant language acquisition”, Labour Economics 10 (3): 265–290 Black, Richard, Michael Collyer, and Will Somerville (2011) “Pay-to-go schemes and other noncoercive return programs: is scale possible?”, Migration Policy Institute Paper. Butschek, Sebastian and Thomas Walter (2013) “What active labour market programmes work for immigrants in Europe? A meta-analysis of the evaluation literature”, ZEW Discussion Papers, No. 13-056 Chauvet, Lisa, Flore Gubert, Marion Mercier, and Sandrine Mesplé-Somps (2013) “Migrants' Home Town Associations and Local Development in Mali”, Mimeo. Chin, Aimee, Leonie Karkoviata, and Nathaniel Wilcox (2011) “Impact of Bank Accounts on Migrant Savings and Remittances: Evidence from a Field Experiment,” mimeo, University of Houston. Chiswick, Barry and Paul Miller (1995) “The Endogeneity between Language and Earnings: International Analyses “Journal of Labor Economics, 13(2): 246-288 Clemens, Michael and David McKenzie (2014) “Why don’t remittances appear to affect growth?”, World Bank Policy Research Working Paper no. Clemens, Michael, Claudio Montenegro, and Lant Pritchett (2009) “The Place Premium: Wage Differences for Identical Workers across the US Border.” HKS Faculty Research Working Paper Series, No. RWP09-004. De Arcangelis, Giuseppe, David McKenzie, Erwin Tiongson, and Dean Yang (2014), “Increasing the Development Impact of Migrant Remittances: A Field Experiment on Educational Finance by Migrant Workers,” working paper, Sapienza University of Rome, World Bank, and University of Michigan. DeParle, Jason (2007) “Western Union Empire Moves Migrant Cash Home.” New York Times, Nov. 22. 40 Doi, Yoko, David McKenzie and Bilal Zia (2014) “Who You Train Matters: Identifying Complementary Effects of Financial Education on Migrant Households,” Journal of Development Economics, 109: 39-55. Dumont, Jean-Christophe and Gilles Spielvogel (2008) “Return migration: a new perspective”, pp. 161-222 in OECD International Migration Outlook 2008. OECD, Paris. Duquette-Rury, Lauren (2014) “Collective remittances and transnational coproduction: the 3x1 program for migrants and household access to public goods in Mexico”, Studies in Comparative International Development 49: 112-39. Esipova, Neli, Julie Ray and Anita Pugliese (2011) Gallup World Poll: The Many Faces of Global Migration. Gallup and International Organization for Migration Report. Freund, Caroline and Nikola Spatafora. (2005). “Remittances: Transaction Costs, Determinants, and Informal Flows.” World Bank Policy Research Working Paper 3704. G8 (2009), Final Declaration on Responsible Leadership for a Sustainable Future, Heads of State Summit, L’Aquila. http://www.g8italia2009.it/static/G8_Allegato/G8_Declaration_08_07_09_final,0.pdf Garcia Zamora, Rodolfo (2007) “El Programa Tres por Uno de remesas colectivas en México: Lecciones y desafíos”, Migraciones Internacionales 4(1): 165-172. Gibson, John and David McKenzie (2014) “The Development Impact of a Best Practice Seasonal Worker Policy”, Review of Economics and Statistics 96(2): 229-43 Gibson, John, David McKenzie, and Hala Rohorua. (2006). “How Cost Elastic are Remittances? Evidence from Tongan Migrants in New Zealand.” Pacific Economic Bulletin, 21(1): 112–28. Gibson, John, David McKenzie and Bilal Zia (2014) “The impact of financial literacy training for migrants”, World Bank Economic Review 28(1): 130-161. Gibson, John and David McKenzie (2012) “The Economic Consequences of “Brain Drain” of the Best and Brightest: Evidence from Five Countries”, Economic Journal, 122: 339-75. Hall, Joan (2010), Ten Years of Innovation in Remittances: Lessons Learned and Models for the Future, Independent Review of the Multilateral Investment Fund Remittance Portfolio, Inter- American Development Bank. Hanson, Gordon (2010) “International Migration and the Developing World”, pp. 4363-4414 in Dani Rodrik and Mark Rosenzweig, editors: Handbook of Development Economics, Vol. 5, The Netherlands: North-Holland. Hayfron, John (2001). Language Training, Language Proficiency and Earnings of Immigrants in 41 Norway. Applied Economics 33(15), 1971–1979. International Organization for Migration (IOM) (2005) Evaluation of IOM Reintegration Activities Targeting Vulnerable Returnees and Displaced Populations in Angola. IOM, Geneva. International Organization for Migration (IOM) (2006), “Mass and Micro-Information Campaign Awareness Impact Assessment - Information Campaign to Combat Trafficking in Women and Children in Cambodia Project,” Joint evaluation MOWA/IOM, March. International Organization for Migration (IOM) (2011), “IOM Migrant Training Programmes Overview, 2010-2011.” http://www.iom.int/jahia/webdav/shared/shared/mainsite/activities/facilitating/IOM_Migrant_Tr aining_Programmes_Overview_2010_2011.pdf International Labor Organization (ILO) (2013), “Pre-Departure Training Programme, Republic of the Philippines,” accessed May 1, 2014. http://www.ilo.org/dyn/migpractice/migmain.showPractice?p_lang=en&p_practice_id=72 Iskander, Natasha (2005) “Social learning as a productive project: The Tres por Uno (three for one) experience at Zacatecas, Mexico”, pp. 249-263 in Migration, Remittances, and Development. OECD, Paris. Joona, Pernilla Andersson and Lena Nekby (2012). “Intensive Coaching of New Immigrants: An Evaluation Based on Random Program Assignment”, Scandinavian Journal of Economics 114(2): 575-600. Koen Jonkers (2008) “A comparative study of return migration policies targeting the highly skilled in four major sending countries”, MIREM project analytical report mirem-ar 2008/05 http://cadmus.eui.eu/bitstream/handle/1814/9454/MIREM%20AR2008_05.pdf?sequence=1 Kuschminder, Katie (2013) “Knowledge Transfer and Capacity Building Through the Temporary Return of Qualified Nationals to Afghanistan”, International Migration, forthcoming. Le Blang, David (2011) “Harnessing the Diaspora: Dual Citizenship, Migrant Remittances and Return”, Mimeo. University of Virginia, Department of Politics. Lowell, B. Lindsay (2001) “Policy responses to the international mobility of skilled labour”, International Migration Papers 45, http://www.ilo.org/wcmsp5/groups/public/---ed_protect/--- protrav/---migrant/documents/publication/wcms_201774.pdf Mazzolari, Francesca (2009) “Dual citizenship rights: do they make more and richer citizens?”, Demography 46(1): 169-91. McKenzie, David (2007) “Paper Walls are Easier to Tear Down: Passport Costs and Legal Barriers to Emigration”, World Development, 35(11): 2026-2039 42 McKenzie, David, John Gibson, and Steven Stillman (2013) “A land of milk and honey with streets paved with gold: Do emigrants have over-optimistic expectations about incomes abroad?” Journal of Development Economics, 102: 116-27. McKenzie, David, John Gibson, and Steven Stillman (2010). “How Important is Selection? Experimental vs. Non-Experimental Measures of the Income Gains from Migration.” Journal of the European Economic Association, 8(4): 913–945. McKenzie, David and Hillel Rapoport (2007) “Network effects and the dynamics of migration and inequality: theory and evidence from Mexico”, Journal of Development Economics, 84(1): 1-24 McKenzie, David, Caroline Theoharides and Dean Yang (2014) “Distortions in the International Labor Market: Evidence from Filipino Migration and Wage Responses to Destination Country Economic Shocks,” American Economic Journal: Applied Economics 6(2): 49-75 McKenzie, David and Christopher Woodruff (2014) “What are we learning from business training evaluations around the developing world?” World Bank Research Observer, 29(1): 48- 82. McKenzie, David and Dean Yang (2012) “Experimental approaches in migration studies” pp. 249-269 in Carlos Vargas-Silva (ed.) Handbook of Research Methods in Migration. Edward- Elgar: Northampton, MA Naidu, Suresh, Yaw Nyarko and Shing-Yi Wang (2014) “Worker mobility in a global labor market: Evidence from the United Arab Emirates”, Mimeo. NYU. Norwegian Agency for Development Cooperation (Norad) (2010), “Evaluation of the International Organization for Migration and its Efforts to Combat Human Trafficking.” Omar Mahmoud, Toman, Victoral Licuanan, Andreas Steinmayr, Erwin Tiongson, and Dean Yang (2013), “Harnessing the Development Benefits of International Migration: A Randomized Evaluation of Enhanced Pre-Departure Orientation Seminars for Migrants from the Philippines,” ongoing study. http://www.3ieimpact.org/en/evidence/impact-evaluations/details/720/ Orozco, Manuel (2004). “The Remittance Marketplace: Prices, Policy, and Financial Institutions,” Pew Hispanic Center Report. http:// pewhispanic.org/fi les/reports/28.pdf. Orozco, Manuel, Elizabeth Burgess, and Landen Romei (2010). “A Scorecard in the Market for Money Transfers: Trends in Competition in Latin America and the Caribbean.” http://www.thedialogue.org/PublicationFiles/Scorecard %20Report%206.18.2010_FINAL.pdf Pires, José, Ashraf El Nour and Lindsay McMahon (1996) “Survey among qualified African nationals under the bridging fund scheme of the programme for the return and reintegration of qualified African nationals (RQAN) – Phase III”, International Organization for Migration report, Geneva. 43 Plewa, Piotr (2012) “The Effects of Voluntary Return Programmes on Migration Flows in the Context of the 1973/74 and 2008/09 Economic Crises”, Comparative Population Studies 37(1- 2): 147-176 Rinne, Ulf (2013) "The Evaluation of Immigration Policies" pp. 530-51 in Amelie F. Constant and Klaus F. Zimmermann (Eds.), International Handbook on the Economics of Migration, Cheltenham, UK, and Northampton, USA: Edward Elgar Publishing. Ruhs, Martin (2013) The Price of Rights: Regulating International Labor Migration. Princeton University Press, Princeton, NJ. Sandgren, Claes (2001) “The IOM Project ”Reconstruction, Capacity Building and Development through the Return of Qualified Nationals to Bosnia and Herzegovina” -- An Evaluation”, Report to the International Organization for Migration, Geneva. Sarvimäki, Matti and Kari Hämäläinen (2012): “Assimilating immigrants. The impact of an integration program.” http://cream.conference- services.net/resources/952/3365/pdf/MGDNF2013_0192.pdf Seshan, Ganesh and Dean Yang (2014) “Motivating Migrants: A Field Experiment on Financial Decision-Making in Transnational Households”, Journal of Development Economics, Vol. 108, May, pp. 119-127. Strand, Arne, Arghawan Akbari, Torunn Wimpelmann Chaudhary, Kristian Berg Harpviken, Akbar Sarwari and Astri Suhrke (2008) “Return in Dignity, Return to What? Review of the Voluntary Return Programme to Afghanistan”, CHR. Michelsen Institute Working Paper. Thomas-Hope, Elizabeth (2004) “Migration Situation Analysis, Policy, and Programme Needs for Jamaica”, report for the United Nations Population Fund, http://countryoffice.unfpa.org/caribbean/drive/2004_migration_study_jamaica.pdf Tornea, Vivian (2003) “Reintegration program for migrant workers”, Review of Women’s Studies 13(2): 203-220. Torres, Federico and Yevgeny Kuznetsov (2006) “Mexico: Leveraging Migrants’ Capital to Develop Hometown Communities”, pp. 99-128 in Y. Kuznetsov (ed.) Diaspora Networks and the International Migration of Skills. World Bank Institute, Washington D.C. Transatlantic Forum on Migration and Integration (TFMI) (2012) “Welcome home? Challenges and Chances of Return Migration” http://www.gmfus.org/wp- content/blogs.dir/1/files_mf/1358538301tfmireturnmigration.pdf 44 Wanigaratne, D.G. (2006) “An Evaluation of the UNDP Transfer of Technology through Expatriate National (TOKTEN) Program”, Mimeo. ftp://209.160.24.160/app/erc/app/documents/3205/report/TOKTEN%20Final%20Evaluation%20 Report%20(2).doc [accessed March 26, 2014]. Weiermair, Klaus (1976) “The Economic Effects of Language Training to Immigrants: A Case Study” International Migration Review, 10(2): 205-219 World Bank (2006) Global Economic Prospects 2006: Economic Implications of Remittances and Migration. World Bank, Washington, D.C. Yang, Dean (2006) “Why Do Migrants Return to Poor Countries? Evidence from Philippine Migrants’ Responses to Exchange Rate Shocks,” Review of Economics and Statistics, 88(4): 715- 735. Yang, Dean (2011) “Migrant Remittances,” Journal of Economic Perspectives, 25(3): 129-152. 45 Table 1: Examples of Pre-Migration Policies Policy Examples what does it do? Potential impact on development Evidence Pre-departure training Training and orientation of Philippines' Pre- Eases migrant workers in legal Departure adjustment to new Non-existent, but one Increases worker productivity and rights, customs Orientation country and work study (Omar earnings; allows workers to escape and language Seminars (PDOS); environment, and Mahmoud et al) abuse overseas, IOM migrant training protects in cases of ongoing. productive sessions abuse skills, etc. Evidence from RCTs among Indian workers in Qatar (Seshan and Yang 2014), Indonesian Included in Financial Improves financial migrants and their Philippine PDOS; literacy training decision-making among Raise savings and investment; families (Doi, small programs for workers and migrants and their improve ability to cope with risk McKenzie, Zia 2014), implemented as part their families families left behind and among Pacific of research studies Island migrants to Australia and New Zealand (Gibson et al 2014) Policies to facilitate migration Strong evidence of positive effects from Philippines with 25 Legal agreement to the New Zealand RSE Bilateral countries allow labor migration Allows more individuals to program through migration Pacific Islands with of set numbers or types migrate abroad matched difference-in- agreements New Zealand of workers differences (Gibson and McKenzie, forthcoming) Suggestive evidence Lower passport Makes it less costly to Makes it less costly to from regressions with costs and Many countries migrate migrate other controls procedures (McKenzie, 2007) Many countries don't have these barriers, Suggestive evidence but a number restrict Remove exit Makes it less costly to Allows more individuals to from regressions with women migrating, barriers migrate migrate abroad other controls and others require (McKenzie, 2007) govt permission for all emigration 46 Table 2: Examples of Policies While Migrants are Abroad Policy Examples what does it do? Potential impact on development Evidence Rights policies for migrant workers Sets minimum wages that Can raise the income of the Evidence shows a trade-off between the Requiring migrants to be Comprehensive migrants must be paid; migrants receiving the policy, number of individuals who get to migrate paid the same wages as economic potentially also gives migrants but reduce the demand of and the economic rights they receive (Ruhs, nationals of the rights access to unemployment employers for labor, and openness 2013; destination country benefits and welfare programs of governments McKenzie et al, 2014). freedom to leave, Protects migrants from abuse, No empirical evidence to measure impacts of Ensures migrants receive income freedom from abuse, gives them the ability to leave, changes in these rights, but theory strongly Basic rights promised, and protects against ensuring migrants receive and ensures they get paid for suggests they should be beneficial for physical and sexual abuse pay for the jobs they do work done development Financial access for migrants and their families Evidence from RCT among migrants from El Facilitating El Salvador's Banco Allows migrants to set up Raises asset accumulation and Salvador (Ashraf et al forthcoming) that offering remittance-linked Agricola accounts offered accounts in home country and allows self-insurance via buffer migrants accounts in home country can raise savings accounts in to migrants in US remit funds directly into them stock accumulation savings, but only when migrants have joint or origin country sole ownership of the accounts Facilitating savings Matricula consular Evidence from RCT among Mexican migrants Formal identification document Raises asset accumulation and accounts for (consular ID card) issued (Chin et al 2011) that issuing matricula consular facilitates bank account opening allows self-insurance via buffer migrants in the host to Mexican migrants by raises account opening and savings in host in US stock accumulation country Mexican consulate country Table 2 continued on next page 47 Policy Examples what does it do? Potential impact on development Evidence Policies to encourage communal remitting and home-town associations Promising evidence that HTAs increase public good provision Setting up home- joins migrants together to from time series, difference-in-differences, Can increase public goods like town Malian HTAs in France potentially Heckman selection water, roads, and electricity associations send collective remittances (Beauchemin and Schoumaker, 2009; Chauvet et al. 2013; Duquette-Rury, 2014). No evidence on policies to increase set up of HTAs Can lead to more remittances No evidence to show that match leads to offers migrant association a being Matching funds more funding being sent than would be sent government match for each sent for community needs. May for collective 3x1 program in Mexico otherwise; evidence that program can end up dollar crowd out family remittances. remittances being regressive in funding they send May crowd out other uses of and politically targeted (Duquette-Rury, 2014). public funds Integration policies for migrants Strong associations between language Can increase job prospects, and Norway, Germany, teaches migrants to speak proficiency Language training hence earnings of migrants; Sweden destination country language and earnings in many countries, but little may help in social integration evidence on impact of training. Mixed. Strong impacts for unemployed individuals language training, job search Can increase job prospects, and in Finland in a regression-discontinuity Migrant introduction Norway, Germany, assistance, teaches civics and hence earnings of migrants; (Sarvimäki and Hämäläinen, 2012); benefits don't programs Sweden values of host society may help in social integration exceed costs in randomized experiment adding job counselling to basic program in Sweden (Joona and Nekby, 2012). 48 Table 3: Examples of Types of Return Migration Policies Policy Examples what does it do? Potential impact on development Evidence Policies to remove regulatory, bureaucratic, and informational barriers allows migrants to maintain Can enable migrants to earn more Promising. Cross-sectional 84+ countries rights while abroad, and make it easier associations and difference- Dual citizenship allow of a citizen in their home country for them to return to their home in-differences (Le Blang, 2011; and in destination country country Mazzolari, 2009). Over 1,000 allows migrants to retain access Can increase incomes of return Pension portability bilateral to retirement benefits accrued migrants, and removes a Non-existent. agreements even if they return disincentive to return Jamaica returning help migrants find out about Can remove informational barriers Job fairs and migrant databases resident program job opportunities in home to migration, allow migrants to Non-existent. Bulgaria's job fairs country find better jobs at home Can remove information barriers allow refugees to see whether UNHCR programs preventing return migration, Go-and-see visits home communities are now Non-existent. IOM programs allow refugees to return to home suitable for return communities Policies changing financial and other incentives for return Non-existent, although take-up IRRANA, available to pay for travel and logistics for Not clear, since alternative is Assisted Voluntary Return rates suggest relatively little Afghans in Norway rejected asylum seekers to return usually deportation demand from many migrants Spain's Voluntary Lowers cost of return for those Return Program, Aide pays migrants to return to their intending to return anyway, Non-existent, although take-up Pay-to-go schemes au home countries potentially provides capital that rates suggest relatively little Retour in France can be used to set up livelihoods demand from many migrants Potentially helps transfer Extremely limited. Qualitative connects highly-skilled emigrants knowledge, ideas, and evidence reveals some mixed Temporary return of qualified TOKTEN to home country institutions for technology from abroad, with effects, scale of programs migrants TRQN 3 months of consulting/ spillover benefits to others typically too low for large knowledge-sharing in home country effects. Potentially helps transfer Extremely limited. Qualitative provides funding, and other knowledge, ideas, and evidence reveals some mixed Permanent return of qualified Return of Qualified incentives to bring highly-skilled technology from abroad, with effects, scale of programs migrants African Nationals migrants back to countries of spillover benefits to others typically too low for large origin in home country effects. 49 Table 3 continued: Policy Examples what does it do? Potential impact on development Evidence Policies to make return migrants more productive and adjust reintegration easier Potentially helps overcome Provides start-up funding and credit constraints and skill The Philippines' loans for setting up businesses; Reintegration support programs constraints to allow migrants Non-existent reintegration program business or vocational training; to earn more. May help migrants psycho-social assistance re-adjust to home society. translate and accredit make it easier for return migrants Argentina's RAICES Recognition of skills earned abroad qualifications to be rewarded for skills learned Non-existent program earned abroad abroad, increasing their incomes 50