A PROFILE OF THE WORLD S YOUNG DEVELOPING COUNTRY MIGRANTS. David J. McKenzie # Development Research Group, The World Bank

Transcription

1 A PROFILE OF THE WORLD S YOUNG DEVELOPING COUNTRY MIGRANTS David J. McKenzie # Development Research Group, The World Bank Abstract Individual level census and household survey data are used to present a rich profile of the young developing migrants around the world. 12 to 24 year olds are found to comprise a large share of the flow of migrants, particularly in migration to other developing countries, with the age distribution of migration peaking in the late teens or early twenties. The paper examines in detail the age and gender composition of migrants, whether or not young migrants move alone or with a parent or spouse, their participation in schooling and work in the destination country, the types of jobs they do, and the age of return migration. The results suggest a high degree of commonality in the youth migrant experience across a number of destination countries. In particular, recent developing country young migrants tend to work in similar occupations all around the world, and are more concentrated in these occupations than recent older migrants or young immigrants who arrived at an earlier age. Nevertheless, there is also considerable heterogeneity amongst youth migrants: 25 percent of recently arrived 18 to 24 year olds are attending school in their destination country, but another 29 percent are not working or in school. This illustrates both the potential of migration for building human capital, and the fear that lack of integration prevents it from being used. # I thank the editor and two anonymous referees for comments. dmckenzie@worldbank.org; Phone (202) ; Mailing Address: The World Bank, MSN MC3-307, 1818 H Street N.W., Washington D.C , USA

2 1. Introduction Globalization has led to much faster movements of goods, technology, money, and ideas across borders, but large barriers remain to the movement of people. Despite these barriers, many people move, and by 2005 an estimated 190 million of the world s population lived outside their country of birth. 1 Large income gaps coupled with diverging demographics between developed and developing countries may amplify the pressures for migration. Holzmann (2005) estimates that without further migration the labor force in Europe, Russia, and high-income East Asia and the Pacific is projected to fall by 43 million between 2005 and 2025, while it is projected to rise by 19 million in China, 77 million in Latin America, 82 million in the Middle East, North Africa and Turkey, 93 million in low and middle income East Asian and Pacific countries, 211 million in Sub-Saharan Africa and 292 million in South and Central Asia. It has long been observed that people are more likely to migrate while young, so if more migration is to occur, it is likely to be youth doing it. 2 Indeed, in the 1930s Dorothy Thomas concluded that perhaps the only generalization that could be made in regard to differentials in internal migration was that migrants tended to be young adults or people in their late teens (quoted in Lee, 1966). Despite this general observation, very little is known about the characteristics of young migrants. National statistical agencies typically only publish very aggregated information on the age structure of migration, seldom providing cross-tabulations of age with other characteristics of interest such as country of 1 Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, Trends in Total Migrant Stock: The 2005 Revision. 2 We will discuss in detail the definition of youth in Section 3. We use the World Bank definition of youth as 12 to 24 year olds (World Bank, 2006), but break up most statistics into three age groups: 12-14, 15-17, and

3 origin. Furthermore, aggregate statistics do not provide information on whether youth are migrating alone or with others, their participation in education and the labor market, the types of jobs they work in, and whether or not they ever return to their home countries. All these factors are important for helping to assess the potential impact of migration on young migrants and on the development of their sending countries, as well as their impact on receiving countries. This paper aims to be a first step in filling these knowledge gaps, by using microdata from national censuses and large scale household surveys in order to provide a detailed description of the lives and circumstances of young migrants from developing countries. Detailed data are available for migration to developed and developing countries, allowing comparison in South-North and South- South migration patterns. Some of the main patterns observed are: Youth aged 12 to 24 average one-third of the flow of migrants, and one quarter of the stock of immigrants. The share of youth in the flow is greatest in migration to developing country destinations, and lowest for migration to countries such as Canada, which rely on skill-intensive admission criteria. The share of females among the flow of youth migrants varies from 39% in South Africa to 61% in Mexico, averaging 50 percent across all countries looked at. Young migrants are less likely to be accompanying a parent in migration when moving to developing countries rather than developed, and are less likely to be accompanying a parent the older they are. On average 77 percent of recently arrived 12 to 14 year old migrants are living with a parent, compared to 52 percent of 15 to 17 year olds and only 14 percent of 18 to 24 year olds

4 Young female migrants are much more likely to be married and accompanying a spouse than young male migrants. There are large differences in the share of recently arrived 18 to 24 year old migrants in school across countries. Approximately half of such migrants to Canada and the United Kingdom are studying, compared to only 25 percent of migrants to the United States and less than 2 percent of migrants to Cote d Ivoire. In some countries, one finds over 25 percent of male 18 to 24 year olds and over 50 percent of female 18 to 24 year olds that are neither in school nor working. Recently arrived migrant 18 to 24 year olds who are working tend to be more heavily concentrated in a few occupations than both older migrant workers and the stock of 18 to 24 year old immigrants. The age of return migration is also very young in many countries, with the median return migrant being aged 25 to 30. This leaves considerable time for working in the home country. The rest of the paper is structured as follows. Section 2 outlines the theoretical reasons to expect migrants to be young. Section 3 discusses the definition of youth and migrant and outlines the data used. Age and gender patterns of migration are presented in Section 4. Section 5 examines whether migrants live with parents and are married, while Section 6 looks at participation in school and work. Section 7 considers the jobs young migrants work in, while Section 8 estimates age profiles of return migration. Section 9 concludes

5 2. Migration Theory and Age of Migration Migration theory offers reasons to expect high youth participation in migration due to individual, family, and community factors: Individual factors. The classic economic explanation for the greater tendency of the young to migrate is that migration is an investment, requiring individuals to incur costs to generate the returns from higher income (Sjaastad, 1962). Costs include the financial costs of moving, finding a job, and forgoing earnings and the psychic costs of leaving familiar surroundings and adapting to a new labor market. The expected returns depend on the wage, the probability of obtaining a job (Todaro, 1969), and the length of time working overseas. Young people are likely to have both higher lifetime returns and face lower costs from moving. Expected returns can be higher because they have more of their human capital in education than in job-specific skills than do older workers and longer working lives. The forgone earnings from migrating are likely to be less for youth, especially in countries with high levels of youth unemployment and strict seniority rules that lower wages more for the young. The psychic costs of moving may also be lower, since age tends to raise investments in family and other contacts (Simon 1986, Lundborg 1991). Family factors. A key insight of the new economics of migration literature (Stark and Bloom 1985, Stark and Levhari 1982) is that the decision to migrate is often a decision of the family, not just an individual, particularly in developing countries, where imperfect credit and insurance markets create a rationale for migrating to diversify risk and finance costly household investment activities. Households can send one of their members and count on remittances to help them cope with shocks, such as financial crises and natural - 5 -

6 disasters. 3 They will select that member not just based on who has the greatest individual gains from migrating but also consider the household functions the member performs, and the likelihood of the member remitting money. In many societies parents exhibit greater control over daughters than sons, so young women may be especially likely to be sent for family reasons. Several researchers argued that this is so in the Philippines, where the majority of migrants are women. 4 Community factors. Once some young people have migrated, community factors make it more likely that other youth will migrate too. One reason for this is the migrant social network, which lowers the costs and increases the benefits of migrating. Since youth are more likely to migrate due to the reasons just discussed, a young potential migrant is more likely to have a recent migrant in his or her peer network than an older individual, and so may be more likely to benefit from the migrant network. Over time a culture of migration can then develop in a community, with migration becoming a rite of passage for youth, and with those not migrating considered lazy and unenterprising (Massey et al, 1998). 3. Definitions and Data 3.1. Definitions Youth is the transitional phase from childhood to adulthood, where important life transitions crucial for human capital formation, such as finishing school, beginning work, getting married, and leaving home typically occur (World Bank, 2006). The defining age 3 See, for example, Yang and Choi (2005) for evidence that Filipino households experiencing rainfall shocks receive more remittances and McKenzie (2003) for evidence showing an increase in remittances received by households during the Mexican peso crisis of See Massey et al. (1998). Lauby and Stark (1988) argue this is also the case for rural-to-urban migration in the Philippines

7 range for youth varies across countries and agencies. Two common definitions are ages 15-24, used by the United Nations, and ages 12-24, used by the World Bank (2006). We adopt the latter age range for calculating the share of youth amongst all migrants, and then disaggregate the data into three age ranges for our more detailed analysis of the schooling, marital status, and employment of young migrants. These age ranges are 12 to 14, 15 to 17, and 18 to 24. The age cutoff of 18 is particularly salient for migration, since it is the age at which many migrant-sending countries treat individuals as an adult for the purpose of obtaining a passport and leaving without parental permission (McKenzie, 2007) and the age at which many migrant-receiving treat individuals as adults for work and immigration purposes. Furthermore, it is the age at which individuals typically finish secondary schooling, so that migration after age 18 may be for tertiary education. Nevertheless, these age cutoffs are no means constant across countries. For example, South Africa considers 16 the adult age for obtaining a passport and has a school leaving age and minimum age of employment of 15. The age range covers such cases. Finally, the age range 12 to 14 covers individuals who are legally children in most countries, but who may in some countries still end up migrating by themselves. We define migrant based on country of birth. We distinguish between the stock of youth migrants, defined as all individuals aged 12 to 24 and born abroad, regardless of the age of migration; and the flow of youth migrants, defined as individuals aged 12 to 24 who were born abroad and have recently arrived in the destination country. Ideally we would measure recently arrived as within the last one to two years, although in the cases of Argentina and Mexico, the Censuses only reveal - 7 -

8 whether or not individuals have migrated within the past 5 years. We will present some basic statistics on the share of youth in the stock and flow of migrants, and then focus on flow migrants to examine the characteristics and circumstances upon arrival of individuals who migrate when aged 12 to 24. The stock, in contrast, will contain many individuals who migrated as young babies or children, and therefore is not informative about the characteristics of individuals who migrate when aged 12 to Data Table 1 summarizes the main sources of data used in this paper. We utilize public use microdata from eleven Censuses from 1999 to 2001, and 2002 household survey data for Côte d Ivoire. 5 The choice of countries to include was dictated by data availability and a desire to include as many different regions as possible. The sample includes six developed countries (Canada, Greece, Portugal, Spain, the United Kingdom, and the United States), four middle income developing countries (Argentina, Costa Rica, Mexico and South Africa), and two poor Sub-Saharan African countries (Côte d Ivoire and Kenya). Overall this yields microdata on a stock of over 300,000 young migrants, and a flow of over 60,000 young migrants. We define developing countries broadly to include all countries not classified as high income by the World Bank, based on a GNI per capita threshold of $11,116 or more. For practical purposes this means developing countries do not include Western Europe, Australia and New Zealand, the United States and Canada, oil-rich Gulf States, rich Caribbean islands, Hong Kong, Singapore, Korea and Japan. Since the origins of migrants vary greatly across destinations, the last column of Table 1 shows the major 5 Data from Côte d Ivoire are from the May 2002 Enquete Niveau de Vie des Menages (Household Living Standards Survey), and were obtained from the World Bank

9 sources of youth immigrants in each country. We can thus, for example, consider the results for Argentina as largely reflecting the experiences of youth from neighboring South American countries that have moved to Argentina. Likewise in South Africa and Mexico, we see youth migrating from poorer neighboring countries to take advantage of the work and educational opportunities available in their nearby neighbors. The costs of migration are usually lower, both in terms of travel costs, language barriers, and overcoming administrative barriers, when moving from one developing country to a neighboring one, which is why many youth from Lesotho and Mozambique move to South Africa rather than, say, Europe. These reasons also help explain why individuals might move from neighboring African countries to Kenya and Côte d Ivoire. In these cases, the neighboring countries have similar levels of development, and one would expect to see flows of migrants in both directions. Census data from Uganda in 2002 reveal this to be the case: 11 percent of all immigrants to Uganda are from Kenya, with Congo, Sudan and Rwanda all accounting for larger shares of immigrants due to war-driven migration from these latter countries. The major missing migration flows are migration from Central Asia to Russia, migration from South Asia to the Gulf states, and migration to India from neighboring South Asian countries. Microdata from these countries was unavailable. However, I did obtain customized cross-tabulations from the 2003 Census in Oman which provides some information on migrants to a Gulf state. Finally, in Section 8, I supplement the main dataset with data on return migrants from household surveys in Albania, Mexico, Pakistan, and the Philippines. This data will be discussed in more detail in Section

10 A final caveat is that Census data vary in their coverage of illegal migrants. Census data do not ask about legal status, preventing separation of the data along this dimension. Data from sending households in both Mexico and Albania suggest that youth are more likely to migrate illegally (World Bank, 2006), which likely reflects the greater risk-taking propensity of youth, and fewer legal opportunities for their migration. To the extent that Censuses undercount illegal migrants, the share of youth amongst total migrants will be even higher than calculated here. These caveats not withstanding, the data here provide the most comprehensive picture of youth migration currently available, and encapsulate a wide range of migrant experiences of interest. 4. Age and Gender Patterns of Migration Figure 1 then uses this microdata to plot kernel densities of the age distribution of the migrant flow in different countries around the world. This is supplemented by Table 2, which gives the share of the stock and flow of migrants who are aged 12 to 24 in each country, and by Table 3, which gives the share of the stock and flow of young migrants who are female. Figure 1 shows the age distribution of the flow of migration from developing countries peaks just before or at age 24 in many destination countries. The age distribution peaks earliest in mainly developing countries, such as Costa Rica, South Africa, Kenya, Argentina, and females in Côte d Ivoire. The borders are more porous in these countries, and young migrants from neighboring countries face relatively less border restrictions to migration than they do in traveling to developed countries. The United States also has a young age distribution, consistent with large-scale illegal migration from Mexico. The

11 age distribution peaks at around 24 in Spain, Greece, and Portugal, and peaks latest for Canada, which uses a highly skill-intensive points system to select migrants. Table 2 shows that in the majority of cases 12 to 24 year olds constitute a larger share of the migrant flow than they do of the stock, suggesting a good number of migrants stay for considerable periods in their host countries. On average, youth constitute one-third of the migrant flow. However, there is considerable diversity by gender and country. The share of 12 to 24 year olds in the migrant flow varies from only 20 percent in Canada to 46 percent among males in South Africa and 50 percent among females in Côte d Ivoire. For comparison with the microdata, cross-tabulations from Oman show that 10 to 24 year olds comprise only 9.5 percent of the stock of male migrants, and 19.7 percent of the stock of female migrants. This is lower than any of the countries shown in Table 2, and reflects the policy in Gulf states of hiring migrant workers and not allowing them to bring dependents. Significant differences in the share of youth in the migrant flow also exist within a single destination country. In the United States, countries whose migrants mostly come through skilled and education categories have lower youth shares: only 17 percent of the flow of Chinese immigrants are aged 12-24, compared with migrant flow youth shares of more than 40 percent from Mexico, Honduras, Guatemala, and El Salvador, for whom family migration and illegal channels are more important. The age distribution of migrants is similar by gender in most countries. The main exceptions are Côte d Ivoire and Canada, and appears to be due to a large share of female migrants being married in these countries, and females tending to marry older men among migrants coming to these countries. Table 3 shows the share of females amongst

12 the flow of youth migrants also varies a lot by country from 39 percent in South Africa to 61 percent in Mexico. Two potential explanations for large differences in the gender composition of migration across receiving countries are differences in the amount of marriage migration occurring, and differences in the labor demand for male-dominated vs female-dominated jobs across countries. There is a correlation of between the share of the 18 to 24 flow who are female (Table 3) and the proportion of the flow of 18 to 24 female migrants who are married (Table 4). Thus in Argentina, 61 percent of the 18 to 24 flow are female, and only 17 percent of the female 18 to 24 flow are married. Countries such as South Africa, where labor demand has typically been for construction and mining work, have a lower female share than in Argentina, where demand is high for domestic workers. We now concentrate exclusively on the flow of migrants for the remainder of the paper, since this the group of 12 to 24 year olds who have migrated as youth, rather than as babies or children. 5. Travelling alone or along? Given this large share of youth amongst the world s migrant population, one question which arises is whether these young migrants are migrating alone, or simply accompanying the migration decision of other family members. Even youth who migrate alone may be migrating as the result of a family decision to send one member abroad, yet such migration is still very different to a teenager accompanying a parent or a wife accompanying a husband. To explore such patterns, Table 4 reports the proportion of recently arrived youth migrants who are living with a parent and the proportion who are

13 married. This is done separately for three age ranges: 12 to 14, 15 to 17, and 18 to 24, since one would expect the degree of autonomy to change with age. A large fraction of recently arrived 12 to 14 year olds and many recently arrived 15 to 17 year olds are accompanying or joining migrant parents. The data allow us to say whether a migrant youth is living in the same household as their mother or father, but not whether they physically moved country with their parent. Nevertheless, if a young person migrates with their parent, or migrates to join them overseas, in both cases the migration is tied to the location decision of another family member. 6 We see young migrants are more likely to be living with parents when moving to developed countries, which enforce more strongly age restrictions on work and which allow children under 18 to join their parents through special family reunification immigration categories. Over 90 percent of 12 to 14 year olds moving to Canada and the United Kingdom live with at least one parent, compared with less than 60 percent migrating to South Africa and less than 50 percent migrating to Côte d Ivoire. A much smaller percentage of 18 to 24 year olds migrate or live with their parents: less than 20 percent in all countries, except Canada, which again shows how most migration to Canada is through high skilled categories which exclude youth as the principal migrants. Although we have only stock data for Oman, we see there that most youth do not live with their parents: 22.1% of year old males and 23.5% of year old females live with their parents. 6 In a minority of cases the parent may not be a migrant, even though their child is if the parent is a native and their child is born abroad. Such cases also clearly represent a case of the young person not making the migration journey without parents

14 The immigration systems of most countries also provide for fairly easy migration of the spouse of an existing citizen or legal resident. Young female migrants often migrate to accompany or join their migrant spouse, and young male migrants are much less likely to be married. Of 18 to 24 year old females migrating to Côte d Ivoire 74 percent are married compared with 38 percent of males, and 45 percent of females this age migrating to the Canada are married, compared with only 12 percent of men. For Oman, crosstabulations show 19.6 percent of the stock of year old migrant men are married, compared to 43.7 percent of the stock of year old migrant women. 6. Students, Workers, or Neither? Youth migration can serve to build human capital if young migrants attend school in the country they move to, and also through work experience. In contrast, a key policy concern in many countries is a lack of integration of young immigrants, manifested in terms of high numbers of unemployed out-of-school youths. Table 5 investigates this in practice by examining the proportion of young migrants who attend school, work, and who do neither. Again I separate into three age groups, and by gender. Table 5 shows that the vast majority of recently arrived 12 to 14 year old migrants are attending school, except in Côte d Ivoire, where less than 20% of all females are, and Costa Rica, where less than 60% of both males and females are. However, even among 12 to 14 year olds, 10% of males and 8% of females are not attending school in the U.S. among new migrants. This may reflect concerns about apprehension for illegal migrants. Consistent with this, one can compare Indians and Chinese, who are mostly all legal migrants in the U.S., to Mexicans and Salvadoreans, who have higher incidences of illegal presence. Over 97% of newly arrived 12 to 14 year old Chinese and 94% of 12 to

15 14 year old Indians are attending school in the U.S., compared to 84-85% of Mexicans and Salvadoreans. Among recently arrived 18 to 24 year olds, one finds over half are attending school in Canada and the United Kingdom, reflecting the importance of migration for education in these countries. However, there are also sizeable proportions of 18 to 24 year olds attending school in Mexico and South Africa, showing regional migration for higher education. In fact, a higher proportion of male 18 to 24 year old migrants to South Africa are attending school than recently arrived male migrants to Spain and the United States in this age range. In Oman, 16.3% of the stock of males 15 to 24 and 20.1% of the stock of females are attending school. Table 5 also shows male migrants tend to have a higher likelihood of working than young female migrants, with the exception of the United Kingdom and Canada, which have roughly equal proportions. This gender difference is also seen in Oman, where 75.7% of the stock of migrant males aged are working, compared to 45.5% of the stock of female migrants aged Many of the surveys do not ask about work of 12 to 14 year olds. The available data shows a high incidence of child labor among recently arrived 12 to 14 year olds in Côte d Ivoire and Kenya, with very low incidences in Greece and Spain. Despite the idea that 18 forms a legal threshold for work migration, one sees large numbers of 15 to 17 year old recently arrived migrants working in some countries: 61 percent of males in Costa Rica, 47 percent of males in Greece, and 40 percent of males in the U.S. In comparison, male 15 to 17 year old recently arrived migrant employment rates are less than 20% in Canada and Spain

16 Youth who are neither working nor in school are often the cause of most concern for policymakers. The last columns of Table 7 shows that the proportion of recently arrived young migrants involved in neither activity is very large in some countries. As one would expect, the proportions are generally higher for young women, who may be involved in raising children or other household activities. However, the rates are still high among males: 20 percent or more of recently arrived male 18 to 24 year olds are idle in Argentina, Kenya, South Africa, Spain and the United States. In contrast, Canada, Portugal, and the United Kingdom have much lower proportions of young migrants who aren t in either school or work. 7. What types of jobs do they work in? The individual level data can also be used to look at the types of jobs young migrants work in. Table 6 reports the most common 3 occupations for recently arrived male and female migrants aged 18 to 24 in each country for which data is available. The table also gives the percent of all recently arrived 18 to 24 year old migrants who are working in this occupational category. Young men tend to work in physically intensive jobs like construction and agricultural labor. Young migrant women are most likely to be domestic workers, cashiers, sales clerks, waitresses and cooks. In Oman, the most common jobs for youth are also service occupations. Many of these jobs are considered to be of low status in developed countries, and offer little in the way of career advancement. Sociologists have argued that this creates a structural demand for workers who see employment as solely a means of income generation (see Massey et al for a review). This role was traditionally filled

17 by teenagers in developed countries, whose lower labor-force participation therefore creates additional demand for immigrant youth to fill these positions. Table 6 shows a high concentration of recent 18 to 24 migrants in just a few occupations. This is particularly true for women, with 50 percent or more of recent female migrants in Argentina and Côte d Ivoire working in Domestic Service. One may ask whether this clustering is specific to all recently arrived immigrants, and whether it is greater for recently arrived youth than immigrant youth who arrived at earlier ages. To formally test this, I calculate a Herfindahl Index to measure the degree of occupational concentration. This is the sum of the squared proportions of workers in each occupation category. The index takes values between 0 and 1, with 0 being the least concentrated and 1 indicating complete concentration, with all workers working in the same occupation. This can only be done accurately for surveys with large samples and rich occupational coding, restricting the analysis to the seven countries in Table 7. Calculating this index over all 475 occupation categories listed in the 2000 U.S. Census gives a Herfindahl index for recent male youth of 0.028, which is higher than the indices for older recent migrants and for the stock of native youth (Table 7). We also see male migrant youth to be more concentrated in just a few jobs in Spain and South Africa. The flow is more concentrated in a few occupations than the stock in all countries. For males, we find that recent young migrants are more concentrated in a few occupations than recent 30 to 50 year old migrants. In every case except the U.S. we find recently arrived 18 to 24 female migrants to have a higher occupational concentration than recently arrived 18 to 24 male migrants. However, the 18 to 24 year old females are not necessarily more concentrated into a few

18 occupations than recently arrived 30 to 50 year olds. Thus it appears that all newly arrived female migrants only have a few occupations that are readily open to them. 8. Age at Return While migration offers young adults the opportunity to acquire education and earn higher incomes abroad, sending countries often worry that many of these migrants will never return. The proportion of migrants who return is very difficult to measure with the existing data available in most countries. However, the few studies which have been able to measure return show that a high proportion of migrants do return. For example, in the U.S. Jasso and Rosenzweig (1982) estimate that up to 50 percent of the migrants who entered the U.S. in 1971 had returned by Dustmann (2005) summarizes several other studies, which find rates of return of 80 percent in Switzerland and two-thirds in Germany. Migration return rates can be quite high when the country of return is close by. In a study of migrants from Western Mexico, Reyes (1997) estimates that about 50 percent of all immigrants return within two years, and by 10 years, almost 70 percent had returned. Given that many migrants return, a key question is then at what age do they return? Individuals who return while relatively young will have more of their working lives to spend in their home countries, potentially using skills acquired abroad. In contrast, individuals who return to their home countries only after retirement may have greatest impact through the savings they bring back. Several household surveys ask questions on last place of residence 12 months or 5 years ago, making it possible to look at the age of return for recent return migrants

19 Figure 2 plots the age distribution of return migration in five countries, showing that migrants tend to return at fairly young ages, rather than at retirement. 7 The median age of return migration is 24 in Mexico, 28 in South Africa, 30 in Pakistan, 31 in Albania and 34 in the Philippines. This includes the return migration of some individuals who migrated as young children, and does not tell us the duration of time spent abroad. Three of the household surveys also ask when migrants left their home countries, as well as when they returned, allowing calculation of durations of migration, and age of return conditional on leaving as a youth. The median (mean) time abroad for youth migrants from Mexico is 3 years (5 years); from Pakistan 4 years (6 years) and from Albania 7 months (19 months). The median age of return for youth migrating between age 12 and 24 in all three countries is 24 to 25 years. This shows that half of youth migrants who ever return, do so while still aged 12 to 24. The majority of others return before age 40, meaning that youth migrants who return to their home countries typically have many years of working life in their home countries. 9. Conclusions This paper shows that young people aged 12 to 24 constitute a large share of the migrant flow approximately one-third on average, and up to 50 percent in some cases. Moreover, we find that the age distribution of return migration peaks at 25 to 28 in five countries, so that a large share of return migrants are also of prime working age. The rich 7 Data on Albania are from the 2005 Albanian Living Standards Measurement Survey, which includes 1419 return migrants; Data on Mexico are from the 1997 ENADID, and have return migration from the U.S. for 4,917 individuals; Data on 96 return migrants to Pakistan are taken from the 2001/02 Pakistan Rural Household Survey; Data on 4,548 return migrants to the Philippines are from the waves of the Survey on Overseas Filipinos; and Data on 2123 return migrants to South Africa are from the 2001 South African Census

20 descriptive detail provided in this paper offers a basis for understanding better the characteristics and experience of these youth migrants which should be of interest to policymakers in both sending and receiving countries. The descriptive analysis in this report suggests several areas which warrant policy attention and further academic study. Firstly, while youth have a high tendency to migrate, youth constitute a much lower share of migrants to developed countries with migration systems that select on skills. Several other countries appear to be considering targeting high-skilled workers more in their migration policies, with the points-system announced by the United Kingdom recently one such example. It should be recognized that such systems offer few opportunities to the large number of young people in the world who wish to migrate temporarily, and therefore a challenge for policymakers is how to offer more opportunities. Secondly, the analysis shows that the large majority of young migrants are either in school or are working. One interesting finding is that there are high shares of migrant youth attending school past age 18 in Mexico and South Africa. This shows the potential for South-South migration to be beneficial in developing human capital. Nevertheless, one concern is that there are also sizeable numbers of older youth who are neither in school or working. Further work needs to explore the reasons for this lack of activity, and help develop policies to better integrate immigrate youth into the labor market. The relationship between lack of integration, the likelihood of return migration, and the level of disaffection among immigrant youth is something the data here can not speak to, and is an open area for future work

21 References Dustmann, Christian (2005) Why go back? Return motives of migrant workers, Mimeo. Holzmann, Robert "Demographic alternatives for aging industrial societies: Enhanced immigration, labor force participation, or total fertility rate." Paper presented at the G-20 Workshop on Demographic Challenges and Migration. Jasso, Guilermina and Mark Rosenzweig (1982) Estimating the emigration rates of legal immigrants using administrative and survey data: the 1971 cohort of immigrants to the United States, Demography 19: Lauby, Jennifer, and Oded Stark "Individual Migration as a Family Strategy: Young Women in the Philippines." Population Studies 42(3): Lee, Everitt S "A Theory of Migration." Demography 3(1): Lundborg, Per "An Interpretation of the Effects of Age on Migration: Nordic Migrants' Choice of Settlement in Sweden." Southern Economic Journal 58(2): Massey, Douglas S., Joaquín Arango, Graeme Hugo, Ali Kouaouchi, Adela Pellegrino, and Edward Taylor Worlds in Motion: Understanding international migration at the end of the millennium. Oxford: Oxford University Press. Minnesota Population Center (2007) Integrated Public Use Microdata Series International: Version 3.0. Minneapolis: University of Minnesota, available online at McKenzie, David J "How do households cope with aggregate shocks? Evidence from the Mexican Peso crisis." World Development 31(7):

22 McKenzie, David J Paper walls are easier to tear down: Passport costs and legal barriers to migration, World Development 35(11): Reyes, Belinda (1997) Dynamics of Immigration: Return Migration to Western Mexico, Public Policy Institute of California. Ruggles, Steven, Matthew Sobek, Trent Alexander, Catherine A. Fitch, Ronald Goeken, Patricia Kelly Hall, Miriam King, and Chad Ronnander (2004) Integrated Public Use Microdata Series: Version 3.0 [Machine-readable database]. Minneapolis, MN: Minnesota Population Center, Simon, Julian L "Basic Data concerning Immigration into the United States." Annals of the American Academy of Political and Social Science 487: Sjaastad, Larry A "The Costs and Returns of Human Migration." Journal of Political Economy 70(5): Stark, Oded, and David E. Bloom "The New Economics of Labor Migration." American Economic Review Papers and Proceedings 75(2): Stark, Oded, and David Levhari "On Migration and Risk in LDCs." Economic Development and Cultural Change 31(1): Todaro, Michael (1969). "A Model of Labor Migration and Urban Unemployment in Less Developed Countries". American Economic Review, Yang, Dean and HwaJung Choi (2005). "Are Remittance Insurance? Evidence from Rainfall Shocks in the Philippines". Department of Economics, University of Michigan, Mimeo. World Bank (2006) World Development Report 2007: Development and the Next Generation, World Bank: Washington D.C

23 Figure 1: Age Patterns of Migrant Flow Around the World by Gender United States South Africa Kenya Costa Rica Age Males Females Age Males Females Age Males Females Age Males Females Age Males Spain Females Argentina Age Males Females Age Males Mexico Females Age Males Greece Females Cote d'ivoire Canada Portugal Age Males Females Age Males Females Age Males Females Note: United Kingdom not shown due to clumping of age categories in the public use data.

24 Figure 2: Distribution of Age of Return Migration Density Age Mexico South Africa Philippines Albania Pakistan

25 Table 1: Summary of Data Sources Number of Developing Country Immigrants in Sample Country Source Flow measure Total Stock Stock aged 12 to 24 Flow aged 12 to 24 Main sources of youth immigrants Argentina 2001 Census (10% sample) last 5 years Paraguay, Bolivia, Peru, Brazil, Chile Canada 2001 Census (2.7% sample) last 2 years India, China, Middle East, South America Costa Rica 2000 Census (10% sample) last 2 years Nicaragua, Panama, Colombia, Cuba Cote d'ivoire 2002 ENVM (household survey) last 5 years Burkina Faso, Mali, Ghana, other Africa Greece 2001 Census (10% sample) last 2 years Albania, Romania, Georgia, Bulgaria Kenya 1999 Census (5% sample) last year Uganda, Tanzania, Ethiopia, Somalia Mexico 2000 Census (10.6% sample) last 5 years Guatemala, Colombia, Cuba, Honduras Portugal 2001 Census (5% sample) last year Brazil, Angola, Cape Verde, Mozambique South Africa 2001 Census (10% sample) last 2 years Mozambique, Lesotho, Zimbabwe, Botswana Spain 2001 Census (5% sample) last 2 years Ecuador, Morocco, Colombia, Venezuela United Kingdom 2001 Census (3% sample) last year Rest of Asia, Africa, Eastern Europe, India United States 2000 Census (5% sample) last 2 years Mexico, India, El Salvador, Guatemala TOTALS Sources: Argentina, Costa Rica, Greece, Kenya, Mexico, Portugal are from the IPUMS-International, Minnesota Population Center (2007). Canada data obtained directly from Statistics Canada Cote d'ivoire household survey data obtained from the World Bank South Africa data obtained directly from Statistics South Africa. Spain data obtained from the Instituto Nacional de Estadistica (INE), United Kingdom data are from the sample of anonymised records (SARS) provided by the Cathie Marsh Centre for Census and Surveys Research, University of Manchester United States data are from IPUMS (Ruggles et al., 2004).

26 Table 2: Share of Developing Country Immigrants who are Aged 12 to 24 Males Females Males & Females Flow Stock Flow Stock Flow Stock Argentina Canada Costa Rica Cote d'ivoire Greece Kenya Mexico Portugal South Africa Spain United Kingdom United States Unweighted Average Table 3: Share of Developing country youth Immigrants who are female All 12 to to 24 Flow Stock Flow Argentina Canada Costa Rica Cote d'ivoire Greece Kenya Mexico Oman n.a n.a. Portugal South Africa Spain United Kingdom United States Unweighted Average Note: Share for Oman is for 10 to 24 year olds.

27 Table 4: Proportion of Flow of Youth Migrants Living with a Parent and Married by Age Living with a Parent Married 12 to to to to to to 24 Argentina Males n.a Females n.a Canada Males n.a Females n.a Costa Rica Males Females Cote d'ivoire Males s.s s.s Females Greece Males Females Kenya Males Females Mexico Males Females Portugal Males Females South Africa Males Females Spain Males Females United Kingdom Males Females United States Males n.a Females n.a Notes: See data appendix for definition of recent migrant Data for UK is for groupings 12 to 14, 15 to 19, and 20 to 24 due to SARS age groupings. n.a. denotes that marital status is not asked for this age group s.s denotes small sample.

28 Table 5: Proportion of Recently Arrived (Flow) Youth Migrants Attending School, Working or Neither by Age Attending School Working Not in School or Working 12 to to to to to to to to to 24 Argentina Males n.a n.a Females n.a n.a Canada Males n.a n.a Females n.a n.a Costa Rica Males Females Cote d'ivoire Males s.s s.s s.s Females Greece Males n.a. n.a. n.a n.a. n.a. n.a. Females n.a. n.a. n.a n.a. n.a. n.a. Kenya Males Females Mexico Males Females Portugal Males n.a n.a Females n.a n.a South Africa Males Females Spain Males Females United Kingdom Males n.a n.a Females n.a n.a United States Males n.a n.a Females n.a n.a Notes: See data appendix for definition of recent migrant School attendance only asked for those 15 and over in Canada, so presumed at 100% for those under 15. Data for UK is for groupings 12 to 14, 15 to 19, and 20 to 24 due to SARS age groupings. n.a. denotes not asked for this age group. s.s. denotes small sample.

29 Table 6: Top 3 Occupations for 18 to 24 Year Old Recent (Flow) Migrants Males Females % of recent % of recent youth youth Job migrants Job migrants Argentina Construction/Building worker 19.0 Domestic Service 63.9 Manufacturing worker 12.0 Sales 6.2 Self-employed construction 8.0 Restaurant worker 3.0 Canada Sales & Service occupations 21.6 Sales & Service occupations 19.8 Machine operators 13.2 Retail trade sales 17.8 Clerical occupations 9.0 Machine operators 12.2 Costa Rica Agricultural day laborer 32.4 Cleaning personal 56.1 Mining and construction day laborer 18.2 Restaurant Worker 9.8 Construction Laborer 7.7 Shop Saleswomen 7.0 Cote d'ivoire Agricultural Laborer 26.5 Domestic Worker 50.5 Self-employed non-agriculture 21.0 Self-employed non-agriculture 21.8 Domestic Worker 17.5 Farmer 11.9 Greece Agriculture and fishing 20.2 Domestic help and cleaning 32.4 Mining and construction 10.8 Agriculture and fishing 14.2 Builder 10.5 Waitress 8.2 Portugal Mining and construction 25.4 Housekeeping and restaurant work 28.7 Builder 22.3 Domestic worker and cleaner 22.8 Restaurant worker 4.2 Shop Saleswomen 11.8 South Africa Agricultural Laborer 21.2 Domestic Worker 23.4 Construction Worker 8.0 Agricultural Laborer 17.7 Shop Salesman 7.4 Street Vendor 8.5 Spain Agricultural Laborer 21.6 Domestic Worker/Cleaner 35.4 Construction Worker 12.0 Restaurant Worker 15.5 Skilled Construction 9.5 Agricultural Laborer 8.3 United Kingdom Elementary admin and service 17.9 Elementary admin and service 18.1 Science and Technology Professionals 11.9 Caring Personal Service Occupations 16.3 Sales 10.5 Administrative Occupations 10.1 United States Construction Worker 7.7 Cashier 8.3 Agricultural Laborer 7.4 Maid 6.3 Grounds Maintenance 5.8 Shop Saleswomen 4.6 Notes: Percentages are Conditional on Working.

30 Table 7: Herfindahl Indices of Occupational Concentration Flow of Stock of Flow of Number 18 to to to 50 Distinct Migrants Migrants Migrants Occupations Argentina Males Females Costa Rica Males Females Greece Males Females Portugal Males Females South Africa Males Females Spain Males Females United States Males Females Notes: Indices are calculated for workers within each group.