Policy Research Working Paper 9740 Temporary Migration for Long-term Investment Laurent Bossavie Joseph-Simon Görlach Çağlar Özden He Wang Social Protection and Jobs Global Practice & Development Research Group July 2021 Policy Research Working Paper 9740 Abstract In the presence of credit constraints, temporary migration emigration, reduces the age at which workers depart, abroad provides an effective strategy for workers to accumu- and reduces the duration of their time abroad, which late savings to finance self-employment when they return together lead to higher savings and domestic self-employ- home. This paper provides direct evidence of this link and ment. Reducing the interest rate for entrepreneurial loans its effects on workers’ employment trajectories by using a reduces migration and savings levels, undercutting the new, large-scale survey of temporary migrants from Ban- positive effects on business creation at home. Correcting gladesh. It constructs and estimates a dynamic model that workers’ inflated perceptions about overseas earnings poten- establishes connections between asset accumulation and tial reduces emigration rates and durations, triggering a credit constraints, and, thus, between workers’ migration decrease of both repatriated savings and self-employment and self-employment decisions. Interlinked impacts also in Bangladesh. The findings, which have implications for emerge from simulations of three key policy interventions migrant-sending countries, highlight the need for policies that target migration costs or domestic credit constraints to take into account the linkages between migration and for entrepreneurship. Lowering migration costs increases self-employment decisions. This paper is a product of the Social Protection and Jobs Global Practice and the Development Research Group, Development Economics.. 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://www. worldbank.org/prwp. The authors may be contacted at lbossavie@worldbank.org, josephsimon.goerlach@unibocconi.it, cozden@worldbank.org, and hwang21@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 ∗ Temporary Migration for Long-term Investment Laurent Bossavie1 , Joseph-Simon G¨ orlach2 , C ¨ ¸ aglar Ozden1 , and He Wang1 1 World Bank 2 Bocconi University, CReAM, IGIER, IZA and LEAP JEL codes: J61, D15, F22, O13 Key words: Temporary migration, entrepreneurship, credit constraints, developing countries, migration costs ∗ We would like to thank Nelly Elmallakh and participants at the 13th Migration and Development conference, the Economics of Migration online seminar series, Migrations and Organizations Conference at the University of Pennsylvania, Social Protection and Labor Workshop at the World Bank and colleagues at Bocconi University for helpful comments and suggestions. We are also very thankful to Rubaba Anwar for providing excellent field coordination support during the implementation of the survey in Bangladesh, and to the Bangladesh Bureau of Manpower, Employment and Training for granting clearance to carry out the Bangladesh Return Migrant Survey (BRMS). Funding for this research was provided by the Rapid Social Response Multi-Donor Trust Fund, the DEC Research Support Budget (RSB), Knowledge for Change Program (KCP) and the Multi-Donor Trust Fund for International Trade within the World Bank. 1 Introduction Temporary migration is part of life in many regions of the world. One of the most prominent examples is the migration of millions of mostly low-skilled workers from South and Southeast Asia to oil-rich Persian Gulf countries. Every year these workers depart with temporary employment visas and short-term contracts. Once their contracts expire, workers are no longer allowed to stay, and they return home (OECD 2008; World Bank 2018). Academic and policy research on the subject has largely focused on how these migrants increase their current income and support their families at home while they are abroad.1 In this paper, we examine temporary migration from a different angle: We argue and present compelling evidence from Bangladesh that such temporary migration episodes are an integral part of workers’ savings and investment strategies implemented over a lifetime, with important ramifications on the overall economic outcomes of origin countries. Answering questions on temporary labor migration decisions and related economic out- comes requires a dynamic framework that accounts for individuals’ intertemporal optimiza- tion decisions.2 Data are needed to reveal migrants’ complete labor market trajectories over time and across borders, and to show their investment and savings behavior. Standard na- tional surveys and administrative data sources do not provide these data, as they focus on current labor market outcomes, but do not ask detailed questions about workers’ employ- ment and financial histories. Thus, static analyses based on these data sources are unable to identify the dynamic linkages between key decisions: whether, when, and where to migrate; how long to stay; how much to save; what employment options to pursue back at home; and whether migration provides the means to overcome credit constraints, as we argue is the case in Bangladesh. In this paper, we identify and examine the implications of these linkages by estimating a dynamic model of decisions on emigration, return, and self-employment. We then use the estimated model to evaluate the intertwined effects that arise from key changes in migration or domestic lending policies. Specifically, we examine the effects likely to emerge as a result of a decrease in migration costs, a decrease in interest rates for self-employment loans, or the provision of more accurate information on overseas earnings to would-be migrants. Our simulations show that a reduction in migration fees, or an increase in wages earned overseas raises emigration, but decreases its duration while increasing overall savings and entrepreneurship rates in Bangladesh. We further show that core development policies, such as loan subsidies for entrepreneurs, can trigger a decrease in the extent and duration of 1 See Mobarak, Sharif, and Shrestha (2020) for a recent study on temporary migrants to Malaysia and the evidence on the impact on their families. 2 See Dustmann and G¨ orlach (2016) for a review of the literature. 2 migration, undercutting the incentive to migrate and to accumulate savings. Moreover, we show that providing accurate information on wage levels in destination countries can reduce both emigration rates and durations, leading to lower levels of repatriated savings and less domestic business creation after return. Our model is closely tailored to the institutional context. Temporary migration is per- vasive in Bangladesh, and domestic economic activity strongly depends on migrants’ remit- tances and repatriated savings. Our model captures the key determinants that together influence the decisions about whether to emigrate, the age of departure, the destination, the duration, and the employment decision upon return. Demand for migrant labor in the main destination countries varies with oil revenues, which in turn limits individual migration op- portunities. In addition, labor contracts may be canceled prematurely, creating income and employment risks for migrants. Migration choices are further influenced by pre-migration employment outcomes, migration fees and other costs, access to credit, and the costs and returns of entrepreneurship. For instance, the duration of a worker’s migration is affected by migration costs and the extent to which these are covered on credit, but also by the level of assets required to start better-paying self-employment activities in Bangladesh. In turn, credit conditions influence a worker’s ability to access self-employment in Bangladesh, but also the incentive to migrate, and the amount of savings repatriated. Thus, cost-benefit anal- yses of implemented policies must account for potentially offsetting changes in migration, savings and investment behavior. Our main data source is a new and unique survey of returning migrants that was specifi- cally developed to answer these questions. The Bangladesh Return Migrant Survey (BRMS) was designed with the idea that temporary migration is an integral part of workers’ life-cycle career. Hence, it provides very detailed information on the entire employment and migration history of a sample of 5,000 migrant workers who had returned to Bangladesh at the time of the data collection. The survey is one of the very few comprehensive surveys on temporary migrants globally, and to the best of our knowledge, it is the first of its kind to have been conducted in a country in which emigration is based almost exclusively on regular tempo- rary work contracts.3 The dataset includes detailed information on migrants’ personal and family backgrounds, their labor market outcomes before migration, their expectations prior to departure about earning and saving prospects abroad, migration expenditures, sources of income and savings, employment histories at the foreign destination, and labor market activities and earnings after their return to Bangladesh. 3 For less regulated contexts, in which emigration is more often undocumented, the Mexican Migration Project (MMP) and the Egypt Labor Market Survey (ELMS) have a similar design. While the Egyptian survey covers all workers in the country and is not specifically focused on temporary migrants, it includes a module on past migration of household members currently living in the Arab Republic of Egypt. 3 We complement the data from the BRMS with nationally representative household and firm surveys. Our modeling of both emigration and return choices accounts for selection into the sample of return migrants. The Bangladeshi Household Income and Expenditure Survey (HIES), which provides information on both non-migrants and current migrants, allows us to model the initial emigration decision. We make labor demand an explicit part of our model to account for the fact that emigration is restricted by the requirement to obtain a work visa. We document and utilize the fact that the volume of Bangladeshi emigration strongly depends on oil prices.4 All major destination countries for Bangladeshi migrants are oil exporters, and we show that migrant numbers track fuel revenues very closely. Subject to the probability of obtaining a work visa and to not being dismissed in the destination country, migrant supply depends on individuals’ unobserved location preferences, which govern emigration and migration duration choices. We identify these preference parameters by fitting our structural model to data on both emigration rates and the time migrants spend abroad. Our original data and analysis provide novel and direct evidence on the linkages between self-employment and migration. Comparing (within individual) activities before and after migration, we document a strong shift into self-employment upon return. Our survey directly asks how self-employment has been financed. Around half of the return migrants in our sam- ple report that savings from abroad were their main source of financing. Migration-related expenses in Bangladesh, while typically lower than the capital needed for self-employment, are nevertheless substantial. We find that while low-skilled workers can often borrow to pay for migration expenses, they can rarely borrow to finance self-employment activities. One explanation is that migrant workers hold a formal job contract that makes loan repay- ment more secure and shirking more difficult compared to a loan repayment that depends on entrepreneurial profits. The credit market can be thought of as segmented, with a clear distinction between migration loans and entrepreneurship loans, with higher or even pro- hibitive interest rates for the loans supporting entrepreneurship. We document a causal, positive effect of earnings accumulated abroad and duration of migration on the probability of self-employment after return. Why is self-employment, then, so attractive for low-skilled workers? Even though it has been argued that self-employment is a last resort for workers in some developing-country contexts, evidence from Bangladesh indicates that the vast majority of the self-employed have voluntarily chosen this activity (Gutierrez, Kumar, Mahmud, Munshi, and Nataraj 2019). 4 The positive association between oil prices and emigration to the six member countries of the Cooper- ation Council of the Arab States of the Gulf (also known as the Gulf Cooperation Council, GCC) has also been documented by Wahba (2015) for Egyptian migrants. 4 While informal wage employment tends to be readily available for workers, self-employment often generates higher income. However, it also requires accumulating sufficient assets in the presence of tight credit constraints.5 To motivate our focus on self-employment as an objective of many migrants from Bangladesh, we also show that alternative mechanisms, such as asset accumulation for marriage, are not supported by our data. We use the estimated structural model to evaluate the effects of three core policy param- eters on temporary migration and self-employment outcomes. First, we consider a subsidy that lowers the cost of migration. In our dynamic model, this policy intervention not only raises the prevalence of emigration, but also shifts it to younger ages, and reduces the time migrants spend abroad. On the one hand, an earlier return implies lower savings. On the other, conditional on the level of repatriated assets, emigration at a younger age and with an earlier return implies a longer payoff period that makes entrepreneurial investments more profitable for younger people. We find that the net effect of these mechanisms is a sizable increase in overall repatriated savings and business creation. Second, we examine a loan subsidy that reduces the interest rate for entrepreneurial loans. While this naturally makes self-employment a more attractive and accessible option, the policy’s effect is undermined by an endogenous adjustment in migration behavior. Specifically, we estimate that cutting the current entrepreneurial lending rate in half would lower the emigration rate by 5.4 percent, and reduce the average migration duration by 6.4 percent. Taken together, these imply a decrease of 8.4 percent in asset repatriation, limiting the positive effect of lower interest rates for entrepreneurial loans on business creation. Finally, we investigate an information treatment that aligns individuals’ expectations about their foreign earnings potential with the average earnings for different migrant groups observed in our data. Our model predicts that information about the true expected wages paid at a destination decreases emigration by a sizable 13 percent. An implication from this finding is that the interests of individu- als and policy makers may not align in such a situation; individuals naturally benefit from making from better-informed decisions, but policy makers aiming to boost business creation may, in fact, favor a situation in which individuals overestimate foreign wages because it encourages emigration and asset formation. In sum, our joint modeling of migration and entrepreneurship decisions highlights the importance of these endogenous responses, which need to be accounted for in the cost-benefit analyses of different policy interventions. The conditions that characterize the migration experience for workers in Bangladesh differ markedly from those that characterize migration from Mexico to the United States, and from Africa and the Middle East to Europe – the regions that are the focus of much of the existing 5 Self-employment options in our context primarily consist of relatively simple occupations such as owning a small store, driving a taxi, or running a small business. 5 research. Migration from Bangladesh is almost exclusively regular and temporary, mediated by officially registered agencies. Entry into the main destination countries in the Persian Gulf and Southeast Asia is strictly conditional on holding an employment contract. Due to tight enforcement and severe punishments, overstays and undocumented stays are very rare.6 If an employment contract is renewed, migrants may extend their stay, but, otherwise they must return home. Temporary migrants from Bangladesh, like most migrants from other countries in South and Southeast Asia, also pay high upfront fees to the intermediary agents who match workers with employers in destination countries. We show in this paper that these costs play a critical role in the timing and duration of emigration, and, subsequently, in the self-employment patterns in Bangladesh. The relationship between migration and self-employment under credit constraints has been investigated by Rapoport (2002) and Djaji´ c (2010) at a theoretical level. We build on their insights, and extend the empirical literature on temporary migration and entrepreneur- ship in several ways. First, we model emigration, the destination choice, and return migra- tion jointly with the self-employment decision, accounting for financial constraints. Among the earlier and influential papers that link migration to entrepreneurship, McCormick and Wahba (2001) focus on emigration from the Arab Republic of Egypt, while Dustmann and Kirchkamp (2002) focus on migrants who returned from Germany to Turkey. Piracha and Vadean (2010) investigate the relation for Albania; Wahba and Zenou (2012) and Mah´ e (2019) again for Egypt; and Br¨ uck, Mah´e, and Naud´ e (2018) for the Kyrgyz Republic. Several papers aim at isolating the causal relationships between migration and financial investments. These include Mesnard (2004) for Tunisia; Yang (2006), Yang (2008), and Khanna, Theoharides, and Yang (2020) for the Philippines; Woodruff and Zenteno (2007) for Mexico; and Batista, McIndoe-Calder, and Vicente (2017) for Mozambique.7 Whereas most papers restrict attention to the link between return migration and entrepreneurship in the country of origin, Wahba (2015) accounts for the selection of migrants from the non- migrant population and for the selectivity of returns. We draw on the insights developed in this literature, and extend them. Our paper is the first to directly estimate a dynamic model of the relationships between migration and self-employment, and to explicitly account for credit constraints, migration costs, and the risks of lower-than-expected earnings abroad, as well as for the risk of exogenous contract termination. Methodologically, our paper contributes to the literature that uses dynamic life-cycle 6 Southeast Asian countries such as Singapore apply very strict deportation measures to address overstays. Among the GCC countries, overstays are punished by imprisonment in Saudi Arabia and the United Arab Emirates. 7 Also see Wahba (2014), Rapoport and Docquier (2006) and Naud´ e, Siegel, and Marchand (2017) for related surveys. 6 models to examine the determinants and effects of international migration. Previous papers that have estimated such models to study temporary migration include Kirdar (2012), Klee- mans (2015), Girsberger (2017), Lessem (2018), G¨ orlach (2020), and Adda, Dustmann, and G¨orlach (forthcoming). This structural economics literature so far has not examined the role of entrepreneurship and self-employment in migration decisions.8 Our context of tem- porary, formal migration that is strictly conditional on an employment contract differs from the contexts of prior work in this literature; nevertheless, this is the norm shared by many other sending countries in South and Southeast Asia. Our paper, therefore, provides new insights on the dynamics of migration of millions of low-skilled workers from similar lower- income origin countries where migration is highly regulated, and self-employment represents an attractive option over informal wage jobs. The paper is organized as follows: Section 2 provides the contextual background on tem- porary labor migration from Bangladesh. Section 3 presents the new dataset and descriptive statistics for our sample. Section 4 introduces a set of stylized facts that inform and motivate the dynamic model laid out in Section 5. Section 6 discusses estimation and identification of this model. Section 7 presents our results. Section 8 concludes. 2 Background and Context Bangladesh is one of the main countries of origin of low-skilled labor migrants. It ranked fifth worldwide, with an estimated 7.8 million workers working abroad in 2018, the year before our survey was conducted (Ahmed and Bossavie 2021). In relative terms, the incidence of migration among the working-age population is also high. As of 2017, about 13 percent of the total working-age male population of Bangladesh (ages 15-64) was currently or in the past employed overseas. Oil-exporting Gulf Cooperation Council (GCC) member countries are the main destinations for these migrants. More than 750,000 workers have emigrated an- nually in recent years, and over time the magnitude of migration outflows has risen steadily with the labor demand in the main destination countries. Though long-term trends show that emigration from Bangladesh is rising, there is noticeable year-to-year volatility. Figure 1 shows that these variations are closely tied to fluctuating oil prices. Fuel exports are the single-most important source of income and a major determinant of labor demand in the destination countries of most Bangladeshi migrants. Although emigration correlates with oil prices, wages do not (Figure 1). This indicates that labor demand poses the binding con- straint, whereas supply at prevailing wage differences between destinations and Bangladesh is rather elastic – a feature we exploit in our estimation of the model (see Section 5). 8 orlach (2016) for a survey of the structural literature on temporary migration. See Dustmann and G¨ 7 Temporary labor migration from Bangladesh is almost entirely low skilled, reflecting the average education level in the underlying working-age population (Barro and Lee 2013). The gender distribution among migrants is strongly skewed, with men representing the vast ma- jority of migrants.9 The prevalence of male labor migration is driven by the low labor-force participation of women in Bangladesh (Rahman and Islam 2013), combined with concentra- tion of foreign labor demand in brawn-based occupations.10 Figure 1: Migration from Bangladesh and global oil prices Source: Bangladesh Return Migrant Survey (BRMS) for the number of migrants who left overseas in a given year and wage levels overseas; OPEC Reference Basket for crude oil prices. Note: The crude oil price for a given year is the inflation-adjusted annual average price in US dollars. Wages abroad for a given year are calculated as the inflation-adjusted average wage of migrants overseas in that year. Due to labor market regulations and strict residency laws in the main destination coun- 9 The share of women among temporary migrants has increased in recent years, boosted by a 2015 bilateral agreement with Saudi Arabia. Still, over the period from 1991 to 2014, women represented 4 percent of total emigration from Bangladesh, according to the administrative data from the Bureau of Manpower, Employment and Training. 10 According to the 2016/2017 Bangladesh Labor Force Survey, the labor-force participation of women between the ages 15-64 was 35 percent, compared to 88 percent for men in the same age group. 8 tries (GCC members and Southeast Asian countries), emigration from Bangladesh is tempo- rary by design. Acquisition of citizenship or permanent residency in the destination countries is effectively impossible, irrespective of migrants’ occupation, education, nationality, or dura- tion of stay (Wahba 2015; Fargues 2011; Fargues and De Bel-Air 2015). Low-skilled migrants’ stay is strictly conditional on a valid employment contract. These contracts are typically of fixed duration and tied to a specific employer, but contract renewals are possible to ex- tend duration, again conditional on the continued agreement of the employer. The initial duration of the work permit for low-skilled migrants is one year in Malaysia, and two years in Singapore, and, in both countries, these can be renewed for a total duration of up to 10 years. There is no explicit cap on total duration of stay for migrants in the GCC countries. However, because their residence at the destination is conditional on holding a contract, migrants cannot stay after they have retired. Similarly, a job loss or the expiration of the contract means mandatory and automatic return to the home country. Furthermore, low- skilled migrants are almost never allowed to take their families with them. This generates additional costs to migrating, and further incentives to return. In addition to the high levels of international migration from Bangladesh, internal rural- urban migration is also widespread. Although transportation costs have been shown to be important (Bryan, Chowdhury, and Mobarak 2014), the financial constraints for overseas migrants are considerably tighter, and complemented by strictly enforced legal restrictions. 3 Data 3.1 The Bangladesh Return Migrant Survey (BRMS) The primary data source in this paper is a newly collected dataset with unique information on full employment histories, migration expenditures, expectations and demographic char- acteristics of migrants. The Bangladesh Return Migrant Survey (BRMS), conducted by the World Bank in 2019, consists of a sample of 5,000 temporary migrants who had returned from employment overseas to Bangladesh at the time of the survey. It is one of the largest datasets on temporary migration conducted to date and, to the best of our knowledge, it is the first of its kind in a country from where migration is almost exclusively based on tem- porary worker contracts. It also is the first comprehensive survey on temporary migrants who have returned to Bangladesh, one of the major migrant sending countries globally. The survey was designed for an analysis of the economic activity of recently returned, temporary migrants in rural and semi-urban areas of Bangladesh. It covers all districts in the country.11 11 Bangladesh consists of eight divisions which are divided into 64 administrative districts. 9 Eligibility for the BRMS was restricted to migrants who had returned from employment overseas to Bangladesh since 2010. This restriction was put in place to reduce possible issues that might arise regarding the accuracy of migrants’ recollection of past migration experiences, and, at the same time, to allow for some variability in the timing of return of the migrants. The sampling frame used for the survey was the Bangladesh Census of Population and Housing for 2011. The survey used a two-stage, stratified-cluster sample design: In the first sampling stage, a total of 250 unions throughout Bangladesh were randomly selected from the rural and semi-urban unions of the census. In the second sampling stage, households with returned migrants were randomly selected from a listing carried out in each selected union to identify households with return migrants. The dataset has several features that enable us to fill important data gaps identified in prior work. First, the survey was designed with the understanding that temporary migration decisions are part of a life-cycle-optimization process by the workers. As a result, the survey includes detailed retrospective questions on the entire employment histories of migrants both in Bangladesh (before and after the migration) and while they were abroad. The data allow us to construct the full employment trajectories of the workers in the sample with detailed information on each migration episode, including destination country, dates and duration of stay, labor market outcomes (such as wages and occupation), and the reasons for returning. The survey records the costs of each migration episode, finely disaggregated by cost categories and the source of financing. Such detailed data are rarely collected in household surveys that include migration modules.12 Those expenses play a critical role in migration decisions in the case of Bangladesh and other South Asian countries, where these costs are known to be very high relative to workers’ earnings (International Labour Organization 2015; Farole, Cho, Bossavie, and Aterido 2017).13 The survey asks workers about the expectations that they held prior to their migration experience regarding labor market outcomes in the destination countries, such as wages and savings. As a result, we can compare their expectations with actual outcomes. Borjas and Bratsberg (1996) postulate that lower-than-expected earnings may be an important reason for return migration from the United States, and McKenzie, Gibson, and Stillman (2013) provide empirical evidence on over-optimistic expectations for migrants from Tonga. Even 12 One exception is the Labor Market Panel Survey for Egypt, one of the few labor market surveys in sending countries that include a detailed module on past migration, and collects information on intermediary fees paid by migrants. Migration from Egypt, however, is distinct from the migration from Bangladesh in that the former includes a larger share of undocumented migrants, many of them headed to European countries. 13 In a recent study, Mobarak, Sharif, and Shrestha (2020) examine a government-mediated program between Bangladesh and Malaysia where the migration costs are significantly lower. The migrants are chosen via lottery which allows the authors to address the selection problems directly and show how migration improves the welfare of the families. The authors attribute the gains, at least in part, to the low fees. 10 though a person’s recall might be imperfect, our data are the first that enables to link pre- migration expectations to realized earnings on an individual level for temporary migrants. Finally, the survey collects additional information on migrants’ most-recent migration episode, such as remittances sent, monthly expenses and savings, professional and personal difficulties encountered, and overall impressions. The survey includes a detailed module on demographic characteristics, enterprises and assets for the household. 3.2 The Household Income and Expenditure Survey (HIES) Because the BRMS only covers households with return migrants, we complement it with a nationally representative household survey that samples both non-migrants and current international migrants. The Household Income and Expenditure Survey (HIES, 2016-2017 wave) was designed to be representative at the national and division levels. The HIES collects detailed data on each household member, including labor market outcomes such as employment status, earnings, and sector of employment. The data are comparable to the labor market information captured by the BRMS. Information on non-migrants and current international migrants is collected through an “absentee module” answered by the household members who stayed in Bangladesh. The HIES also includes information on the debt and assets of the household, as well as detailed expenditure information that allows us to calculate household consumption and savings rates. The Census of Population and Housing 2011 was used as the sampling frame for both the HIES and the BRMS. While the HIES covers both rural and urban areas, the BRMS sampling was designed to capture a representative sample of recently returned migrants in rural and semi-urban areas of Bangladesh. We thus restrict the HIES sample to rural and semi-urban areas in all of our comparisons. The HIES was carried out two years earlier than the BRMS.14 This is an advantage in our context as current migrants in 2016 are likely to be more comparable to the return migrants captured in 2019 by the BRMS. Indeed, among the BRMS respondents, the median returned migrant had been back in Bangladesh for two years. ¨ Ahmed, Ahmed, Bossavie, Ozden, and Wang (2020) show that the sample composition of temporary migrants in the HIES and the BRMS are very similar, and that the characteristics and destinations of migrants in the two surveys are also aligned with those in administrative data covering the entire population of temporary migrants from Bangladesh.15 14 The Bangladesh HIES was carried between April 2016 and March 2017. 15 The Bureau of Manpower, Employment and Training in Bangladesh publishes aggregate data on the number, composition and destinations of legal migrants from Bangladesh by year of departure. 11 Table 1: Summary statistics BRMS HIES Return migrants Non-migrants Current migrants Mean S.D. Mean S.D. Mean S.D. Socio-econmonic characteristics Male† 0.96 0.20 0.48 0.50 0.96 0.19 Rural/Semi-urban 0.99 0.10 0.95 0.22 0.99 0.11 Age currently 37.9 8.2 35.3 11.4 34.1 8.8 Age category 18-24 0.03 0.17 0.21 0.41 0.14 0.34 25-34 0.35 0.48 0.28 0.45 0.39 0.49 35-44 0.39 0.49 0.25 0.43 0.32 0.47 45-54 0.20 0.40 0.19 0.39 0.13 0.34 55-59 0.03 0.17 0.07 0.25 0.02 0.14 Years of schooling 6.6 4.0 5.6 5.0 7.3 3.6 Education level Illiterate 0.16 0.37 0.30 0.46 0.08 0.26 Some primary (1-5) 0.24 0.43 0.25 0.44 0.26 0.44 Some secondary (6-9) 0.36 0.48 0.21 0.41 0.38 0.48 Above some secondary (10-12) 0.22 0.41 0.17 0.37 0.25 0.43 Tertiary (16+) 0.02 0.15 0.06 0.24 0.03 0.16 Married currently 0.88 0.32 0.77 0.42 Married before migration 0.58 0.49 Migration costs Total costs, 2010 constant dollar 4,203 2,121 Costs by usage, 2010 constant dollar Intermediary fees 2,318 2,195 Visa/Passport 998 1,544 Government fees 51 224 Other costs 746 1,386 Fraction of migrants who borrowed 0.56 0.50 Share of costs financed by borrowing* 0.76 0.30 Stay abroad Destination country GCC** 0.74 0.44 0.62 0.49 Southeast Asia 0.17 0.38 0.21 0.41 Other country 0.08 0.28 0.17 0.38 Age at departure 28.5 7.2 28.1 8.1 Duration of stay at destination 6.5 5.6 6.3 5.2 Return earlier than planned 0.25 0.43 Annual income, 2010 constant dollar 5,299 4,703 After return/Currently Age at return 34.9 8.1 Years since return 2.7 2.6 Employment status currently*** Not working 0.17 0.38 0.16 0.36 Waged worker 0.29 0.45 0.56 0.50 Self-employed 0.54 0.50 0.28 0.45 Annual income, 2010 constant dollar 1,280 499 1,163 1,174 Observations 4,709 47,137 3,571 Note: Sample is restricted to males aged 18-59. †Here, the mean is the share of males in the full sample. *Conditional on having taken a loan to migrate. **For columns 3 to 6, the GCC does not include Bahrain, which is coded as “others” in the original HIES data. ***These statistics exclude individuals who have been back in Bangladesh for less than a year at the time of the survey. 12 3.3 Descriptive Statistics Table 1 reports the summary statistics of the BRMS sample of return migrants.16 A clear majority of migrants are men. Women represent only 4 percent of returnees in the survey sample; this level is consistent with statistics from administrative data from the Bureau of Manpower, Employment and Training (BMET). The low level of female migration can be explained by the low labor-force participation of women in Bangladesh, and by difficult work conditions in the destination countries (International Labour Organization 2015; Farole, Cho, Bossavie, and Aterido 2017). In addition, there is social pressure on women to stay behind because they bear household responsibilities, and because low-skilled temporary migrants are not allowed to take their families with them. Given the small share of female migrants in the sample, combined with the fact that post-migration patterns are likely to differ between genders, we restrict our working sample to men. Most temporary migrants in the sample have some secondary schooling, while 17 percent have never attended school and only 2 percent have some tertiary education. The average years of schooling in the male sample of return migrants is 6.5 years, which is low compared to average educational attainment across the world (Barro and Lee 2013). Returning migrants have a higher level of schooling than the non-migrant, male, working-age population (5.8 years), but their schooling levels are lower than those of migrants who are currently overseas (7.6 years) according to the HIES 2016. These differences are largely explained by the older age of returnees compared to current migrants and increasing education levels across cohorts. About 75 percent of temporary migrants in the sample have returned from GCC coun- tries. Saudi Arabia, with about 25 percent of migrants, is the largest destination and closely followed by the United Arab Emirates. Malaysia represents 13 percent of temporary mi- grants, and is the largest destination in Southeast Asia. Only a small minority of migrants in our sample were employed in other high-income countries with Singapore as the leading destination. Return migrants in the sample have been back, on average, for 2.7 years; with a median time since return of two years. The mean current age in the sample is 38, with average departure and return ages of 29 and 35, respectively. Therefore, most workers do not migrate overseas immediately upon entering the labor force, but often work for more than a decade before going abroad. We observe substantial variation in the duration of migration episodes in our sample (Table 1). The median and mean duration of stay in all destinations are 4.7 years and 6.5 years, respectively. Close to a quarter of migrants stayed abroad for less than two years, 16 For a detailed description of the BRMS and its sample characteristics, see Ahmed, Ahmed, Bossavie, ¨ Ozden, and Wang (2020). 13 while a similar proportion stayed for more than nine years. A quarter of the migrants in the sample returned earlier than they expected or before the end of their employment contract. This implies that, among those surveyed, many who returned came back involuntarily. Figure 2: Incidence of unplanned returns among all return migrants Source: Bangladesh Return Migrant Survey (BRMS). Note: Returns are categorized as unplanned if migrants report that their main reason for returning was being laid off by their employer, having visa issues, or being expelled from the destination country. Statistics are reported for the top five destinations of return migrants in the BRMS sample. As shown in Figure 2, there is a negative relationship between the share of unplanned returns and duration of stay overseas. Temporary migrants who stayed overseas for only a year or two often returned because they were laid off, or because they had Visa issues. The share of forced returns is about 40 percent among migrants who returned within a year overseas, compared to less than 20 percent among those who returned after eight years or more. This pattern holds across different destinations. Repeated temporary migration, as captured at the time of the survey, is uncommon. In the surveyed population, 97 percent of returning migrants had made one trip abroad, excluding short visits during holidays. We should note that long migration episodes include multiple contracts, but there are few 14 interruptions and returns in between. Relatively high fixed costs of migration are likely to be the driving factor behind the limited incidence of repeated migration. Figure 3: Temporary migration costs and annual household income at the time of departure Source: Bangladesh Return Migrant Survey (BRMS). Note: Annual household income before migration and migration costs are denoted in 2010 constant USD. The last bar in each distribution includes annual household incomes or migration costs above $10,000. Temporary migration represents a considerable investment for workers. Total expenses for any migration episode are quite large compared to earnings in Bangladesh or even overseas (Figure 3). The median total cost reported by return migrants is approximately equal to one year of foreign earnings, three years of earnings of a wage worker in Bangladesh, or more than two years of household income. Intermediary fees are by far the largest item, representing 55 percent of total costs, followed by visa fees, which account for roughly 20 percent of costs. As a result of these large upfront costs, most migrants in the sample report that they had to borrow to finance their migration. 15 Figure 4: Earnings of temporary migrants in Bangladesh and abroad Source: Bangladesh Return Migrant Survey (BRMS). Note: Wages are annual and measured in 2010 constant USD. Statistics are for individuals who were employed before migration and after return to Bangladesh. The last bar of the distribution includes annual wages above $15,000. High migration expenses appear to be worthwhile for many workers; the median wage abroad is roughly three times workers’ pre-migration wages in Bangladesh (Figure 4). While almost all migrants are employed in low-skilled jobs, we observe differences in wage levels across different destinations. For instance, the median wage in Malaysia – the destination with the highest wages among the top five destinations – is about 45 percent higher than it is in Qatar. Finally, expectations of migrants about their own earnings potential abroad are even higher. As shown in Figure 5, prior to their departure, migrants systematically overestimate the wages they will earn abroad by a large margin. In our empirical analysis, we show how reducing this overestimation would affect migration choices and self-employment in Bangladesh. 16 Figure 5: Expected and actual earnings of temporary migrants overseas Source: Bangladesh Return Migrant Survey (BRMS). Note: Earnings are annual and measured in 2010 constant USD. Data on wage expectations prior to departure were collected retrospectively at the time of the survey. The last bar of each distribution includes annual wages above $15,000. 4 Stylized Facts on Temporary Migration and Entrepreneurship This section presents several stylized facts that motivate the analytical model, estimation and policy simulations. Stylized Fact 1: The rate of self-employment is significantly higher among returning migrants. Three main patterns emerge when we analyze the share of working-age men who are self- employed at a given age and by migration history. As Figure 6 shows, first, self-employment increases with age for both those who did and did not migrate. This pattern is compatible with the existence of credit constraints. Workers need to accumulate a certain level of assets 17 to cover the start-up expenses of self-employment and entrepreneurial activities. Second, prior to their move abroad, migrants have self-employment rates that are very similar to those of non-migrants across all ages (blue and red lines in Figure 6). These two groups of workers may differ in various dimensions, yet their propensity to become self-employed is almost identical until after the migration experience. Third, and most importantly for this paper, the rate of self-employment of return migrants is considerably higher than the rate among non-migrants, and the rate for the same migrants prior to their migration experience. This difference emerges at any age (green line versus the red and blue lines). The gap in self- employment rates between return migrants and non-migrants is largest at younger ages.17 For instance, 65 percent of 30-year-old return migrants in our sample are self-employed, compared to 28 percent of workers of the same age who have not migrated. While this difference has been pointed out in different contexts, our data facilitate a comparison not only with non- migrants, but also with migrants’ own activity prior to migration. This within-individual comparison strengthens the case that migration is instrumental in business creation in the origin country. Across all ages, the rates of self-employment for return migrants are high, reaching levels of between 60 and 70 percent. The gap in self-employment between those who did and did not migrate narrows with age, but it never closes. Data from the BRMS support the pattern. They show that over 90 percent of the enterprises owned by former migrants were established after their return from overseas work. In other words, self-employment patterns are not driven by other forces, such as returning migrants taking over the family businesses. This descriptive evidence highlights the role played by temporary migration in accelerating transitions of workers into self-employment. Faster asset accumulation is the main channel behind this observation (see Section 4 for detail). While self-employment can be a last resort for workers in certain contexts (Gindling and Newhouse 2014), evidence from Bangladesh indicates that the vast majority of the self- employed choose this type of work for other reasons. Gutierrez, Kumar, Mahmud, Munshi, and Nataraj (2019) report that 82 percent of the self-employed state that the ability to work independently and the possibility of earning higher incomes are the main reasons behind their decisions. The same study also finds that the self-employed stay in the same activity much longer than casual laborers and wage employees in the private sector. Self-employment thus appears to be an “absorbing state” for many workers in Bangladesh. 17 Figure 6 includes both dependent and self-employed workers in agriculture. Excluding that sector slightly reduces the self-employment rate but otherwise the pattern carries over. 18 Figure 6: Share of self-employment among the employed working-age population Source: Bangladesh Return Migrant Survey (BRMS) for migrants; Household Income and Expenditure Survey (HIES) for non-migrants. Note: Statistics are for males aged 20-55. To smooth the curve, each point is the mean of a two-year age cohort. Figure 7 shows that the self-employed earn higher incomes than wage workers, at any given age. Median monthly earnings of migrants after their return is 168 USD for en- trepreneurs with paid employees, and 101 USD for self-employed individuals with no other paid employees. By contrast, median earnings of return migrants who work as casual labor- ers is 83 USD.18 These earnings patterns by employment status are consistent with those of workers in Bangladesh who did not migrate, as shown in the Household Income and Expendi- ture Survey (HIES) and in Gutierrez, Kumar, Mahmud, Munshi, and Nataraj (2019). Figure 7 highlights another important distinction: Income levels of wage workers start to decline after age 45, presumably due to the physically demanding nature of many of the low-skilled jobs. By contrast, self-employed workers can sustain both their labor market participation rates and higher income levels for much longer, until age 65 in most cases (as reported in the HIES). These patterns are strongly indicative of the attractiveness of self-employment compared to wage employment in maximizing lifetime welfare.19 18 Earnings are adjusted to 2010 constant dollars. 19 Consistent with this evidence, analyses of a sample of workers from seven developing economies under- 19 Figure 7: Earnings of workers after return, by type of employment Source: Bangladesh Return Migrant Survey (BRMS). Note: Earnings are annual and in 2010 constant USD. The curves report average unconditional monthly wages after applying a local polynomial smooth. Due to the small number of observations in the tails, the sample is restricted to males aged 25-56. Stylized Fact 2: Entrepreneurs face tight credit constraints, while migrants are able to borrow to finance migration expenses. As in many other developing countries, individuals in Bangladesh who want to start a busi- ness need initial capital, but face credit constraints.20 Although self-employment appears to be the preferred employment option among many low-skilled workers in Bangladesh, most in- dividuals enter it at a relatively later age. In the HIES data, the median age of self-employed individuals in Bangladesh is 42, compared to 33 for wage employees, and 35 for daily labor- ers. In addition, the share of self-employed workers increases steadily with age, from about 20 percent at age 20 to close to 50 percent at age 55 (Figure 6).21 By contrast, Figure 6 taken by Giambra and McKenzie (2021) find that self-employed individuals are less likely to emigrate. 20 For cross-country evidence on credit constraints to self-employment in developing economies, see Beck (2007). 21 Using a large sample of workers in developing countries around the world, Gindling and Newhouse (2014) also find evidence that the rate of self-employment in developing economies increases with age, until workers reach the mid-forties, when the trend flattens out. 20 above shows a high self-employment rate among returnees, even at younger ages. These patterns are consistent with the self-employed needing time to accumulate savings, and sug- gest the existence of credit constraints preventing individuals from starting self-employment activities at a younger age. Figure 8: Primary source of start-up capital for self-employment Source: Bangladesh Return Migrant Survey (BRMS) Note: For each non-agricultural enterprise owned by the household, a return migrant chooses two sources in response to the question, “What were the main sources of start-up capital for this enterprise?” This graph shows the distribution of the primary sources listed by respondents. Several other data sources provide more direct evidence of credit constraints faced by individuals seeking self-employment. According to the 2010 Bangladesh Informal Firms Survey by the World Bank, the average start-up cost of a self-employment activity represents about two-and-a-half years of the average household’s income. The 2010 Survey of Firms in Bangladesh reports that only 10 percent of current employers funded their start-up capital through bank loans, a finding also supported by Mahmud (2006). For the specific population of returned migrants studied in our sample, 70 percent of self-employed individuals used their own savings – primarily savings accumulated while working abroad – as the main source of financing (Figure 8). By contrast, 19 percent of self-employed returned migrants in the 21 BRMS sample report using loans from private lenders as their primary source of capital. There is not much variation in these shares across education groups. While there is strong evidence of severe credit constraints faced by start-up and self- employment activities in Bangladesh, migrants often use credit to pay for their upfront, migration-related expenses and fees. Figure 9 shows that around 80 percent of the migrants in our sample used loans to finance more than half of their migration expenditures. Among those who borrowed, more than 35 percent of them covered all of their migration costs through a loan. On average, among the migrants who borrowed to finance their expenses, 60 percent of the total costs were covered by a loan. Figure 9: Distribution of the share of total migration costs financed by a loan Source: Bangladesh Return Migrant Survey (BRMS). Note: The graph shows the distribution of the loan-to-migration-cost ratio, conditional on borrowing. The last bar to the right of the distribution captures loan amounts that are more than over twice the migration costs, which represent less than 1% of the number of observations in the sample. There are several reasons why workers are able to borrow for migration, but not for self-employment. First, it is significantly less risky for lenders to finance migration expenses than to finance a start-up. As we have previously discussed, low-skilled workers going to the Persian Gulf or Southeast Asian countries cannot migrate without a valid contract that 22 specifies a wage and initial duration for their employment. Although workers face some uncertainty as to the monetary wage actually paid (as opposed for instance to benefits in kind that may be subtracted), such contracts send a strong signal to lenders that the migrant will have the requisite income to pay back the loan. In addition, migrants cannot settle permanently in these countries, and they almost never migrate with their families – effectively guaranteeing that they will return home. By contrast, earnings from self- employment are more uncertain and riskier. The likelihood of defaulting on a loan to finance a new business is therefore higher than for a loan to finance migration. Second, the agency problem faced by lenders is more pronounced in the case of entrepreneurship. Migrants’ earnings abroad are easy to verify and quantify because of the formal nature of migration arrangements. By contrast, it is difficult for lenders to verify self-employment earnings, which tend to have a high informal share due to the nature of these ventures (such as small shops). The markets for migration loans and entrepreneurship loans can thus be thought of as two separate credit markets, in which interest rates on entrepreneurship loans are significantly higher, and possibly prohibitive. Stylized Fact 3: Duration of stay overseas increases with migration costs and wages. There is a strong, positive association between migration costs, wages abroad, and the dura- tion of stay at the destination country. Figure 10 shows that, first, migrants who pay higher upfront migration expenses tend to stay longer at the destination. This finding is intuitive if workers are migrating with the aim of achieving a minimum level of net savings during the migration episode. Holding everything else equal, an increase in migration costs raises the length of stay required to recover the upfront expenses and to reach the desired level of savings. Second, migrants who earn higher wages at the destination country also stay there longer. The literature on the relationship between migrants’ wages and duration of stay suggests that this relationship could go in either direction. On the one hand, higher wages can increase incentives to stay longer to maximize lifetime earnings. On the other hand, migrants with a target level of savings can reach this target earlier when they earn higher wages. Figure 10 indicates that the first channel tends to prevail, particularly when upfront costs are high. We do not find strong evidence in the data that the risk of forced return due to visa issues or contract termination decreases for migrants who earn higher wages. Nevertheless, we account in the model for the possibility that the risk of forced return differs for higher- skilled migrants. 23 Figure 10: Duration of stay abroad, as a function of migration costs and wages overseas Source: Bangladesh Return Migrant Survey (BRMS). Note: Wages are grouped into four quartiles according to monthly wages abroad in 2010 PPP-adjusted USD. Migration cost quartiles are based on total migration costs in 2010 PPP-adjusted USD. Stylized Fact 4: Higher earnings abroad increase the likelihood of self-employment after return. The likelihood that a return migrant becomes self-employed increases with his net cumulative earnings abroad. These earnings are a function of migration duration, migration costs, and wages abroad. Figure 8 indicates that earnings abroad are the main source of the investment capital for workers who became self-employed after their return. In line with this, Figure 11 shows that both monthly wages abroad and duration of stay increase the probability of self-employment after return. 24 Figure 11: Share of self-employment after return by duration of stay and monthly wage abroad Source: Bangladesh Return Migrant Survey (BRMS). Note: The sample is restricted to return migrants who were employed at the time of the survey and had been back to Bangladesh for at least a year. Wages are grouped into four quartiles according to the monthly wages abroad in 2010 PPP-adjusted USD. To complement this descriptive evidence, we regress self-employment status after return on workers’ total earnings overseas, controlling for observable factors. Formally, the equation we estimate is the following: Self Empidot = α + β ln(Cum.Earningidot ) + Xit θ + γo + δd + ηt + idot , (1) where the dependent variable Self Empidot is a binary variable taking the value one if in- dividual i from division o becomes self-employed after returning from country d in year t, and zero otherwise. Xit is a vector of control variables, including age and age squared at the time of the survey, a dummy for self-employment before migration, and education-level dum- mies. γo , δd and ηt represent origin-division, destination-country, and year-of-return effects, respectively. The explanatory variable of interest is the (log) of cumulative earnings abroad, 25 which are obtained by multiplying a worker’s monthly wage in the destination country in PPP Bangladeshi takas by the duration of stay abroad.22 Columns (1) and (4) of Table 2 report the estimates for the coefficient of interest, β , when Equation 1 is estimated using a simple OLS estimation. Both columns show that there exists a positive and statistically significant correlation between total earnings abroad and the likelihood of self-employment after return, both in the full sample and among working individuals. The OLS coefficients, however, may not be interpreted as causal because total earnings at destination are likely to be endogenous to the desire to become self-employed after return. In particular, the optimal migration duration and the occupational choice after return may be determined simultaneously (Dustmann and Kirchkamp 2002). To address this endogeneity, we instrument total earnings overseas by exploiting specific characteristics of the Bangladeshi migration context, which centers on providing labor for oil-exporting countries. Specifically, we use the growth rate in oil prices, interacted with the oil dependency of the GDP of the destination country. The oil-price growth rate is calculated as the ratio of oil prices in the year when the individual returned to Bangladesh over the price during the year when the individual left Bangladesh. The destination-specific, oil-GDP dependency is measured by the mean of the annual oil rent over GDP in a destination country for all years from 1996 to 2017. Formally, the instrumental variable for total earnings abroad, denoted Zidt , is constructed as: Zidt = OilGrowthit × OilGDPd (2) The intuition behind this instrumental-variable strategy is that exogenous fluctuations in oil prices affect migrants’ duration of stay by increasing demand for immigrant labor, and thus increasing the likelihood that their labor contracts will be extended. The power of the instrument thus resides in the fact that oil prices affect migrants’ total earnings through adjustments in the demand for labor, rather than the wages paid for labor. Indeed, we do not observe any visible association between overseas wages and oil prices; yet, demand for foreign labor appears very responsive to oil prices (Figure 1). Such adjustments through quantities as opposed to wages (of immigrant labor) in response to GDP shocks in destination countries has also been confirmed by McKenzie, Theoharides, and Yang (2014) for migration from the Philippines. Importantly, since we condition on full sets of year and destination effects, only the interaction of oil-price growth and the share of oil revenue in GDP is used for identification. 22 The PPP rate half way through the migration episode is used to convert foreign currency to Bangladeshi takas. 26 Table 2: Earnings abroad and self-employment after return, reduced-form estimation (1) (2) (3) (4) (5) (6) Full sample Working individuals only OLS 2SLS First Stage OLS 2SLS First Stage Dependent Variable Self-employed Self-employed ln(Cum.Earn.) Self-employed Self-employed ln(Cum.Earn.) ln(Cum. Earning abroad) 0.031*** 0.112*** 0.040*** 0.110** (0.008) (0.032) (0.009) (0.047) Oil price growth × Oil rents/GDP 0.855*** 0.824*** (0.119) (0.112) Other controls Yes Yes Yes Yes Yes Yes Year of return FE Yes Yes Yes Yes Yes Yes 27 Origin division FE Yes Yes Yes Yes Yes Yes Destination country FE Yes Yes Yes Yes Yes Yes F-statistics of excluded instruments 51.9 54.3 Observations 3354 3354 3354 2775 2775 2775 Note: * p<0.10, ** p<0.05, * p<0.01. Standard errors clustered at the destination-origin level are reported in parentheses. The sample is restricted to males between the ages 18 to 59, and excludes individuals who have been back in Bangladesh for less than a year. Control variables include age and squared age at the time of survey, educational attainment dummies, and a dummy for self-employment prior to migration. Oil-price growth is the ratio of the oil price at the time of return to Bangladesh over the one at the time of departure from Bangladesh. One possible identification issue regarding the exclusion restriction may be that fluc- tuations in oil prices directly affect the relative attractiveness of wage and self-employment activities in the home country (El-Mallakh and Wahba 2021). This might occur, for instance, when wage and self-employment work are concentrated in different sectors in Bangladesh. Two arguments mitigate this concern: First, our estimation conditions on a full set of year ef- fects, so that only the interaction with destination country-specific oil shares in GDP provide the identifying variation. Second, in our sample, the majority (two-thirds) of returned migrants living in rural and semi-urban areas work either in small-scale agriculture or retail businesses. According to the 2016 industry-level Input-Output tables for Bangladesh, both sectors have a very small share of their input values coming from “Coke or refined Petroleum”.23 We do, however, observe a slightly higher importance of the transportation sector for the self-employed (21 percent) compared to wage workers (15 percent) in the sample. Since transportation is more dependent on oil-derived inputs compared to other sectors, we also run our IV estimation on a sample that excludes people employed in transportation. The results of this robustness check are reported in Table A2, and show that the IV coefficients are virtually identical when the transportation sector is excluded. We next investigate changes in the share of self-employment among non-migrants over time, and a possible association with fluctuations in oil prices. We explore these potential connections to provide further evidence that oil prices do not directly affect the relative attractiveness of self-employment and wage work. We use several waves of the Bangladesh Labor Force Survey, which is nationally representative and available at several points in time between 2000 and 2017. We focus on workers who have not migrated overseas. As shown in Figure A1, the fraction of those who did not migrate and who are self-employed in rural and semi-urban areas (the areas covered by the BRMS) is fairly stable over time. In short, we do not observe any noticeable association between the share of self-employment in Bangladesh and large fluctuations in oil prices over time. As the instrument primarily operates through its effect on the duration of stay abroad, another potential threat is that duration of stay may affect self-employment odds after return in ways that extend beyond effects on total earnings abroad. This could occur, for example, if a longer duration abroad reduces the likelihood of finding a wage job back in Bangladesh, due to a loss of social networks and contacts with local employers. However, as we have previously discussed, self-employment is typically preferred by workers in Bangladesh. Therefore, it is unlikely that workers who spent a longer time abroad would turn to self-employment because they encounter more frictions in finding a wage job. To provide further evidence on this, 23 The ratio is around 1.2 percent in agriculture and 0.2 percent in retail. 28 we also check whether there are systematic differences in employment spells according to the duration of stay abroad of return migrants. As shown in Figure A3, we do not find any systematic association between duration of stay overseas and duration of unemployment after return. These results do not seem compatible with the hypothesis that return migrants who stayed overseas longer are more likely to opt for self-employment because they cannot access wage employment back home. Columns (2) and (5) of Table 2 display our estimation results using this instrumental- variable approach. The first-stage coefficient of the oil-price instrument reported in columns (3) and (6) has the expected sign: A positive growth rate in oil prices during the migrant’s stay abroad increases the total value of earnings overseas in local currency. The coefficient is statistically significant at the one percent level, and the F-statistic for the excluded in- strument suggests a strong first stage. The estimated effect of cumulative earnings overseas on the likelihood to become self-employed after return is sizable, with a semi-elasticity of around 0.1. As shown in Table A2, these findings are virtually unchanged when we exclude return migrants who are employed in the transportation sector, which uses a higher fraction of oil-derived inputs than other sectors. These stylized facts jointly suggest that a key driver of temporary migration from Bangladesh is the desire to work overseas to accumulate sufficient financial savings to fund self-employment upon return. In other contexts, human capital accumulation abroad has been pointed out as an important factor for temporary migration and post-return decisions (McCormick and Wahba 2001). Consistent with this, a wage premium for returning migrants has been esti- mated, for instance, for migrants from Hungary (Gang and Yun 2000), Albania (De Coulon and Piracha 2005), West-Africa (De Vreyer, Gubert, and Robilliard 2010), Mexico (Lacuesta 2010; Reinhold and Thom 2013), Romania (Ambrosini, Mayr, Peri, and Radu 2015) and Egypt (Marchetta 2012; Wahba 2015; El-Mallakh and Wahba 2017; El-Mallakh and Wahba 2021). In the context of migration from Bangladesh, however, human capital accumulation is unlikely to be the main motivation. The vast majority of migrants from Bangladesh are em- ployed in low-skilled jobs in the destination countries. These occupations generally rely on physical strength, providing limited opportunity for learning and human capital accumula- tion. In addition, the occupational patterns of individuals before, during, and after migration do not seem to be compatible with the human-capital accumulation hypothesis (Table A3). For example, while two-thirds of temporary migrants in the sample were employed in the construction sector at their destination abroad, but only 10 percent after returning home (Panel A). In the HIES data, this ratio is only slightly lower (7 percent) for those who did not migrate. In addition, 24 percent of the returning migrants were employed in construction 29 prior to their migration, suggesting that experience accumulation occurs beforehand. The ratio of migrants employed in construction, however, is significantly lower after return. Like- wise, only 14 percent of return migrants had worked in retail or agriculture while overseas; by contrast, more than two-thirds of migrants were employed in one of these two sectors when they returned to Bangladesh. Furthermore, migrants who were employed in the construc- tion sector overseas do not transition into that same sector at significantly higher rates after return, compared to individuals who were employed in transport and utility sectors (panel B of Table A3). Similarly, temporary migrants employed in transport and utility sectors while abroad do not disproportionately transition into that sector after return. Overall, descrip- tive evidence on sectoral patterns lends little supports for human-capital accumulation as a driver of temporary migration patterns from Bangladesh. Another possible motivation for temporary migration from low-income countries is the improvement of marriage market prospects through asset accumulation overseas. The mar- riage and migration patterns of temporary migrants in the BRMS sample, however, do not seem compatible with this hypothesis. Appendix Figure A2 depicts the shares of married individuals before and after migration, at a given age. If migrants were moving overseas to improve their marriage prospects, one would expect the share of those temporary migrants who are married after migration to be higher than the share who are married before depar- ture. However, as shown in Figure A2, the share of individuals who are married before and after migration, at any given age, is virtually the same.24 This descriptive evidence does not support the idea that improvement of marital prospects is a major mechanism driving temporary migration decisions. 24 Unreported results show that the same holds when we disaggregate the sample by education groups. These results are available upon request. 30 5 Model Motivated by the above evidence, this section presents a dynamic model that captures em- igration, return migration, and wage and self-employment decisions of temporary migrant workers across multiple destinations and time periods. Each individual i in the model is identified by his education level ei and age ait at time t.25 While education is fixed over the lifetime of the individual, age increases by one in each time period (year). Every in- dividual makes several interrelated decisions in each period t. First, he chooses a location lit ∈ {B, M, O, Q, SA, U AE } to live and work in that period. The individual has the option of staying in Bangladesh (B ), or choosing to work in one of five foreign destinations. These are, in alphabetical order, Malaysia (M ), Oman (O), Qatar (Q), Saudi Arabia (SA) and the United Arab Emirates (U AE ), and together account for the destinations of 82 percent of all migrant workers in our sample. Modeling the choice between alternative destinations is important because, for example, when conditions in one country deteriorate, migrant flows divert elsewhere (Bertoli, Moraga Fern´ andez-Huertas, and Ortega 2013). If the individual migrant already resides in a foreign destination, d ∈ {M, O, Q, SA, U AE }, he chooses whether to extend the stay for another period at the same location, or to return to Bangladesh.26 As it is rare among Bangladeshi migrants workers, we rule out the possibility of moving directly from one foreign destination to another. The second critical decision of the individual is about his employment type, denoted by sit ∈ {S, W } where the option S indicates self-employment, and W indicates wage employ- ment. Emigration, return or self-employment decisions have a financial component. If a wage worker in Bangladesh decides to emigrate to a foreign destination, or switch to self- employment, then he has to undertake an investment which needs to be financed via personal savings, borrowing, or a combination of both. If the individual stays in his current location and employment, he continues to allocate that period’s income between savings and con- sumption, which, in turn, determine his stock of assets and future investment options. We now describe each of these decisions and variables in more detail. 5.1 Employment, Assets, Costs and Constraints l l,W Wage employment in any given location l yields labor income wit = wl (ei , ait , αit ) which l,W is a function of the education level ei , current age ait , and unobserved random factors αit ∼ 25 We use male pronouns to refer to the migrants because more than 95 percent of those who work abroad are men, as seen in Table 1. 26 We use the phrase “destination” and superscript d to refer to the five foreign countries that host a large majority of the migrants. The phrase “location” and the superscript l refer to any one of these countries, as well as Bangladesh. In other words d ∈ {M, O, Q, SA, U AE } and l ∈ {B, M, O, Q, SA, U AE }. 31 2 N (0, σl,W ). When working in one of the foreign destinations, d ∈ {M, O, Q, SA, U AE }, migrant workers face an exogenous risk that their contract will be terminated, in which case, they will be forced to return to wage employment in Bangladesh. The probability of d termination and forced return, δit = δ d (ei , ait , yit , f uelRevdt ) is specific to the destination and, again, depends on an individual’s age and education level, as well as on the time since migration to the destination country, denoted by yit , and that destination’s revenues from fuel exports. In parallel, wage workers in Bangladesh also face a risk of unemployment, B which we denote by δit = δ B (ei , ait ). Migration duration naturally does not enter into this expression since the individual is at home. Individuals are assumed to be risk-neutral, utility maximizers, and we adjust wages in Bangladesh by these employment probabilities. In principle, wages in both Bangladesh and foreign destinations are equilibrium outcomes. Yet, Figure 1 shows that the real wages paid to Bangladeshi migrants in our sample vary neither with international oil prices nor with the scale of migration, possibly due to highly elastic immigrant labor supply. Instead, wages are quite stable over the time period we consider. As such, we can rule out any major effects of migration on wages in destination countries. Within Bangladesh itself, migrants are a much smaller fraction of the workforce than in destination countries. Furthermore, estimates in the literature suggest that the wage effects of emigration are generally small. Dustmann, Frattini, and Rosso (2015), for instance, estimate a wage elasticity of 0.02 for the effect of emigration on wages in Poland.27 Thus, we also disregard equilibrium wage effects in Bangladesh, which we believe to be of second order relative to the mechanisms investigated in the paper. Our data suggest that potential migrants tend to overestimate their future, foreign wages prior to their emigration. In line with this observation, we assume the emigration decision is d based on destination-specific expected wages, denoted by w ˜it = E[w˜ d |ei , ait ]. However, once they are abroad and start to work, migrants realize their actual wages for period t, denoted d as wit . All future decisions are then based on the latter. Self-employment is an option for wage workers who are in Bangladesh and it requires I an initial investment Ce . We let this investment vary across education groups to account for the heterogeneity of businesses operated by individuals with different education levels. S In return, self-employment generates profits πit = π (ei , ait , αit ) per period, which again vary with the education level and the age of the individual, as well as with unobserved factors S 2 given by αit ∼ N (0, σS ). Because the overwhelming majority of migrants in our survey are 27 Overall, however, there is little consensus on the size of the effect of emigration on wages, with estimates ranging from small, positive effects when estimated for different member countries of the Organisation for Economic Co-operation and Development (OECD) by Docquier, Ozden, and Peri (2014) to negative effects estimated by Aydemir and Borjas (2007). 32 contract wage workers, we assume that the self-employment option is available only when workers are in Bangladesh. Migration costs and labor demand at destination are critical in shaping emigration decisions and their timing. The main reason for emigration for most migrants is the ability to earn higher wages. However, emigration requires an upfront payment of fees and expenses, d which we denote by Cit = C d (ei , ait ). The amount is specific to destination country d ∈ {M, O, Q, SA, U AE } and dependent on the individual’s age and education level. While they d are abroad, migrants further suffer from an education and destination-specific disutility ηe in each period, arising from their separation from family and friends who stay in the home country. Migration is also constrained by skill-specific labor demand in destination countries. As documented in Figure 1, aggregate migration is strongly related to the price of oil, which is the main source of revenue in all major destination countries. We account for this dependence by specifying a function that relates revenues from oil to the share of individuals with education-level e who can locate a job and obtain a work visa in destination d, conditional on having the desire and financial ability to move there. The probability λd et of securing a visa is related to the level of fuel revenues f uelRevdt in destination d and year t through the relationship ln(λd d et ) = φe + ψ ln(f uelRevdt) ). 28 Figure 1 also shows that the real wages paid to migrants do not track oil prices, but are roughly constant over the time period we consider, in spite of the strong variation in oil prices. Credit constraints and asset accumulation link self-employment and migration de- cisions. Both self-employment and emigration to a foreign country generate higher incomes (in expectation), but require upfront investments. Individuals can finance these investment amounts from their current savings, or through borrowing or a combination of the two op- tions. The level of private assets accumulated by an individual at time t is denoted by Ait , and yields real interest at a rate rA . Assets Ait can have a negative value, implying that the individual is in debt, in which case a higher lending rate rL > rA accrues. Many migrants finance their trip through loans, but not all individuals may have access to credit. We assume that an individual with education level e can cover migration fees and expenses through credit with probability pe . Without access to credit, migration costs have to be fully financed from individual savings. Hence, with probability 1 − pe , migration to destination d requires that Ait ≥ C d (ei , ait ). In the case of investment for self-employment, we 28 Fuel revenues are measured in USD, deflated to 2010 prices, and we restrict realizations of λd et to andez-Huertas Moraga, and [0, 1]. Importantly, as for the case of Filipino migrants analyzed by Bertoli, Fern´ Keita (2017), labor demand across the main destinations for Bangladeshi migrants is correlated, and our specification of λdet allows for this. 33 assume that a maximum of 50 percent of the investment costs can be covered via borrowing. This ratio is based on the guidelines of the Microcredit Regulatory Authority (MRA) in Bangladesh, and reflects the credit constraints faced by entrepreneurs as previously discussed (see also Battaglia, Gulesci, and Madestam 2021). Becoming self-employed requires that I Ait ≥ 0.5Ce . We observe the share q of individuals who inherited a business, and in the model let this fraction of individuals start their working life as entrepreneurs without having I to accumulate the upfront investment cost Ce . The savings rate is given by ρl , which is the share of income saved when the individual is in location l. In our calibration, we take this parameter directly from the data. Subject to the inequality constraints above, the stock of assets of an individual then accumulates according to the following equation: Ait+1 = (1 + r)Ait + ρl l i (1[sit = W ]wit + 1[sit = S ]πit ) − 1[lit−1 = B ∩ lit = d]C d (ei , ait ) I − 1[sit−1 = W ∩ sit = S ]Ce . (3) In this expression, ρl i is the location-specific savings rate, sit is the indicator variable on whether the individual is self-employed (sit = S ) or a wage worker (sit = W ), C d is I the destination-specific migration fee, and Ce is the education specific investment cost for self-employment. If his asset level is negative (because he is still repaying a loan), then the individual pays interest at a rate r = rL . If the individual has a positive level of assets, then he earns returns at the rate of r = rA < rL . Hence, an individual will use all of his savings before borrowing to finance self-employment or migration expenses. The expression 1[lit−1 = B ∩ lit = d] is an indicator variable that is equal to 1 if an individual is in Bangladesh in period t − 1 and migrates to destination country d at the beginning of period t. Therefore, the individual has to pay the migration fee C d (ei , ait ) for that destination. Similarly, the expression 1[sit−1 = W ∩ sit = S ] defines an indicator variable that is equal to one if an individual is a wage worker in period t − 1, becomes self-employed in period t, and has to pay I the self-employment investment cost Ce . Finally, we assume that individuals own an initial stock Ae,0 of assets, depending on their education level e at the beginning of their working life. 34 5.2 Welfare Utility in each period consists of three components: (i) Consumption, cit , is given by income minus savings. The income might come from wage employment abroad or in Bangladesh l (given by wit for location l), or from the profits from self-employment, πit . A share ρl i of this income is saved. (ii) Individuals experience an education- and location-specific disutility l from migration, ηe . This is equal to zero if the individual is living in Bangladesh, but may take other values if he is in a foreign country. (iii) There are transitory taste shocks εlit for a location l, as well as for sectoral choices (when in Bangladesh), denoted by εs it . Consumption is given by cit = (1 − ρl l l i )(1[sit = W ]wit + 1[sit = S ]πit ). (4) In this expression, 1[sit = S ] is an indicator variable that is equal to 1 if the individual is self-employed as explained in equation (3). 1[sit = W ] is the corresponding indicator variable for wage employment. With consumption defined as above, the flow utility for individual i in period t at location l is l uit = cit + ηe + εl s it + 1[lit = B ]εit , (5) where 1[lit = B ] indicates whether the individual is in Bangladesh. Wages earned in foreign destinations and the corresponding consumption are transformed accounting for price level differences between destinations and Bangladesh. This is in line with most families staying behind, and financing consumption through remittances sent by the migrant. It also implies that a higher purchasing power of the destination country’s currency in Bangladesh facilitates faster asset accumulation and impacts migration duration (Yang 2006; Yang 2008; Akay, Brausmann, Djaji´ c, and Kırdar forthcoming; Albert and Monras 2020). Individuals maximize their discounted expected lifetime utility over the sequences of location and self-employment choices, and subject to the evolution of the vector of state S W variables Ωit = {t, ei , ait , Ait , sit , lit , yit , αit , αit , εs l it , εit } as explained above. The maximized lifetime value at time t is, then, given by T Vit (Ωit ) = max β τ −t E[uiτ (Ωiτ )], (6) {l,s} τ =t where individuals live until period T . B,W Value functions depend on location and employment choices. We first define Wit , which denotes the value for a wage worker who is currently in Bangladesh and faces proba- 35 bility λd et that he can obtain a work visa for foreign destination d. All values in the section depend on the vector of state variables Ωit , which for better readability we omit from the notation. With probability pe , agents have access to a migrant loan, and their value includes the choice between all locations B,W B Wit = E max{ Vit , λM ˜M M B et Vit + (1 − λet )Vit , λO ˜O O B et Vit + (1 − λet )Vit , λQ ˜Q Q B et Vit + (1 − λet )Vit , λSA V˜ SA + (1 − λSA )V B , et it et it AE ˜ U AE λU et Vit + (1 − λU et AE B )Vit }. (7) In this expression, the first term is the value in Bangladesh. The subsequent terms are sums of values in each destination and Bangladesh, weighted by the probability of being able to migrate to that destination. If the agent has no credit access (with probability 1 − pe ), his value is given by (7) only if Ait ≥ C d (ei , ait ). Instead, when assets cannot cover the cost B,W B of migration, the agent cannot migrate, and the value is simply given by Wit = Vit . B The value in Bangladesh, Vit , depends on the decision between wage and self-employment, which are subject to the individual’s asset level. If an individual has enough assets Ait to I meet the investment cost, Ait ≥ 0.5Ce , he has the option to become self-employed. If he does not have enough assets, an individual will continue as a wage worker. The value of being in Bangladesh is, therefore, given by  E max{V B,W , V B,S } , if A ≥ 0.5C I B it it it e Vit = (8) V B,W I , if Ait < 0.5Ce , it B,W B,S where Vit and Vit denote values of wage and self-employment at home, respectively. The value attributed to wage employment in Bangladesh is, in turn, given by B,W B,W B,W Vit B = (1 − ρB )wit + βWit +1 + εit , (9) whereas, the value for self-employment is B,S B,S B,S Vit = (1 − ρB )πit + βVit +1 + εit . (10) We assume that self-employed individuals do not migrate (because they would have to 36 leave their businesses behind), and that they stay self-employed from that period forward. This is supported by our data, which show that only a small fraction of the respondents report being self-employed at the time of their migration. It also is consistent with findings for other developing countries, see Giambra and McKenzie (2021). Accordingly, the continuation value B,W in equation (9) is given by Wit +1 , since wage workers can change occupations and locations, and their continuation value involves this optimizing decision. By contrast, the continuation B,S value in equation (10) is given by Vit +1 , as individuals stay self-employed. The expected value before migration to destination d depends on (potentially bi- ased) wage expectations. This wage, which individuals expect to obtain if they were to move from Bangladesh to destination d at time t, is ˜it V d = (1 − ρd )w d ˜it d + ηe ˜ d + εd + βW (11) it+1 it d In this expression, as a reminder, w ˜it ˜ d |ei , ait ] is the wage that potential migrants = E[w d expect to earn before migration takes place, and ηe is the education specific disutility. Finally, ˜ d Wit+1 is the (expected) continuation value before the migrant arrives in destination d and observes the true wage. It is defined as ˜ it W d d = (1 − δit B,W ˜ d )E max{Vit d B,W , Vit } + δit Vit , (12) d where δit is the probability of losing the job at the destination and being forced to return to Bangladesh. Having defined all of its components, l∗ is the optimal location decision that B,W maximizes the expression in equation (7) and gives Wit . d The value after migration to destination d depends on the realized wage wit , and enters into the return migration decision. It is given by d Vit d = (1 − ρd )wit d + ηe d + βWit d +1 + εit , (13) where continuation values after migration to destination d (when the actual wage is observed) is given by d d B,W d d B,W Wit = (1 − δit )E max{Vit , Vit } + δit Vit . (14) 37 6 Identification and Estimation Several components of our model are directly observed in the data. These include the cost d l Cit of migration to destination d, earnings wit and profits πit by individual characteristics at location l, saving rates ρl , and earnings expected prior to emigration, w d ˜it . Furthermore, the survey contains information on the reason for return migration, including whether a work contract has been terminated. We use this information to compute the corresponding d probabilities δit within each group of migrants. Appendix B, and in particular Tables A4 and A5 provide further details. In the Bangladesh Informal Firms Survey 2010, we observe the share q of businesses that are inherited. Finally, rA and rL are, respectively, the interest rates earned on savings, or paid on migrant loans; these are obtained from Mallick (2012) and Berg, Emran, and Shilpi (2013). We estimate the remaining structural parameters of the model with by method of sim- ulated moments, minimizing the distance between informative moments computed for a population of agents simulated from the model, and the counterpart of these moments ob- served in the data. In total, we jointly estimate 52 parameters pertaining to the demand in each destination country for workers of a given education level (parameters determining the share λd et of would-be migrants who obtain a work visa), agents’ destination-education- d specific disutility from migration ηe and their initial stock of assets Ae,0 , the share pL e of I would-be migrants who can finance migration on credit, as well as the cost Ce of setting up a business. The latter three sets of parameters are education-specific, and identified by having the model match the observed asset level, the fraction of migrants who report hav- ing financed their migration on credit, and the self-employment share in our sample. The remaining parameters are identified by observed migration patterns. Note that the intensity and distribution of emigration from Bangladesh to different destination countries are affected both by labor demand λd d et and by agents’ preferences ηe . To disentangle the two effects, we target both emigration shares and migration durations conditional on having migrated to a d given destination. The disutility ηe from staying abroad determines both the emigration and the migration duration decision. Motivated by Figure 1, the share λd et primarily affects emi- gration. The termination or expiration of work permits is observed directly and accounted d for through δit . Building on the evidence reported in Figure 1, we let λd et vary with destination countries’ fuel revenues. We proceed under the following assumptions: (i) Fuel revenues are exogenous; (ii) The labor supply of Bangladeshi migrants always exceeds labor demand in destination countries (or is very elastic, as suggested by the wage trend depicted in Figure 1); (iii) The elasticity of the probability of locating a job in a destination with respect to fuel revenues 38 equals the elasticity of migration with respect to fuel revenues in a destination; that is, the relationship between fuel revenues and migrant demand can be identified as an estimate of ψ in an estimating equation ln(migrantsdt ) = φd + ψ ln(f uelRevenuesdt ) + udt . (15) We thus feed a regression estimate of ψ into the structural model, and estimate intercepts φd e for each migrant group by letting the model match the respective observed emigration rate. The vector (ψ, φM U AE 1 , ..., φ4 ) then parameterizes λd et , the probability of locating a foreign job. Estimating a life-cycle model requires an assumption on agents’ expectations about the paths for fuel revenues in destination countries over time. We fit quadratic trends to fuel revenues in the main destination countries over the 2000-2017 period, and assume that agents’ expectations are based on these trends. M M Taken together, We estimate the 52-element structural parameter vector θ ≡ (η1 , η2 , ..., O U AE η1 , ..., η4 , φM U AE 1 , ..., φ4 I , A1,0 , ..., A4,0 , C1 I , ..., C4 , p1 , ..., p4 ) of the model in Section 5 by us- ing method of simulated moments, which minimizes the distance between moments mm simulated from the model, and the corresponding data moments md from the HIES and BRMS samples.29 Specifically, we minimize the estimation criterion crit(θ ) = (md − mm (θ )) W (md − mm (θ )) , where W is a diagonal weighting matrix with the inverse variances of the data moments on the diagonal. The moments targeted are listed in Appendix Table A6, and Figure 12 shows the model’s good overall fit. The figure plots moments simulated by the model against their empirical counterparts. For better visibility, we use log scales, and indicate the different groups of moments in the graph. The precise magnitudes for all moments are shown in Appendix tables A7-A9. 29 The simulation draws individuals from the sample-specific, age-education distribution, and mirrors the empirical sampling scheme of the BRMS in selecting former (simulated) migrants who have returned since 2010. 39 Figure 12: Model fit: estimation moments Note: The figure shows the overall fit of the model by plotting moments simulated from the model against the corresponding moments observed in the data (using log scales). Model moments are based on a simulation of 100,000 individuals. Data moments are computed from BRMS and HIES data, see text. Each circle represents one targeted moment, with groups of moments indicated in the graph. For precise magnitudes, see Appendix Tables A7-A9. Table 3 lists our estimates for the initial stock of assets of agents with different levels of education, the investment costs for self-employment, and the probabilities that agents of different education levels have to access credit to finance migration expenses. While all moments contribute jointly to the estimation of the model’s parameters, the last column indicates the moment related most directly to the identification of each parameter. The total initial stock of assets strongly varies by education level, ranging from around 5,000 USD (adjusted for PPP) for individuals without any secondary education to 9,800 USD for higher-skilled individuals. These initial asset levels are considerably lower than the savings that are accumulated over time, and measured by the time of the survey after individuals return from abroad. Investment costs required for self-employment (that are compatible with the self-employment rates in our data) do not vary much by education level, averaging slightly over 30,000 USD (adjusted for PPP). Finally, Table 3 shows the probabilities that agents have access to finance a migration. These range from 54 percent for the least-educated individuals to almost 100 percent for those who have completed high school education. These probabilities are identified by matching the observed shares borrowing for migration. Recall, however, that an actual migration also requires agents to locate a foreign job. 40 Table 3: Structural parameter estimates: initial stock of assets, investment costs, and credit- access parameters Parameter Point estimate Standard error Identifying moment A0 1 5390.5 (2048.2) stock of assets, education level 1 0 A2 5012.9 (203.6) stock of assets, education level 2 A0 3 7891.4 (206.5) stock of assets, education level 3 A0 4 9756.1 (246.0) stock of assets, education level 4 I C1 31.53 (4.43) share self-employed, education level 1 I C2 29.65 (2.95) share self-employed, education level 2 I C3 35.40 (0.29) share self-employed, education level 3 I C4 32.73 (6.20) share self-employed, education level 4 p1 0.543 (0.046) share borrowing, education level 1 p2 0.572 (0.048) share borrowing, education level 2 p3 0.771 (0.026) share borrowing, education level 3 p4 0.980 (0.066) share borrowing, education level 4 Note: Asymptotic standard errors are in parentheses. Assets and investment costs are denoted in 1,000 PPP adjusted USD. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. d We show the larger subsets of parameters pertaining to preferences (ηe ) and foreign labor d 30 demand (φe ) graphically in Figure 13. The left panel of Figure 13 shows that utility losses d ηe from migration are largest for less-skilled migrants in Oman. Note that these losses are conditional on wages and job loss risk at a destination. Given the assumed utility function, d the absolute magnitude of ηe corresponds to forgone utility in thousands of PPP-adjusted d dollars. The displayed estimates are the values for ηe that make the model match observed emigration rates and migration duration across education groups and destination countries. 30 The full list of estimates, their standard errors, and the moment most directly contributing to identifi- cation of each parameter are relegated to Appendix tables A10 and A11. 41 Figure 13: Structural parameter estimates: preferences and labor-demand parameters Note: The left graph shows structural parameter estimates of utility losses from working abroad, by destination and education; the right graph shows estimates of the intercepts φd e in the labor demand functions λd e . Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. All estimates and their standard errors are listed in Appendix tables A10 and A11. The absolute magnitudes of demand parameters φd e are not readily interpretable, as they denote intercepts in a function determining the percentage of visa applications that is granted. The right panel of Figure 13 hence displays – for each education group and destination – probabilities of locating a job offer that would result for a baseline level of oil revenues, which we take to be oil revenues in Malaysia in 2000. These probabilities are generally smallest for individuals with the lowest levels of education, and highest for those who immigrate to Oman.31 31 The apparent outlier, Oman, is a result of empirically high migrant numbers, despite comparatively low wages and a high risk of dismissal and forced return – conditions which the model needs to match. 42 7 Policy Simulations Our model links a worker’s decisions about where to migrate, when to leave, and how long to stay to his financial decisions on self-employment after returning back home. We use the estimated model to evaluate three core policy interventions pertaining to the conditions under which migration and self-employment decisions are made. First, we consider cuts to the cost of migration. Second, we examine a subsidy that reduces the interest rate for self-employment loans. Finally, we investigate an information treatment that aligns individuals’ (overly optimistic) expectations about their foreign earnings potential with the average earnings observed in our data. Our model is flexible regarding the direction of the effects. For instance, lower costs of migration, our first policy intervention, make migration more attractive but the impact on self-employment is not clear. On the one hand, we may observe less self-employment in Bangladesh as workers favor migration as a means to achieve higher incomes. On the other hand, a rise in repatriated savings may contribute to more financing of entrepreneurial ac- tivities. Mobarak, Sharif, and Shrestha (2020), who examined the outcomes for Bangladeshi agricultural workers who went to Malaysia under a specific government-mediated migration program, document a negative short-run net effect of emigration on entrepreneurial activity while migrants are abroad. Anelli, Basso, Ippedico, and Peri (2020) present similar evidence for emigration from Italy. Our simulations predict that a decline in migration costs not only affects the emigration decision, but also its timing and the duration of the employment period abroad. We show that once we account for return migration, these changes lead to sizable increases in repatriated savings and business creation upon return. Similarly, lower interest rates on entrepreneurship loans may discourage emigration be- cause workers simply prefer to borrow more instead of accumulating savings abroad to raise the capital they need. Alternatively, improved credit conditions may lead workers who orig- inally considered self-employment out of their reach to decide to migrate, as a way to save to become entrepreneurs after their return. Our results show that, while a decline in inter- est rates naturally makes self-employment more attractive, the policy’s impact is actually undercut by an endogenous adjustment in migration behavior. This opposing effect needs to be accounted for in a cost-benefit analysis. There are also opposing effects from our third policy intervention, the information treat- ment regarding expected wages abroad. Individuals naturally benefit from better-informed decisions because the migrants with overly optimistic wage expectations do not migrate with the intervention. As a result, the level of disappointment and premature returns decline. At the same time, the overall levels of savings and business creation in Bangladesh also de- 43 cline. In conclusion, our policy simulations clearly show that understanding the extent and direction of these potential responses to policy measures are critical to identifying the likely impact of any intervention. 7.1 Lowering of Migration Costs Costs of migration receive significant attention, especially from policy makers. For many low-skilled temporary migrants from poor countries, the inability to pay these expenses is the main impediment to taking advantage of significantly higher wages abroad. The average migrant in our sample paid an amount that is equivalent to about three years of earnings in Bangladesh. Even if borrowing is an option, high rates of interest further increase the burdens for migrants. Lowering migration costs, either by limiting the rent-seeking behavior of intermediary firms, or through migration subsidies is one of the most frequently demanded policy interventions by migrants, policy advocates and academics (World Bank 2018). Our model shows that an immediate effect of lowering of the cost of migration would be an increase in emigration, and a shortening of the duration of migration episodes. The graphs in Panel A of Figure 14 show the effects that lowering migration costs by half would have on emigration rates over the life cycle of individuals with different education levels. Because liquidity constraints are the most binding for younger workers, the expansionary effect is stronger for them. For example, reducing the cost of emigration by half leads the annual emigration rate to increase from 0.5 percent to 0.8 percent for 25-year-old workers without secondary education. In addition, the peak migration age drops from 33 to 30, largely because workers need less personal savings to pay for fees. Similarly, under the same policy intervention of reducing fees, the annual emigration rate for 25-year-old workers with some secondary education increase from 1.6 percent to 1.9 percent. There is minimal impact for either education group after age 40. Finally, we find that the overall impact on the migration rate is an increase of 29 percent (9 percent) over the lifetime of a worker with less than (at least some) secondary education. 44 Figure 14: Simulated Effects of a Decrease in Migration Costs Panel A - Age of Migration Panel B - Duration of Migration Panel C - Assets and Self-employment 45 Panel D - Domestic Expenditure and Welfare Note: Figures show the estimated effects of a 50 percent cut in the cost of migration. Predictions are based on a simulation of 100,000 individuals from the model. In our dynamic model, migration costs are linked to all other decisions, such as the length of time a migrant chooses to work abroad and when (at what age) he chooses to depart. Everything else equal, and conditional on having their job contracts extended, the migration duration will be shorter if the cost of migration is lower, especially if the amount of debt needed to finance these costs has come down. With a 50 percent reduction in the cost of migration, Panel B of Figure 14 shows a marked leftward shift in the distribution of years spent abroad. The dashed line shows the distribution of migration duration at the baseline of expenses. The dotted line is the new distribution of the original migrants, and the solid line is the distribution of all migrants, including new migrants, under the lower-cost scenario. For less-educated migrants, however, there is a counteracting effect, in that a lower cost of migration facilitates migration at younger ages (see Panel A of Figure 14). With much of their working life ahead, these younger migrants tend to stay abroad for longer periods. In addition to showing a leftward shift in the distribution of migration episodes, the left figure in Panel B also shows a stronger decrease in the share of very short durations (less than two years), deriving from an increase in the share of young migrants. We find that the net effect is that workers now bunch around the median duration of seven years, and there is a 6.8 percent reduction in the duration of the average migration. In the case of workers with some secondary education (right figure of Panel B), there is a uniform shift to the left in the duration, with an average 10.5 percent reduction in the length of stay abroad. 46 Our results show that the combination of (i) lower initial migration costs, (ii) the resulting increased migration of younger workers, and (iii) shorter duration of migration episodes leads to higher level of accumulated savings, with important implications for business creation in Bangladesh. Panel C of Figure 14 illustrates these effects. The left figure shows a pronounced shift with higher levels of accumulated and repatriated assets at younger ages when the migration cost is halved. Original and new migrants, however, show significant behavioral differences in this respect. Original migrants (dotted line) return earlier, and, thus, they return with a lower level of assets. Once we account for new migrants (solid line), the level of repatriated assets per capita of the Bangladeshi population rises significantly. In addition, these assets are repatriated at younger ages. Together, the increase in emigration at younger ages and the ensuing rise in repatriated assets early in life are the main reasons why we observe an expansion of self-employment and entrepreneurship. The graph on the right in Panel C of Figure 14 shows that the same 50 percent decrease in migration costs raises business creation, particularly for younger workers. For example, the annual business creation rate increases from 0.7 percent to 1.1 percent for 35-year old workers, the median age for returning migrants. We see that about half of the expansion is due to increased investment by existing migrants, who return earlier in life. The other half derives from new migrants. The overall effect is a sizeable boost in business creation and self-employment. All of these effects of lower migration expenses – younger migrants, shorter migration periods, faster accumulation of savings, higher and earlier entry into self-employment – jointly contribute to significant welfare gains. We observe a rise in domestic expenditure in Bangladesh (net of migration expenses) over individuals’ life cycles, as shown in the left figure of Panel D of Figure 14. The overall effect is a gain of 0.06 of a standard deviation in the lifetime welfare across the population, including non-migrants. We should note, however, that these gains arise at the higher end of the welfare distribution (right figure of Panel D). There is minimal change on the left tail of the welfare distribution because the very poor still cannot afford to migrate and then move to self-employment, even with a 50 percent decrease in migration costs. We show that it is the middle- and upper-middle-income groups who manage to take advantage of lower migration costs, migrate in higher numbers, become self-employed and realize higher incomes.32 32 Note that welfare gains are lower than income gains, since agents in our model suffer a utility loss while working abroad. 47 7.2 Decrease in Lending Rates Credit constraints faced by workers who would like to seek self-employment and entrepreneurial opportunities are among the main motivations to migrate. Working abroad for higher wages (and under rather demanding conditions), saving some of these extra earnings, and bring- ing them home enable workers to finance self-employment activities upon their return. A policy-relevant question is what happens to migration, savings, and self-employment levels if domestic credit constraints were to be eased. We model such policy interventions through a 50 percent reduction in the interest rate faced by the workers on self-employment loans. As expected, an improvement in the credit conditions for self-employment mitigates em- igration pressures. As Figure 15 shows, a 50 percent cut in the lending rate on loans to finance a business impacts migration duration and asset repatriation. Panel A reveals that such a policy leads the overall emigration rate to decrease by 5.4 percent, distributed rather evenly across age groups. At the same time, migration duration shrinks by 6.4 percent. In sum, lower borrowing rates for entrepreneurship leads to a decline of migration at both the extensive and intensive margins. A situation in which fewer migrants stay abroad for shorter periods of time, in turn, implies a lower level of repatriated assets. The left figure of Panel B shows that assets repatriated by migrant workers decline by 8.4 percent, and that this effect is stronger for workers over age 35. The endogenous adjustment in migration behavior is due to the fact that financing self-employment through loans has now become less expensive. These effects are likely to limit part of the positive, direct effect of lower interest rates on business creation. The right figure in Panel B of Figure 15 shows that without any adjustment in migration, the 50 percent cut in lending rates on business loans would boost the annual business creation rate for a 35-year-old worker from 0.7 percent to about 1 percent. The counteracting effect of less migration and lower levels of repatriated savings instead only leaves a net effect of 0.8 percent for workers at that age. Across the life cycle, this accumulates to a 2 percentage point increase in the self-employment rate in Bangladesh. These patterns demonstrate the importance of a joint consideration of migration and self- employment decisions when designing policies aiming to either foster entrepreneurial activity, or to address international labor migration. 48 Figure 15: Simulated effects of a Decrease in the Lending Rate for Self-employment Panel A - Emigration and Migration Duration Panel B - Assets and Self-employment Note: Figures show the effects of a 50 percent cut in the interest rate for entrepreneurial loans. Predictions are based on a simulation of 100,000 individuals from the model. 7.3 Wage Expectations One of the observations from our survey is that workers who emigrate appear to have biased expectations about their employment opportunities abroad. Workers’ wage and savings expectations prior to migration are systematically higher than the actual wages and savings realized once they go abroad and start working. 49 Figure 16: Simulated effects of correct expectations about earnings abroad Panel A - Emigration and Migration Duration Panel B - Self-employment Note: Figures show the effects of an alignment of individual wage expectations with averages realized wages (conditional on observables). Predictions are based on a simulation of 100,000 individuals from the model. In this section, we use the model to evaluate the effect of an information dissemination policy that would align individuals’ expectations about their earnings potential abroad with the actual mean earnings of the Bangladeshi migrants we observe in the sample. The left figure in Panel A of Figure 16 shows that the implied reduction in expected earnings leads to a sizable reduction in the emigration rate by 13 percent, distributed more or less uniformly across age groups. The reduction in emigration duration (figure on the right) is milder at around 6 percent. These two results imply that the overall impact of correcting the wage 50 expectations is on the extensive margin (fewer migrants) than the intensive margin (shorter episodes). The resulting decline in repatriated savings (left figure in Panel B of Figure 16), in turn, leads to a decline in the rate at which new businesses are created, which accumulates to a 0.8 percentage point reduction in self-employment over the life cycle (see the right figure in Panel B). Whereas individual welfare unambiguously benefits from the better choices agents can make under more accurate information about foreign earnings, overall business creation is actually enhanced by individuals’ optimistic expectations. 8 Conclusions For millions of low-skilled workers from countries in South and Southeast Asia, temporary migration to the high-income countries in the Persian Gulf and East Asia is a critical part of their working lives. Immigrant workers not only earn significantly higher wages while working abroad, but they also accumulate savings that they use to finance self-employment and entrepreneurial activities upon their return. Workers’ decisions – about whether and when to emigrate, where to go abroad, how long to stay, how much to save, and what kind employment to seek upon their return – are intricately linked, as our paper demonstrates. This paper differs from the previous migration literature in two key respects. First, most papers in the migration literature focus on the wage and labor market gains during the mi- gration episode. Largely due to data constraints, these studies tend to employ cross-sectional analyses to answer their questions. By contrast, our paper highlights the interdependence between workers’ decisions at each stage of the life cycle, and the dynamic effects of tem- porary migration on workers’ economic activity upon return. Second, the literature usually focuses on the role that economic conditions at the immigration destination play in deter- mining immigrants’ decisions about when to return to their home countries. Our paper takes a different approach to the migration experience by modeling the central role played by self-employment aspirations upon return. These are important distinctions from the pre- vious work. A joint, dynamic analysis of migration and self-employment decisions is crucial. For example, policies that aim to overcome financial constraints to self-employment might become less effective because endogenous adjustments in migration behavior may offset part of the intended effects on business creation. Indeed, we show that this is the case. The data that underlie our findings come from a unique survey, the Bangladesh Return Migrant Survey (BRMS), which was specifically designed to investigate the patterns and determinants of the temporary migrants’ lifetime employment trajectories, earnings, and welfare. We use these data to construct a dynamic framework that models the tightly intertwined nature of pre-migration labor market out- 51 comes, the timing and duration of migration episodes, workers’ savings levels, and their post-return employment outcomes. We show how the savings accumulated abroad enable workers to overcome the credit constraints they face for entrepreneurship at home. We es- timate our dynamic model using data from the BRMS and the nationally representative Household Income and Expenditure Survey. Using our estimated model, we then explore the effects of changes in exogenous parameters, such as the cost of migration, and the in- terest rates on loans. We examine how these changes simultaneously affect migration and investment decisions, and lifetime welfare. Our analysis of simulated policy changes shows that a decline in migration costs would lead to a rise in emigration and a reduction in the duration of migration which, in turn, contribute to an increase in savings and self-employment. By contrast, a reduction in interest rates for entrepreneurial investment reduces migration and savings levels, undercutting part of the positive effect on self-employment. We find that providing workers with information that corrects widely held, inflated expectations about earnings abroad would reduce the rate and duration of emigration. The result is a decline in repatriated savings and in the rate at which new businesses are created. Though individual welfare benefits from better- informed choices, overall business creation is reduced by lowering of workers’ exaggerated expectations. Throughout these simulations, we observe changes in both the extensive and intensive margins of migration and investment decisions. These results confirm that policy makers must take such dynamic externalities into ac- count when designing their policy interventions. Given Bangladesh’s similarities with other migrant sending, lower income countries, the findings of this paper have relevance for low- skilled temporary migrant workers from many countries of origin. 52 References erˆ Adda, J´ orlach (forthcoming). “The Dy- ome, Christian Dustmann, and Joseph-Simon G¨ namics of Return Migration, Human Capital Accumulation, and Wage Assimilation”. Review of Economic Studies. Ahmed, Amer, Faizuddin Ahmed, Laurent Bossavie, C ¸ a˘ ¨ glar Ozden, and He Wang (2020). “Temporary Migration and Development: A New Dataset from Bangladesh”. Unpublished manuscript. Ahmed, S. Amer and Laurent Bossavie (2021). Towards Safer and More Productive Migration for South Asia. World Bank Group. c, and Murat Kırdar (forthcoming). Akay, Alpaslan, Alexandra Brausmann, Slobodan Djaji´ “Purchasing-Power-Parity and the Saving Behavior of Temporary Migrants”. European Economic Review. Albert, Christoph and Joan Monras (2020). “Immigration and Spatial Equilibrium: The Role of Expenditures in the Country of Origin”. Ambrosini, William, Karin Mayr, Giovanni Peri, and Dragos Radu (2015). “The Selection of Migrants and Returnees in Romania: Evidence and Long-run Implications”. Economics of Transition 23.4, pp. 753–793. Anelli, Massimo, Gaetano Basso, Giuseppe Ippedico, and Giovanni Peri (2020). “Does Em- igration Drain Entrepreneurs?” CESifo Working Paper No. 8388. Aydemir, Abdurrahman and George J Borjas (2007). “Cross-Country Variation in the Im- pact of International Migration: Canada, Mexico, and the United States”. Journal of the European Economic Association 5.4, pp. 663–708. Barro, Robert and Jong Wha Lee (2013). “A New Data Set of Educational Attainment in the World, 1950-2010”. Journal of Development Economics 104, pp. 184–198. Batista, Catia, Tara McIndoe-Calder, and Pedro Vicente (2017). “Return Migration, Self- selection and Entrepreneurship”. Oxford Bulletin of Economics and Statistics 79.5, pp. 797– 821. Battaglia, Marianna, Selim Gulesci, and Andreas Madestam (2021). “Repayment Flexibility and Risk Taking: Experimental Evidence from Credit Contracts”. 53 Beck, Thorsten (2007). “Financing Constraints of SMEs in Developing Countries: Evidence, Determinants and Solutions”. KDI 36th Anniversary International Conference, pp. 26– 27. Berg, Claudia, Shahe Emran, and Forhad Shilpi (2013). “Microfinance and Moneylenders: Long-run Effects of MFIs on Informal Credit Market in Bangladesh”. World Bank Policy Research Working Paper. us Fern´ Bertoli, Simone, Jes´ andez-Huertas Moraga, and Sekou Keita (2017). “The Elasticity of the Migrant Labour Supply: Evidence from Temporary Filipino Migrants”. Journal of Development Studies 53.11, pp. 1822–1834. us Moraga Fern´ Bertoli, Simone, Jes´ andez-Huertas, and Francesc Ortega (2013). “Crossing the Border: Self-selection, Earnings and Individual Migration Decisions”. Journal of De- velopment Economics 101, pp. 75–91. Borjas, George and Bernt Bratsberg (1996). “Who Leaves? The Outmigration of the Foreign- Born”. The Review of Economics and Statistics 78.1, pp. 165–76. uck, Tilman, Clotilde Mah´ Br¨ e (2018). “Return Migration and Self-employment: e, and Wim Naud´ Evidence from Kyrgyzstan”. IZA Discussion Paper 11332. Bryan, Gharad, Shyamal Chowdhury, and Ahmed Mushfiq Mobarak (2014). “Underinvest- ment in a Profitable Technology: The Case of Seasonal Migration in Bangladesh”. Econo- metrica 82.5, pp. 1671–1748. De Coulon, Augustin and Matloob Piracha (2005). “Self-selection and the Performance of Return Migrants: the Source Country Perspective”. Journal of Population Economics 18.4, pp. 779–807. De Vreyer, Philippe, Flore Gubert, and Anne-Sophie Robilliard (2010). “Are there Returns to Migration Experience? An Empirical Analysis Using Data on Return Migrants and Non-migrants in West Africa”. Annals of Economics and Statistics/Annales d’´economie et de statistique 97, pp. 307–328. c, Slobodan (2010). “Investment Opportunities in the Source Country and Tempo- Djaji´ economique 43.2, rary Migration”. Canadian Journal of Economics/Revue canadienne d’´ pp. 663–682. ed´ Docquier, Fr´ ¸ a˘ eric, C glar Ozden, and Giovanni Peri (2014). “The Labour Market Effects of Immigration and Emigration in OECD Countries”. The Economic Journal 124.579, pp. 1106–1145. 54 Dustmann, Christian, Tommaso Frattini, and Anna Rosso (2015). “The Effect of Emigration from Poland on Polish Wages”. The Scandinavian Journal of Economics 117.2, pp. 522– 564. Dustmann, Christian and Joseph-Simon G¨ orlach (2016). “The Economics of Temporary Mi- grations”. Journal of Economic Literature 54.1, pp. 98–136. Dustmann, Christian and Oliver Kirchkamp (2002). “The Optimal Migration Duration and Activity Choice after Re-migration”. Journal of Development Economics 67.2, pp. 351– 372. Fargues, Philippe (2011). “Immigration without Inclusion: Non-nationals in Nation-building in the Gulf States”. Asian and Pacific Migration Journal 20.3-4, pp. 273–292. coise De Bel-Air (2015). “Migration to the Gulf States: The Po- Fargues, Philippe and Fran¸ litical Economy of Exceptionalism”. Global Migration: Old Assumptions, New Dynamics, pp. 139–166. Farole, Thomas, Yoonyoung Cho, Laurent Bossavie, and Reyes Aterido (2017). Bangladesh Jobs Diagnostic. World Bank. Gang, Ira and Myeong-Su Yun (2000). “Returns to Returning”. Journal of Population Eco- nomics 13.1, pp. 57–79. Giambra, Samuele and David McKenzie (2021). “Self-employment and Migration”. World Development 141. Gindling, T. H. and David Newhouse (2014). “Self-Employment in the Developing World”. World Development 56, pp. 313–331. Girsberger, Esther Mirjam (2017). “Migration, Education and Work Opportunities”. IZA Discussion Paper 11028. orlach, Joseph-Simon (2020). “Borrowing Constraints, Migrant Selection, and the Dynam- G¨ ics of Return and Repeat Migration”. Unpublished manuscript. Gutierrez, Italo, Krishna Kumar, Minhaj Mahmud, Farzana Munshi, and Shanthi Nataraj (2019). “Transitions between Informal and Formal Employment: Results from a Worker Survey in Bangladesh”. IZA Journal of Development and Migration 9.1, pp. 1–27. International Labour Organization (2015). “The Homecoming: Profiling the Returning Mi- grant Workers of Bangladesh”. International Labour Organization. 55 Khanna, Gaurav, Caroline Theoharides, and Dean Yang (2020). “Abundance from Abroad: Migrant Income and Long-Run Economic Development”. Kirdar, Murat (2012). “Estimating the Impact of Immigrants on the Host Country Social Security System When Return Migration is an Endogenous Choice”. International Eco- nomic Review 53.2, pp. 453–486. Kleemans, Marieke (2015). “Migration Choice under Risk and Liquidity Constraints”. Un- published manuscript. Lacuesta, Aitor (2010). “A Revision of the Self-selection of Migrants Using Returning Mi- ´ grant’s Earnings”. Annals of Economics and Statistics/Annales d’Economie et de Statis- tique 97-98, pp. 235–259. Lessem, Rebecca (2018). “Mexico-US Immigration: Effects of Wages and Border Enforce- ment”. The Review of Economic Studies 85.4, pp. 2353–2388. e, Clotilde (2019). “Return Migration and Self-employment: Is There a ‘Jack-of-All- Mah´ Trades’ Effect?” Unpublished manuscript. Mahmud, W (2006). “Local Business Dynamics: Ground Realities and Policy Challenges”. Institutional Reform and the Informal Sector (IRIS), p. 13. El-Mallakh, Nelly and Jackline Wahba (2017). “Return Migrants and the Wage Premium: Does the Legal Status of Migrants Matter?” In: Economic Research Forum Working Paper Series No. Vol. 1133. — (2021). “Upward or Downward: Occupational Mobility and Return Migration”. World Development 137, pp. 105–203. Mallick, Debdulal (2012). “Microfinance and Moneylender Interest Rate: Evidence from Bangladesh”. World development 40.6, pp. 1181–1189. Marchetta, Francesca (2012). “Return Migration and the Survival of Entrepreneurial activ- ities in Egypt”. World Development 40.10, pp. 1999–2013. McCormick, Barry and Jackline Wahba (2001). “Overseas Work Experience, Savings and En- trepreneurship Amongst Return Migrants to LDCs”. Scottish journal of political economy 48.2, pp. 164–178. 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, pp. 116–127. 56 McKenzie, David, Caroline Theoharides, and Dean Yang (2014). “Distortions in the inter- national migrant labor market: evidence from Filipino migration and wage responses to destination country economic shocks”. American Economic Journal: Applied Economics 6.2, pp. 49–75. Mesnard, Alice (2004). “Temporary Migration and Capital Market Imperfections”. Oxford economic paper 56.2, pp. 242–262. Mobarak, Ahmed Mushfiq, Iffath Anwar Sharif, and Maheshwor Shrestha (2020). “Returns to Low-Skilled International Migration: Evidence from the Bangladesh-Malaysia Migration Lottery Program”. World Bank Group: Policy Research Working Paper. No.9165. e, Wim, Melissa Siegel, and Katrin Marchand (2017). “Migration, Entrepreneurship Naud´ and Development: Critical Questions”. IZA Journal of Migration 6.1, pp. 1–16. OECD (2008). International migration outlook. OECD publishing. Piracha, Matloob and Florin Vadean (2010). “Return Migration and Occupational Choice: Evidence from Albania”. World Development 38.8, pp. 1141–1155. Rahman, Rushidan Islam and Rizwanul Islam (2013). “Female Labour Force Participation in Bangladesh: Trends, Drivers and Barriers”. International Labour Organization. Rapoport, Hillel (2002). “Migration, Credit Constraints and Self-employment: A Simple Model of Occupational Choice, Inequality and Growth”. Economics Bulletin 15.7, pp. 1– 5. ed´ Rapoport, Hillel and Fr´ eric Docquier (2006). “The Economics of Migrants’ Remittances”. Handbook of the Economics of Giving, Altruism and Reciprocity 2, pp. 1135–1198. Reinhold, Steffen and Kevin Thom (2013). “Migration Experience and Earnings in the Mex- ican Labor Market”. Journal of Human Resources 48.3, pp. 768–820. Wahba, Jackline (2014). “Return migration and Economic Development”. In: International handbook on migration and economic development. Edward Elgar Publishing. — (2015). “Selection, Selection, Selection: the Impact of Return Migration”. Journal of Population Economics 28.3, pp. 535–563. Wahba, Jackline and Yves Zenou (2012). “Out of Sight, out of Mind: Migration, Entrepreneur- ship and Social Capital”. Regional Science and Urban Economics 42.5, pp. 890–903. Woodruff, Christopher and Rene Zenteno (2007). “Migration Networks and Microenterprises in Mexico”. Journal of development economics 82.2, pp. 509–528. 57 World Bank (2018). Moving for Prosperity: Global Migration and Labor Markets. World Bank Group. Yang, Dean (2006). “Why do Migrants Return to Poor Countries? Evidence from Philippine Migrants’ Responses to Exchange Rate Shocks”. The Review of Economics and Statistics 88.4, pp. 715–735. — (2008). “International Migration, Remittances and Household Investment: Evidence from Philippine Migrants’ Exchange Rate Shocks”. The Economic Journal 118.528, pp. 591– 630. 58 Appendix A Other Descriptive Figures and Tables ‘ Figure A1: Share of self-employed workers in Bangladesh and global oil prices Source: Bangladesh Labor Force Surveys 1999-2017 for share of self-employment; OPEC Reference Basket for crude oil prices. Note: The sample is restricted to employed males aged 18-59 in rural or semi-urban areas. 59 Figure A2: Marriage patterns of temporary migrants from Bangladesh Source: Bangladesh Return Migrant Survey (BRMS). Note: The sample is restricted to males aged 22-50. 60 Figure A3: Unemployment spells after return by duration of migration Source: Bangladesh Return Migrant Survey (BRMS). Note: The sample is restricted to employed males aged 18-59 who had stayed in Bangladesh for at least one year since the return. Unemployment spells are the duration of time between return and employment after return, conditional on the age of return, education level and employment status before migration. A local polynomial smooth is applied to this curve. Due to the small number of observations in the tails, duration longer than 13 years (10% of the sample) is treated as 13 years. 61 Table A1: Sample comparison of the HIES 2016/17 and the BRMS 2018/19 (males, age 18-59) HIES non-migrants HIES current migrants BRMS retrun migrants N Mean N Mean N Mean Male (among all age 18-59) 99,140 0.47 3,703 0.96 4,910 0.96 Rural/Semi-urban 47,137 0.92 4,709 0.99 Age category 18-24 47,137 0.21 3,571 0.14 4,709 0.03 25-34 47,137 0.28 3,571 0.39 4,709 0.35 35-44 47,137 0.25 3,571 0.32 4,709 0.39 45-54 47,137 0.19 3,571 0.13 4,709 0.20 55-59 47,137 0.07 3,571 0.02 4,709 0.03 Age 47,137 35.3 3,571 34.1 4,709 37.9 Education Illiterate 46,986 0.30 3,563 0.08 4,709 0.16 Some primary (1-5) 46,986 0.25 3,563 0.26 4,709 0.24 Some secondary (6-9) 46,986 0.21 3,563 0.38 4,709 0.36 Above some secondary (10-15) 46,986 0.17 3,563 0.26 4,709 0.22 62 Tertiary (16+) 46,986 0.06 3,563 0.03 4,709 0.02 Years of education 46,986 5.62 3,563 6.34 4,709 6.54 Years of staying abroad 3,560 6.10 4,709 6.46 Destination country GCC* 3,571 0.62 4,709 0.71 Southeast Asia 3,571 0.21 4,709 0.17 Others 3,571 0.17 4,709 0.12 Employment status ** Self-employed 47,137 0.28 3,333 0.54 Waged worker 47,137 0.56 3,333 0.29 Not working 47,137 0.16 3,333 0.17 Annual income, current BDT 27,792 123,474 3,151 152,292 Annual income, 2010 constant dollar 27,792 1,129 3,151 1,280 Note: *Bahrain is not included, as it is classified as ”Others” in the HIES. 3.6% migrated to Bahrain in the BRMS. **For the BRMS, the sample is restricted to migrants who have lived in Bangladesh for at least one year since return. Table A2: Self-employment and total earnings abroad, excluding the transportation sector (1) (2) (3) (4) (5) (6) Full sample Employed only OLS 2SLS First Stage OLS 2SLS First Stage Dependent Var Self-employed Self-employed ln(Earning) Self-employed Self-employed ln(Earning) ln(Cum. Earning abroad) 0.027*** 0.114** 0.037*** 0.101** (0.009) (0.031) (0.009) (0.049) Oil price growth × Oil rents/GDP 0.866*** 0.832*** (0.112) (0.104) Other controls Yes Yes Yes Yes Yes Yes Year of return FE Yes Yes Yes Yes Yes Yes Origin division FE Yes Yes Yes Yes Yes Yes 63 Destination country FE Yes Yes Yes Yes Yes Yes F-statistics of excluded instruments 60.3 63.8 Observations 3004 3004 3004 2425 2425 2425 Note: * p<0.10, ** p<0.05, * p<0.01. Standard errors clustered at the destination-origin level are reported in parenthesis. The sample is restricted to males age 18-59 and excludes individuals who have been back in Bangladesh for less than a year. Control variables include age and squared age at the time of survey, educational attainment, and a dummy for self-employment prior to migration. Oil price growth is the ratio of oil price at the time of return over the one at the time of departure. Table A3: Distribution and transition of temporary migrants across sectors of activity Panel A: Distribution of temporary migrants by sector of activity Before Migration During Migration After Return Sector % % % Agriculture 22.2 3.0 25.9 Construction 23.8 66.6 9.6 Manufacturing 2.5 5.1 1.7 Retail, Hotel, Restaurant 37.9 11.1 42.1 Transport, Utility 10.4 10.3 19.4 Other services 3.3 4.0 1.4 Total 100.0 100.0 100.0 Panel B: Transitions of temporary migrants across sectors of activity Sector After Return Cons., Rtl., Htl., Trans., Other Agri. Total Manu. Restr. Utility serv. Sector During Migration % % % % % % Agriculture 34.0 11.0 39.0 15.0 1.0 100.0 Construction, Manufacturing 26.1 12.3 40.8 19.5 1.3 100.0 Retail, Hotel, Restaurant 25.3 5.5 51.4 16.2 1.6 100.0 Transport, Utility 23.7 12.0 42.5 20.7 1.2 100.0 Other services 23.7 7.6 38.9 27.5 2.3 100.0 Total 25.9 11.3 42.1 19.4 1.4 100.0 Source: Bangladesh Return Migrant Survey (BRMS). Note: The sample is restricted to employed males aged 20-59. 64 B Details on the Structural Estimation This appendix provides further detail on the structural estimation of the model presented d in Section 5. Table A4 lists the parameters governing how the cost Cit of migration to d l destination d, earnings w ˜it expected there, and the earnings wit and profits πit realized in any location l vary by individual characteristics. Each of these outcomes is specified as a linear function of age and education, with separate functions for each location. The corresponding coefficient estimates are displayed in Table A4. Similarly, the risk of being d dismissed and forced to return, δit , is a function of age and education, and again destination country specific. It furthermore depends on destination d’s revenues from fuel exports in year t and the time a migrant has spent in the destination. The relation of this probability to individual characteristics is estimated through a probit specification. Table A5 further lists observed saving rates ρl in each location, and borrowing and lending rates based on Mallick (2012) and Berg, Emran, and Shilpi (2013). Finally, we let β = 0.95 and consider working lives until age T = 60. In line with our assumption of linear utility, we assume that individuals’ value at the end of life is given by their stock of assets at time e T , and if owning a business, the value of investment cost CI . 65 Table A4: Auxiliary regressions for the model Dependent variable Country Coefficient in the regression SD of residuals Constant Age Some Some Above Duration primary secondary some sec. ln(self-employment income) B 8.236 0.0009 0.0130 0.0292 0.0269 0.335 ln(waged income) B 8.065 0.0012 0.0494 0.0720 0.1038 0.312 ln(wage abroad) M 9.379 -0.0041 -0.0349 -0.0859 -0.0386 0.487 ln(wage abroad) O 9.241 0.0087 -0.0478 0.0418 -0.0395 0.807 ln(wage abroad) Q 8.495 0.0143 0.2333 0.1288 0.2804 0.507 ln(wage abroad) SA 8.951 0.0031 0.1658 0.1970 0.3226 0.602 ln(wage abroad) UAE 8.928 0.0048 0.0578 0.1246 0.2926 0.509 ln(wage expectation) M 9.392 0.0174 -0.0079 -0.0924 0.0500 0.827 ln(wage expectation) O 9.448 0.0090 -0.0245 0.0504 0.1691 0.850 ln(wage expectation) Q 8.810 0.0203 0.0495 0.4443 0.3088 0.921 ln(wage expectation) SA 9.149 0.0174 0.1693 0.2941 0.3651 0.779 ln(wage expectation) UAE 9.358 0.0154 0.0493 0.0987 0.2324 0.700 ln(migration costs) M 9.540 -0.0095 0.0048 0.0247 0.0248 66 ln(migration costs) O 9.390 -0.0043 0.0007 0.0035 -0.0099 ln(migration costs) Q 9.050 0.0026 0.1120 0.2640 0.2040 ln(migration costs) SA 9.795 -0.0096 0.0516 0.0008 0.0155 ln(migration costs) UAE 9.527 -0.0045 -0.0300 0.0281 -0.0213 forced return (probit coefficients) M -2.464 0.0146 0.0788 0.1120 -0.0007 0.0021 forced return (probit coefficients) O -1.900 0.0156 -0.0136 -0.0260 -0.0226 -0.0321 forced return (probit coefficients) Q -1.674 0.0112 -0.0099 0.1010 -0.0209 -0.0365 forced return (probit coefficients) SA -3.019 0.0268 0.1040 0.2870 0.2780 -0.0186 forced return (probit coefficients) UAE -2.400 0.0191 0.0092 0.0137 0.0129 -0.0114 Source: Bangladesh Return Migrant Survey (BRMS). Note: Incomes are annual. All monetary variables are converted into 2010 PPP adjusted dollar. Age in the regressions of self-employment and waged incomes in Bangladesh is the age at the time of the survey; age in the regressions of wages abroad, wage expectation and forced return is the age at the year of return; age in the regressions of migration costs is the age at departure. ”Some primary” is 1-5 years of schooling, ”Some secondary” is 6-9 years of schooling and ”Above secondary” is 10-15 years of schooling but excluding college and above. M stands for Malaysia, O for Oman, Q for Qatar, SA for Saudi Arabia, and UAE for United Arab Emirates. Table A5: Other parameters in the model Parameters Country Value Price level relative to Bangladesh M 1.62 Price level relative to Bangladesh O 1.66 Price level relative to Bangladesh Q 2.30 Price level relative to Bangladesh SA 1.52 Price level relative to Bangladesh UAE 2.31 Saving rate M 0.347 Saving rate O 0.289 Saving rate Q 0.303 Saving rate SA 0.354 Saving rate UAE 0.384 Saving rate B 0.114 interest rate of savings B 0.05 interest rate of loans for migration B 0.22 Note: Relative price level is the ratio of nominal exchange rate and PPP rate between destination country and Bangladesh in 2012, from World Bank Development database. Saving rate in destination countries is the share of remittance and cash taken back to home over total earnings abroad, from BRMS. Saving rate in Bangladesh is 1 minus the share of household consumption expenditures over household incomes, calculated from the HIES. Interest rate of savings and loans is from literature. M for Malaysia, O for Oman, Q for Qatar, SA for Saudi Arabia, and UAE for United Arab Emirates. 67 Table A6: Moments to fit in the model Moments Country By education level Illiterate Some primary Some secondary Above Secondary %Self-employed after return 0.450 0.513 0.529 0.653 Share of borrowing 0.469 0.437 0.427 0.353 Assets at the time of survey (2010PPP$) 13,291 14,086 21,332 30,687 Duration in the destination M 6.807 6.629 6.900 6.824 Duration in the destination O 4.592 4.000 4.060 4.310 Duration in the destination Q 3.293 3.531 3.383 3.992 Duration in the destination SA 9.919 9.768 9.233 9.266 Duration in the destination UAE 5.150 5.950 6.220 6.839 %Emigrants to the destination M 0.0037 0.0139 0.0210 0.0148 %Emigrants to the destination O 0.0019 0.0100 0.0165 0.0081 %Emigrants to the destination Q 0.0015 0.0057 0.0090 0.0082 68 %Emigrants to the destination SA 0.0060 0.0157 0.0295 0.0263 %Emigrants to the destination UAE 0.0017 0.0102 0.0154 0.0120 Note: The percentage of migrants to a given destination is from the HIES 2016-2017; Duration in the destination, the percentage of self-employed after return and assets at the time of survey is from the BRMS. ”Some primary” is 1-5 years of schooling, ”Some secondary” is 6-9 years of schooling and ”Above secondary” is 10-15 years of schooling but excluding college and above. M stands for Malaysia, O for Oman, Q for Qatar, SA for Saudi Arabia, and UAE for United Arab Emirates. Assets do not include the house value. Table A7: Model fit for emigration rates by destination and education Moment Data Std. Dev. Model emigration to Malaysia, education level 1 0.004 (0.001) 0.003 emigration to Malaysia, education level 2 0.014 (0.001) 0.014 emigration to Malaysia, education level 3 0.021 (0.001) 0.020 emigration to Malaysia, education level 4 0.015 (0.001) 0.015 emigration to Oman, education level 1 0.002 (0.000) 0.002 emigration to Oman, education level 2 0.010 (0.001) 0.008 emigration to Oman, education level 3 0.017 (0.001) 0.015 emigration to Oman, education level 4 0.008 (0.001) 0.008 emigration to Qatar, education level 1 0.001 (0.000) 0.002 emigration to Qatar, education level 2 0.006 (0.001) 0.005 emigration to Qatar, education level 3 0.009 (0.001) 0.007 emigration to Qatar, education level 4 0.008 (0.001) 0.005 emigration to Saudi-Arabia, educ. level 1 0.006 (0.001) 0.006 emigration to Saudi-Arabia, educ. level 2 0.016 (0.001) 0.017 emigration to Saudi-Arabia, educ. level 3 0.030 (0.002) 0.029 emigration to Saudi-Arabia, educ. level 4 0.026 (0.002) 0.028 emigration to the UAE, education level 1 0.002 (0.000) 0.001 emigration to the UAE, education level 2 0.010 (0.001) 0.010 emigration to the UAE, education level 3 0.015 (0.001) 0.017 emigration to the UAE, education level 4 0.012 (0.001) 0.009 Note: Data sample standard deviations in parentheses. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. Data source: HIES. 69 Table A8: Model fit for migration duration by destination and education level Moment Data Std. Dev. Model mean years in Malaysia, education level 1 6.807 (0.440) 6.364 mean years in Malaysia, education level 2 6.629 (0.355) 6.264 mean years in Malaysia, education level 3 6.900 (0.295) 5.600 mean years in Malaysia, education level 4 6.824 (0.412) 6.645 mean years in Oman, education level 1 4.592 (0.477) 3.891 mean years in Oman, education level 2 4.000 (0.325) 3.819 mean years in Oman, education level 3 4.060 (0.226) 4.340 mean years in Oman, education level 4 4.310 (0.400) 3.444 mean years in Qatar, education level 1 3.293 (0.777) 2.113 mean years in Qatar, education level 2 3.531 (0.546) 3.420 mean years in Qatar, education level 3 3.383 (0.281) 3.167 mean years in Qatar, education level 4 3.992 (0.568) 4.531 mean years in Saudi-Arabia, educ. level 1 9.919 (0.508) 8.721 mean years in Saudi-Arabia, educ. level 2 9.768 (0.439) 8.491 mean years in Saudi-Arabia, educ. level 3 9.233 (0.354) 7.115 mean years in Saudi-Arabia, educ. level 4 9.266 (0.399) 8.796 mean years in the UAE, education level 1 5.150 (0.301) 5.375 mean years in the UAE, education level 2 5.950 (0.263) 5.618 mean years in the UAE, education level 3 6.220 (0.268) 5.556 mean years in the UAE, education level 4 6.839 (0.310) 6.267 Note: Data sample standard deviations in parentheses. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. Data source: BRMS. 70 Table A9: Model fit for self-employment, asset level and borrowing by education level Moment Data Std. Dev. Model stock of assets, education level 1 13.291 (1.037) 17.404 stock of assets, education level 2 14.086 (0.792) 23.378 stock of assets, education level 3 21.332 (1.186) 24.647 stock of assets, education level 4 30.687 (2.082) 30.455 share borrowing, education level 1 0.469 (0.018) 0.443 share borrowing, education level 2 0.437 (0.015) 0.429 share borrowing, education level 3 0.427 (0.013) 0.396 share borrowing, education level 4 0.353 (0.017) 0.285 share self-employed, education level 1 0.450 (0.024) 0.398 share self-employed, education level 2 0.513 (0.020) 0.496 share self-employed, education level 3 0.529 (0.017) 0.485 share self-employed, education level 4 0.653 (0.021) 0.554 Note: Data sample standard deviations in parentheses. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. Data source: BRMS. 71 Table A10: Structural parameter estimates: preference parameters Parameter Point estimate Standard error Identifying moment M η1 -5.043 (0.774) mean years in Malaysia, education level 1 M η2 -4.414 (0.330) mean years in Malaysia, education level 2 M η3 -3.610 (0.184) mean years in Malaysia, education level 3 M η4 -3.911 (0.436) mean years in Malaysia, education level 4 O η1 -10.935 (1.426) mean years in Oman, education level 1 O η2 -9.680 (0.216) mean years in Oman, education level 2 O η3 -9.740 (0.135) mean years in Oman, education level 3 O η4 -9.890 (0.644) mean years in Oman, education level 4 Q η1 -5.159 (0.185) mean years in Qatar, education level 1 Q η2 -5.738 (0.436) mean years in Qatar, education level 2 Q η3 -4.604 (0.044) mean years in Qatar, education level 3 Q η4 -6.033 (1.404) mean years in Qatar, education level 4 SA η1 -3.856 (0.262) mean years in Saudi-Arabia, educ. level 1 SA η2 -4.782 (0.287) mean years in Saudi-Arabia, educ. level 2 SA η3 -5.252 (0.160) mean years in Saudi-Arabia, educ. level 3 SA η4 -5.617 (0.257) mean years in Saudi-Arabia, educ. level 4 U AE η1 -3.808 (0.850) mean years in the UAE, education level 1 U AE η2 -3.891 (1.757) mean years in the UAE, education level 2 U AE η3 -4.346 (0.015) mean years in the UAE, education level 3 U AE η4 -4.897 (0.337) mean years in the UAE, education level 4 Note: Asymptotic standard errors in parentheses. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. 72 Table A11: Structural parameter estimates: foreign labor demand parameters Parameter Point estimate Standard error Identifying moment φM 1 -8.726 ( 0.262) emigration to Malaysia, education level 1 M φ2 -7.751 ( 0.163) emigration to Malaysia, education level 2 φM 3 -7.404 ( 0.024) emigration to Malaysia, education level 3 φM 4 -7.822 ( 0.235) emigration to Malaysia, education level 4 φO 1 -9.186 ( 0.028) emigration to Oman, education level 1 φO 2 -7.036 ( 1.258) emigration to Oman, education level 2 φO 3 -7.488 ( 0.209) emigration to Oman, education level 3 φO 4 -7.105 ( 0.995) emigration to Oman, education level 4 φQ 1 -9.174 ( 0.038) emigration to Qatar, education level 1 φQ 2 -8.023 ( 0.606) emigration to Qatar, education level 2 φQ 3 -8.597 ( 0.030) emigration to Qatar, education level 3 φQ 4 -8.107 ( 1.606) emigration to Qatar, education level 4 φSA 1 -12.181 ( 0.152) emigration to Saudi-Arabia, educ. level 1 φSA 2 -11.470 ( 0.061) emigration to Saudi-Arabia, educ. level 2 φSA 3 -11.203 ( 0.067) emigration to Saudi-Arabia, educ. level 3 φSA 4 -11.398 ( 0.293) emigration to Saudi-Arabia, educ. level 4 φU 1 AE -11.546 ( 0.000) emigration to the UAE, education level 1 φU 2 AE -9.841 ( 0.483) emigration to the UAE, education level 2 φU 3 AE -9.636 ( 0.024) emigration to the UAE, education level 3 φU 4 AE -10.108 ( 0.407) emigration to the UAE, education level 4 Note: Asymptotic standard errors in parentheses. Education levels 1-4 refer to illiterate, some primary, some secondary and high school degree, respectively. 73