EQUITY & DEVELOPMENT World Development Report 2006 Background Papers Migration and inequality

International migration is a powerful symbol of global inequality, whether in terms of wages, labour market opportunities, or lifestyles. Millions of workers and their families move each year across borders and across continents, seeking to reduce what they see as the gap between their own position and that of people in other, wealthier, places. In turn, there is a growing consensus in the development field that migration represents an important livelihood diversification strategy for many in the world’s poorest nations.1 This includes not only international migration, but also permanent, temporary and seasonal migrations within poorer countries, a phenomenon of considerable importance across much of Africa, Asia and Latin America. 
 
Yet it is also clear that migration - and perhaps especially international migration - is an activity that carries significant risks and costs. As such, although migration is certainly rooted, at least in part, in income and wealth inequalities between sending and receiving areas, it does not necessarily reduce inequality in the way intended by many migrants. Much depends on the distribution of these costs and benefits, both within and between sending and receiving countries and regions. Also important in terms of the aggregate impact of migration on sending societies is the selectivity of migration itself. Clearly if most migrants were to come from the poorest sections of society, and they were to achieve net gains from migration, this would act to reduce economic inequality at least, all other things being equal. But migrants are not always the poorest, they do not always gain, and other factors are not equal.


Introduction
International migration is a powerful symbol of global inequality, whether in terms of wages, labour market opportunities, or lifestyles. Millions of workers and their families move each year across borders and across continents, seeking to reduce what they see as the gap between their own position and that of people in other, wealthier, places. In turn, there is a growing consensus in the development field that migration represents an important livelihood diversification strategy for many in the world's poorest nations. 1 This includes not only international migration, but also permanent, temporary and seasonal migrations within poorer countries, a phenomenon of considerable importance across much of Africa, Asia and Latin America.
Yet it is also clear that migration -and perhaps especially international migration -is an activity that carries significant risks and costs. As such, although migration is certainly rooted, at least in part, in income and wealth inequalities between sending and receiving areas, it does not necessarily reduce inequality in the way intended by many migrants. Much depends on the distribution of these costs and benefits, both within and between sending and receiving countries and regions. Also important in terms of the aggregate impact of migration on sending societies is the selectivity of migration itself. Clearly if most migrants were to come from the poorest sections of society, and they were to achieve net gains from migration, this would act to reduce economic inequality at least, all other things being equal. But migrants are not always the poorest, they do not always gain, and other factors are not equal.
Over twenty years ago, Michael Lipton argued that not only was inequality a major cause of rural-urban migration, but that its 'after-effects' -remittances and return migration -tended to increase interpersonal and inter-household inequality within and between villages. 2 Lipton's observations, based on rural micro-evidence primarily but not exclusively in India, include a number of important insights. For example, he argued that inequalities within sending areas were also crucial in generating migration -more unequal villages send more migrants. In turn, the fact that migrants come from the most productive age groups, that unequal power structures within villages go unchallenged as a result, and that migrants from wealthier backgrounds do better, all conspire to ensure that migration enhances inequality.
Since 1980, further studies have in some cases supported Lipton's work, but in other cases have come to somewhat different conclusions. 3 One reason for such differences is that of methodological variation -which specific economic question is being asked, and the econometric or statistical techniques used to generate estimates of income and income distributions. For example, if remittances are treated as an exogenous transfer, the economic EQUITY & DEVELOPMENT World Development Report 2006 Background Papers question is how remittances, in total, or at the margins, affect the observed income distribution in the sending community. However, if remittances are treated as a potential substitute for home earnings, the economic question becomes how the observed income distribution compares to a counterfactual scenario in which no migration takes place, but including an imputed level of home earnings. A study of migration from Nicaragua using both of these methods suggests that where remittances are considered exogenous, they reduce income inequality, whereas if they are considered a substitute for home earnings, they increase income inequality. 4 Given such difficulties, this paper does not seek to re-affirm or draw a new definitive conclusion on whether migration does or does not act to reduce inequality. Instead, we seek to make three key points. First, given the range of different types of migration, and the varied economic, social, cultural and political contexts within which migration occurs, any overarching conclusion about impacts on inequality is unlikely to be very robust at a global or even a regional level.
Indeed, examples can be found of migration both increasing and decreasing inequality. 5 With this in mind, we seek to explore particular kinds of relationship between migration and inequality that emerge from case study material, to demonstrate how context-specific these relationships are. Second, through these case studies, we also argue that inequality needs to be defined in broader terms than simply income or wealth. Inequality, like poverty, is multidimensional, and can be measured at individual, household, regional and international levels. There are socio-cultural dimensions to inequality, as well as inequalities in access to power, whilst all aspects of inequality are highly gendered.
Finally, we argue that central to outcomes in terms of all of these dimensions of inequality is the role played by a variety of political, economic and social-cultural institutions, since these are often crucial to the ways in which wealth, power and opportunity are distributed within societies. Whilst this partly leads us to consider the kinds of institutional policies and arrangements that can promote more equitable outcomes from migration, we also argue that in an important sense, migration itself can be viewed as an institution that directly structures outcomes for different groups.

Case studies
The paper focuses on case studies across Central America, Eastern Europe, West Africa and South Asia. The case of migration from Mexico to the US, which constitutes one of the largest and probably best-studied of all international migration flows in recent decades, is particularly useful in terms of direct measurement of the effect of emigration on economic inequality. In contrast, analysis of Albania, a country that has undergone a dramatic turnaround from virtually zero emigration in 1990, to a situation today where as many as one in five of the population is estimated to be living abroad, and West Africa, where complex rural-urban, rural-rural and other flows of poor people both within and between countries have generated highly varied outcomes, points to the need to consider different types of migration flow separately.
Then, we turn to three sections on migration in South Asia, which reinforce the need to disaggregate flows, whilst also emphasising the variety of forms of inequality that are influenced by migration. Thus, ethnographic evidence on long-distance migration from Bangladesh and Kerala provides detailed evidence of impacts on social as well as economic inequality; in-depth studies of internal migration in West Bengal and Western India suggest highly contrasting findings on the emancipatory effect of migration on workers; whilst evidence EQUITY & DEVELOPMENT World Development Report 2006 Background Papers from Bangladesh and Sri Lanka considers the extent to which migration flows dominated by women have an impact on gender inequality.
Following these case studies, the paper turns to policy interventions that might influence the impact of migration on inequality, focusing on policies that target institutional arrangements and the way that these impact the inequality-migration-inequality relationship. This may not be the only way of influencing outcomes, but is certainly a more fruitful way forward than simply trying to stop -or indeed encourage -migration itself.

Migration and inequality across the US-Mexico border
The US-Mexico border is the longest between a developed and developing country in the world and there is a long history of migration between the two countries, which can be divided into three significant periods: • The first wave began shortly after the turn of the century, and peaked in the 1920s. This period, when a post-war economic boom in the US was matched by economic devastation in Mexico following the revolution , ended in 1929 with the onset of the Great Depression; • The second wave began in the 1940s and peaked in the late 1950s. This period led on from a series of droughts in Mexico in the 1940s, and was fuelled by US labour shortages.
In particular, the latter led to the Bracero program, a temporary employment initiative to recruit Mexicans for agricultural work, which was initiated in 1942 and renewed annually until 1964; • The third wave began in the mid-1960s, and has continued to the present, reflecting maturation of migrant networks in the US, the mechanization of agriculture in Mexico, and growing labour demand in US cities.
In all, some 4.6 million braceros and 565,000 legal emigrants entered the US during the period 1940-64. Although this scheme was for temporary mobility, not all of them returned, whilst undocumented immigration may have exceeded the number of legal visa holders during this period. 6 Subsequently, Mexican emigration has grown steadily, so that there were nearly 10 million Mexican-born immigrants in the US by 2003, representing around 10 per cent of Mexico's population. 7 There have been several studies here of the impact of these flows across the US-Mexico border on inequality in rural sending communities. For example, a pioneering study from the 1980s 8 in the Patzcuaro region of the state of Michoacan, approximately 2,000km south of the Mexico-Arizona border found that the impacts of migrant remittances on rural income distribution appeared to depend on a number of factors, including: • a village's migration history, • the degree to which migration opportunities were diffused across village households, • the returns to human capital embodied in remittances, and • the distribution of potentially remittance-enhancing skills and education.
Two villages chosen for analysis were only two kilometres apart and were statistically similar in terms of family size, but the difference in their migration patterns was significant -one had EQUITY & DEVELOPMENT World Development Report 2006 Background Papers many internal migrants, whereas the other had a long history of sending migrants to the US. In the former village, the remittances of internal migrants had a favourable effect on the village income distribution. Yet the remittances in this village from the small number of people who had experienced migration to the US, had a profoundly negative impact on income distribution. In contrast, in the second village, it was US to Mexico remittances that were found to have an equalising impact on incomes in the village, whereas remittances from internal migrants, which included a large returns-to-schooling component, and where education was highly correlated with household income, accounted for a comparatively large share of inequality in this village.
One of the limitations of this study -aside from the fact that it was based on quite a small sample -was the fact that it represented only a single snapshot in time. Yet the impact of migration on inequality may change over time. For example, the most direct effect of migration is to increase the welfare of the migrant households, as migrants bring back remittances and savings from abroad. This can either increase or decrease inequality in the community, depending on where the migrant household is located in the overall community wealth distribution. In the Mexico-US case, evidence presented by Massey et al. from a wide range of migratory experiences 9 suggests that the first migrants usually come from the lower-middle ranges of the socioeconomic hierarchy, and are individuals who have enough resources to absorb the cost and risks of the trip, but are not so affluent that foreign labour is unattractive. In this context, remittances tend to open up a wealth gap between middle-income groups, who become wealthier, and the poor, who do not have access to migration.
Once someone has migrated and returned, and that person has direct knowledge of employment opportunities, labour-market conditions, and ways of life in the destination country, they use these understandings to migrate again with fewer risks and costs than before. Every new migrant thus reduces the costs and risks and increases the attractiveness and feasibility of migration for a set of friends and relatives, contributing further to inequality. However, the gradual accumulation of network connections and migratory knowledge across developmental stages creates spill-over effects, which make migration an increasingly common social and economic practice, lower the costs and risks of movement, and make migration less selective. 10 For example, as more people migrate, the mean cost of the coyote (smuggler) to cross the border declines, whilst more households move above the income threshold that constrains migration.
The precise relationship between migration and inequality in sending villages will clearly depend in part on the initial distribution of income. Initial wealth plays two key roles in determining whether a given individual will migrate: increases in wealth generally raise the returns to domestic production, increasing the opportunity cost of migrating, but also relax credit constraints which restrict access to costly migration. As wealth increases, the maximum number of migrants a given household can afford increases, but the optimal number decreases. Migration rates will thus first increase and then decrease with wealth -what is known in the literature as the 'migration hump' in aggregate terms. 11 Yet patterns of community networks also matter a great deal in the decision to migrate.
McKenzie et al. have found that the proportion of other households with migrant experience two years prior to the survey year had a significant positive effect on the probability of migration. 12 Where migration costs are high to begin with, the first network effects tend to increase migration opportunities more for the middle and upper-middle classes. However, as migration EQUITY & DEVELOPMENT World Development Report 2006 Background Papers costs continue to fall through the building of a larger network, this benefits primarily the lower and lower-middle classes in the village, which will tend to reduce inequality. This process gives rise to an inverted U-shaped relationship between migration and inequality.
McKenzie et al. demonstrate this relationship using two data sets: the first consists of data from 57 rural communities in Mexico collected as part of the Mexican Migration Project (MMP), while the second consists of data on 97 rural municipalities from the Encuesta Nacional de la Dinamica Demografica (ENADID). 13 The MMP surveyed five communities in 1982, between two and five communities each year from 1987-97, and fourteen communities in 1998. In general, 200 households were surveyed in each community, with smaller samples taken in communities with less than 500 residents. The study relies on the MMP71 database (71 communities). Since the theoretical model applies best to rural communities and small towns, McKenzie et al. restrict most of the analysis to the 57 communities which had a population below 100,000 in 1990. Both data sets provide detailed information on international migration, but did not include income or consumption data, so inequality is measured instead in terms of data on households' ownership of infrastructure and assets. Although this data does not show increasing inequality at low migration prevalence rates, it does show that an increase in migration prevalence is followed by a decrease in inequality.
The ENADID survey is a national demographic survey conducted in 1992 and 1997 by Mexico's national statistical agency, the Instituto Nacional de Estadística, Geografia e Informatica (INEGI). Approximately 2,000 households were surveyed in each state, with a total sample size of 57,017 households in 1992, and 73,412 in 1997. The ENADID data does show an inverted-U relationship between international migration and inequality, with emigration increasing inequality at lower volumes, and then reducing inequality as volume approaches the levels prevailing in the MMP communities. In particular, the community migration prevalence 14 15 years before, which serves as a proxy of the stock of migration experience in the community, is shown to have significant negative effects on community inequality. This effect is strongest and most significant for asset inequality, but also significant for income and consumption inequality.
If it is true that the inequality-reducing effects of emigration increase over time, as the prevalence of migration becomes more widespread, and potential migrants are able to lower the costs of migration, then one potential policy conclusion is that steps could be taken to enhance this process. However, the institutional and policy framework within which Mexican migration to the US takes place is far from moving in that direction: on the contrary, in response to anti-immigration rhetoric in parts of the US, and fears over the security of the border since 9-11, significant steps have been taken to strengthen the border and limit movement, with serious consequences for the welfare of migrants. 15 An additional insight from the Mexican case concerns the implications of relative and absolute income motives for migrating. 16 Using the Michoacan data referred to above, Stark and Taylor argue that relatively deprived households are more likely to engage in international migration than are households more favorably situated in their village's income distribution. Relative deprivation on the contrary does not have a significant (direct) effect on internal migration, which is more affected by absolute poverty. These findings have interesting implications for EQUITY & DEVELOPMENT World Development Report 2006 Background Papers policy within Mexico. First, a distribution-neutral development policy that increases incomes across the board would reduce the incentive to engage in internal migration for all but the richest households, who have little economic incentive to migrate anyway. However, a distribution-biased policy in Mexico leading to a more equal income distribution could tip a migration balance from international migration to internal migration. 17 Migration and regional inequality: the case of Albania Like Mexico, Albania has experienced considerable migration over the last decade, both across the border to neighboring Greece and Italy, and internally to the main cities of Tirana and Durres. However, unlike Mexico, emigration was illegal from 1945 to 1990, whilst the government restricted urban growth from 1961 onwards, with the result that in 1990, there was virtually no experience of international migration, whilst 64 per cent of the population was located in rural areas, compared to just 27 per cent across Europe as a whole.
This helps to make the massive emigration of Albanians -an estimated 600,000 in 2001, most of them in Greece -and substantial internal movements since that time all the more remarkable. Albanian migrants working abroad -about one in five to one in six of the country's population -are estimated to remit anywhere between $300 million and $1 billion each year, making remittances the country's largest source of external income after aid. 18 Migration is seen by most Albanians as the most effective way of coping with the country's difficult economic conditions, in which over 780,000 individuals -nearly a quarter of the populationwere still below the poverty line in 2002. 19 The Human Development Report of 2000 estimates that 60 per cent of new apartments in the country have been built with remittance money. 20 The transfer of such significant resources might be expected to have helped reduce inequality at an international level. For example, Albania has moved up the UNDP's Human Development Index ranking from 105th in 1995 to 65th in 2003 21 , and its GDP per capita growth rate from 1993-1996 was reported to be consistently over 9 per cent per annum. 22 However, over the period 1990-2002, its per capita GDP has remained at just under 10 per cent of the per capita GDP of its nearest neighbor and principal migrant destination country, Greece, whilst Albania's HDI improvement at least in part reflects statistical adjustments rather than real improvements over the period. Worse, the collapse in 1996 of pyramid schemes in which large numbers of migrants had invested their money led to losses of over $1.2 billion, or more than half of total GDP that year. More generally, there is evidence that migrant remittances tend to be invested in imported items, whilst there appear to have been negative effects on agricultural production, with negative consequences for Albania's balance of payments. 23 Evidence on the impact of migration on inequality within Albania is also mixed, and requires us to disaggregate at least four distinct migration streams: • internal migration, especially to Tirana and Durres, although also to smaller urban areas within the mountain and central regions • temporary external migration to Greece, which was the first and largest external migration stream, and involves migrants mainly from rural areas in the north, centre and south of the country; • temporary, and some permanent migration to Italy and other European destinations, which is dominated more by migration from urban and rural coastal areas; and

EQUITY & DEVELOPMENT World Development Report 2006 Background Papers
• a rising trend of permanent migration further afield, notably to the US and Canada.
Analysis of the Albania Living Standards Measurement Survey (LSMS) suggests some interesting potential conclusions about the effect of these flows both on intra-household inequality, and on regional inequality. Whereas individuals who migrate internally or go to Greece are generally from rural and poorer households, those who go to other European destinations and beyond are generally from better-off and/or urban families, with higher levels of education. 24 In turn, per capita remittances vary significantly across the four groups, with internal migrants remitting the least, and those outside Europe the most (Table 1). According to the Albania Poverty Assessment, this suggests a 'migration-poverty trap', in which poorer people from rural areas move internally or to Greece, from which they gain less that the better off living in urban and coastal areas, who are more able to access better-remunerated migration opportunities to Italy and beyond. As a result, remittances are estimated to make up 14 per cent of households' total incomes in non-poor households, but just 8 per cent in poor households, whilst remittances are a higher proportion of total income in urban areas (16 per cent) than in rural areas (11 per cent). There is also some evidence that these patterns of migration and remittances have enhanced regional inequality in Albania; indeed, migrants abroad tend to concentrate productive investments and much investment in housing in the major cities of Tirana and Durres, regardless of their original region of origin. This in turn may have fuelled the process of rural-urban migration amongst family members. 25 Slightly contradictory evidence is provided by another conclusion of the Albania Poverty Assessment that remittances are highest, in both aggregate and per capita terms, in rural mountain areas, where poverty is greatest. 26 This may be partly explained by the fact that the total volume of migration is higher in rural mountain areas, with a higher proportion of families than elsewhere having two or more migrants. Meanwhile, the inequality-producing effects of migration are probably mitigated somewhat by the levels of migration in Albania, which are high amongst all groups. For example, even marginalized ethnic groups such as the Roma and Evgjit communities are able to engage in international migration and capture a share of remittances, and this tends to limit the growth of income inequality, whilst also constituting an important and effective way of coping with social exclusion. 27 In addition to income and social inequality, the Albanian case also points to some interesting conclusions in relation to gender inequality. Overall, international migration from Albania is dominated by young men, reflecting in part the strongly patriarchal nature of Albanian society. King and Vullnetari note that family structures occupy an important place in the migration process, but are predicated on patriarchal authority in which the person receiving remittances and administering the family budget is usually the male head of household. 28 29 This would suggest that migration does little to reduce gender inequalities -indeed, Mai has argued that since 1990, young Albanian men have 'reinvented' aspects of Albanian patriarchy to serve their own needs and exploit young women in new ways, including through trafficking for sex work in Europe. 30 However, it is also worth noting that data from the LSMS suggest that amongst children of households who have left home, a significantly higher proportion are women within flows outside of Europe (43 per cent), in comparison with Greece (38 per cent) or the rest of Europe (31 per cent). This higher involvement of women in long-distance migration, which holds out the prospect of better economic rewards, is a small indication that for some flows at least, patterns of gender inequality may be eroding.

Migration and intra-village inequality in West Africa
The previous two case studies relate to strong emigration flows from poorer to richer countries that have responded primarily to economic differentials, but it is important to bear in mind that migration is a much broader phenomenon that this. In West Africa, for example, migration has probably taken place in response to drought, climatic variability and low and fluctuating agricultural productivity for centuries, became rather more prominent during colonial times as plantations, mines and new administrative centers created important poles of attraction for workers, and since independence, has ebbed and flowed, in response to both economic expansion and contraction, and to political and social upheavals and associated violent conflict. New, intercontinental, flows have also become much more important, both to former colonial metropoles in Europe, and increasingly to North America and elsewhere.
A distinction between different forms of migration is often necessary in seeking to understand its consequences, and this is especially true in West Africa. An illustration of this is provided by recent work that highlights the relationship between migration and inequality in Burkina Faso, based on a survey of 223 rural households in four villages in the Central Plateau in early 2003. 31 Two villages, Niaogho and Beguedo, are in the more isolated southern part of the plateau 32 , whilst two more, Boussouma and Korsimoro, are in the north, located next to the main road to the capital. In all four villages, household members were found to engage in migration, with continental migrants generally consisting of young men who leave to attempt to find work elsewhere, mostly in Côte d'Ivoire, or in the capital, Ouagadougou. This type of migration was found not to result in remittances, but did reduce consumption pressures faced by the sending household. 33 Intercontinental migration, on the other hand, was in nearly all cases embarked on by young men who leave for southern Italy to engage in horticulture. These men tend to leave alone but may send for their wife and children to come over at a later stage. This form of migration requires high entry costs particularly in the form of transport to the destination.
According to this study: • Households with continental migrants tend to be in the middle-income groups whereas intercontinental migration takes place mainly in the highest income group; • Households with intercontinental migrants earn much more in remittances compared to households with continental migrants; • Households with only continental migrants are less well off in terms of income per capita from agriculture and non-farm activities compared to households with intercontinental EQUITY & DEVELOPMENT World Development Report 2006 Background Papers migrants. This reflects the fact that remittances from continental migration are found to be comparatively low; • Households with intercontinental migrants are better off in terms of productive assets, land and livestock compared to households with continental migrants and households without migrants.
All of this supports the hypothesis that intercontinental migration should be seen as an accumulation strategy undertaken by comparatively wealthy households, leading to increased intra-village inequality. In contrast, less well-endowed households may be willing, but are not able to engage in this form of migration. Instead, they are seen as pursuing continental migration primarily in response to a lack of work and insufficient income.
However, a set of more qualitative studies of migration and mobility in Burkina Faso and neighbouring countries suggest a number of nuances to the above narratives, and other ways in which migration can impact on inequality. First, whilst intercontinental migration is perhaps more likely to lead to accumulation by migrant households than continental migration, this is not necessarily the case, and where migrants to Italy are not successful and do not remit, this can have serious consequences in terms of the upfront costs of such a move. In turn, the ability of people to migrate successfully may depend less on whether they are poor or non-poor, and more on their access to social networks, whether they are migrating to the capital city, a neighbouring country, or to Europe. 34 Second, although continental migration may be a measure to diminish consumption needs, especially in years with low rain or other adversities, its effect can be more problematic, particularly where migrants stay away during the rainy season, with a consequent negative effect on the availability of agricultural labour. 35 In this context, an important further distinction should be between seasonal and more 'permanent' migration, although permanent in this case may include movement of people for a number of years, but followed by eventual return to their village or town of origin. 36 In addition, migration can also be seen as an outcome of the diversity of economic interests within large embedded households common to much of the Savannah belt, and in this sense should be seen less as the result of a 'household decision', and more as a balancing of different interests within the household, or indeed as a result of intra-household contestation. 37 In this sense, migration also impacts on intra-household inequalities, notably between older and younger generations.

International migration and inequality in South Asia
In South Asia, any discussion of the relationship between migration and inequality again needs to take into account the wide range of migration contexts that exist in the region. Because of unequal access to opportunities, international migration in particular often enhances inequality. However, discussion below shows this is not always the case. As in Albania, the region also provides evidence of the consequences of international migration for gender inequalities, in particular where migration flows are dominated by women.
Looking first at international migration from across the region, evidence from both Pakistan 38 and Bangladesh 39 suggests that inequalities are increased, as relatively wealthy individuals and villages that have more access to long-distance migration enhance their position in relation  40 Here, Gardner argues that 'access to bidesh (abroad) has increasingly become the pole around which inequalities are clustered. Not only has it helped to create them, but so too has it become a metaphor for thinking about them'. 41 Her argument relates not only to economic inequalities, but also to broader social and cultural cleavages, with migration becoming one of a number of analytically distinct measures of status and power. Yet although inequality has increased between wealthier households and the very poor, it has decreased between the wealthiest -the elite that used to hold positions of power -and the many poorer households who were often previously dependant on this elite for economic and social support, but have now become much better off.
One reason for this is that the original migrants from Talukpur were by no means the wealthiest -indeed, some were even landless, with land traditionally being the major indicator of wealth in rural Bangladesh. In practice, migration brought with it considerable changes to landownership in Talukpur. Those who already owned land often acquired more fields and moved from being small to large landowners, whilst some of those without land managed to transform their position to become amongst the most prosperous households in the village. In this context, access to international migration, specifically to London, came to result in as much social and economic power as ownership of land. Migration brought more than simply money, in the form of remittances; it also brought with it 'cultural capital' and social prestige. In turn, remittances were also fed into other areas of life, with a wider impact on political, social and economic power.
However, the success of migrants has also worked to increase inequality between those who could, and those who could not migrate. Gardner has argued that: 'those without access to foreign wages have found it increasingly difficult to compete in the struggle for local resources. During the period of most intense migration in the 1960s, when migrants struggled to buy as many fields as possible, local prices rose quickly…. To buy land today, foreign income is crucial'. 42 Indeed, inflation in all sectors of the economy (labour, commodities and technology) has made it difficult for small plots of land to be economically viable. As elsewhere in Bangladesh, many Sylhetis have lost their land and this phenomenon has been intensified in Londoni villages. Yet as polarization grows between migrants and non-migrants, there increasingly appears to be no clear fit between economic position and status classification. The conspicuous consumption of foreign goods, while being the principle differentiation in the village between migrant and nonmigrant households also, according to Gardner, 'reflects the economic dependency of the desh on aid and remittances from overseas and the hegemonic power of international capitalism'. 43 While laying claim to some of this power the migrants are unwittingly pointing to the larger inequalities that structure our world.
In addition to intra-village inequality, some observations can also be made about increased inequality between migrant and non-migrant villages in Sylhet. In particular, villages that have experienced high levels of 'Londoni' migration are now startlingly distinct, filled with stone houses, sometimes two or three stories high, and showing extensive material wealth in contrast to the impoverishment of much of the rest of rural Bangladesh. Education is another clear sign

EQUITY & DEVELOPMENT World Development Report 2006
Background Papers of the increasing inequality between villages, with 'modern' education a major sign of prestige in Talukpur, and levels of literacy in the village much higher -over twice the national average for men, and one and a half times the national average for women, mainly a result of migrant households' increased wealth. 44 Growing temporary migration to the Gulf may also be having similar impacts on inequality, as the poor remain largely excluded from such flows. 45 Turning to Kerala, a rather different research project -the Kerala Migration Study -also provides rich evidence on aspects of the relationship between international and internal migration and inequality in sending areas. Based on a sample survey of some 10,000 households conducted in 1998, it is probably the largest study of its kind in the region. It shows that over the last decade at least, Kerala has experienced a massive increase in migration, which has become one of the most prosperous 'industries' in the state. 46 The authors go on to assert that 'migration has contributed more to poverty alleviation and reduction in unemployment in Kerala than any other factor', with the proportion of the population below the poverty line declining by 12 per cent, and the unemployment rate falling by 30 per cent in the last 25 years. 47 Some support for this comes from the fact that in 1998, remittances made up nearly 10 per cent of the State's GDP, over twice what the state received from the central government by way of budget support, and well over the combined export earnings of the state's two major industries, sea food and spices.
The Kerala Migration Study compares opportunities and living standards between non-migrant households, inter-state migrants who stayed in India, and international migrants, most of whom went to the Gulf. As in other cases, it shows that migration is generally selective of young, educated men, although the proportion of women in both emigrant and inter-state migrant flows was found to be rising. Interestingly, it shows migration patterns that differ remarkably between different religious and ethnic communities -thus Muslims were twice as likely to migrate internationally to the Gulf than the average, whereas Syrian Christians were twice as likely to migrate within India than the average. In contrast, members of scheduled castes were less likely to migrate to either destination. Also, although degree holders were much more likely to migrate than non-degree holders, the reverse was true for those whose fathers were educated. 48 Unfortunately, the study does not allow us to look directly at changing economic inequality, since the economic position of migrants and migrant households prior to migration is not recorded. It does show that those who migrate do better than those who do not. For example, migrant households had higher levels of food consumption, were more likely to have electricity, to own land, and to have invested in land improvement, and were more likely to use private hospitals. However, per capita remittances also appear to follow the educational level of migrants prior to departure, with the most educated sending home the highest amounts. This is reflected in a better performance of Syrian Christian internal migrants, who tended to be better educated, compared to Muslim migrants to the Gulf, whose educational levels were lower. Meanwhile, anthropological evidence from Osella and Osella tends to support the conclusion that the poorest have been excluded from economic investments made by migrant workers. 49

Seasonal labour migration and inequality in India
Our fifth case study concerns the complex, and in some respects contradictory story to be learned from recent studies of internal migration in West Bengal and Western India. In West Bengal, Rogaly 50 Rogaly and Rafique consider the crucial role of remittances in the context of growing inequality especially for the landless households, noting that there are very few cases in which seasonal migration has led to individual upward economic mobility. 51 In Murshidabad district in particular, men usually travel unsolicited, and without prior agreement with an employer and thus face great uncertainty. Almost all migrant households use their cash remittances for food and loan repayments. Seasonal migrants are excluded from government provision of health services and education when away for work. Combined with the nature and intensity of the work and the living conditions, this means that the health of the migrants suffers more than those who stay behind. 52 However this form of seasonal migration in West Bengal has also increased the migrants' aggregate power in the labour market and changed long-standing relations of power more in favour of the labourers themselves. For example, in Puruliya District, there has been a significant shift in relations between the landowning and labouring classes. The increased access to seasonal out-migration for agricultural and other manual workers has contributed to the diversification of livelihoods by these workers and a consequent loosening of their obligations to the landowners. The result has been labour shortages during peak agricultural seasons, leading to changes in the terms of land and labour contracts in the region to the advantage of labourers. 53 These labour market adjustments have allowed workers more choice regarding their employers, giving them more power to gain increased benefits or higher wages.
In contrast, in Western India, the scope for such worker choice is much more limited, since employers often act together. This is the case in Bardoli taluka in South Gujarat, where sugar cultivators, predominantly from a single social group, maintain a 'highly organised recruitment process', reducing the extent of benefits derived by labourers. 54 Meanwhile, Mosse et al. conclude that whilst for a small minority of Bhil migrant households in tribal Western India, there are benefits in the form of saving, investing and meeting contingencies, for the poor majority, migration is again a defensive coping strategy that simply allows them to combat the extreme economic vulnerability. 55 The uniformly poor appearance of Bhil villages disguises significant differences in wealth, status and power within them, which have a bearing on the organisation and outcome of seasonal labour migration. According to Mosse et al., 'it is clear that migration is strongly determined by existing social inequalities (within and between villages), and also has the effect of amplifying these' 56 .
In this region, the success of migration is highly dependent on access to recruiter networks or kin and without these networks the migrant is often left with the worst paid, less reliable forms of labour and are in the most vulnerable positions. Indeed, more generally, the system of recruitment here amplifies debt and dependence and prevents most migrants from working their way out of debt. In this context, Bhil migrants working for the same amount of time end up with very different outcomes. For most, many years of seasonal migration have not led to any long-term increase in assets or reduction in poverty. Migration can also lead to a loss of social position and status if the migrant is absent for a long time and fails to take part in important village events. This will increase marginality from credit and social networks.

EQUITY & DEVELOPMENT World Development Report 2006 Background Papers
However, in a few cases migration can increase a household's earning capacity and its creditworthiness and allow new investments in machinery or land or in social networks to increase social position and for a very few, migration provides the means for upward mobility. For a number of households with greater livelihood security, 'labour migration strengthens their position, especially where labourers can become (through kin networks) gang leaders and "recruiters"'. 57 There are several reasons why different outcomes in terms of inequality can be observed between the two studies of rural-rural migration in West Bengal and rural-urban migration in Western India. In West Bengal, migrants travel in groups to labour recruiting areas where employers and employees collect and here bargaining for wages and benefits takes place. If the migrants are lucky enough to arrive on a day with a high employer to employee ratio then their bargaining power can be quite high. Indeed, there are even cases that are reported of migrant labourers foot-dragging or walking out of a job, in the knowledge that alternative opportunities might exist.
In contrast, in the case study of tribal Western India by Mosse et al., seasonal migration is from rural to urban areas. Within the urban informal labour market work is highly differentiated along 'ethnic' lines and the Bhils occupy a distinct employment sector that is low paid and unskilled. They are unwilling or unable to risk competing in other labour markets, and instead are recruited directly from their villages. In this situation, much more than in the case of West Bengal, it is neighbourhood and kin networks that are involved in the structures of employment and this reduces competition between employers and keeps wages low. It also removes the opportunity for mobility that is seen in the West Bengal case and ensures the employer a carefully selected, accountable and amenably dependent workforce.
The degree to which workers can make choices in the West Bengal labour market would not be possible in such a structured casual labour market with a socially and economically dependant workforce as in Western India. Thus unlike West Bengal, the high demand for casual labour does not translate into stronger bargaining power for Bhil migrants. 58 This means that it is also less likely that migration can have an emancipatory effect. On the contrary, Mosse et al. argue that 'through migration these workers move into a world in which hierarchical distinctions (Bhil/non-Bhil; tribal/non-tribal) are more keenly felt, generalised and amplified; a world in which the culture of poverty is newly experienced as jati (caste/'ethnic') discrimination '. 59 In this situation, links to recruiters and their kin are crucial and the labour recruitment process can ensure unequal access to earning. Those who have no stable relationships or networks necessary to get regular work are pushed into the most casual and poorly paid work. Also without such networks, the migrant will have problems finding living arrangements, they are more vulnerable to intimidation, under or non-payment of wages and isolation and loneliness.
In addition, migration in the Western India case appears to have done little to emancipate Bhil women. Gender structures tend to remain intact and women have to continue domestic roles alongside waged labour. Women and older people have a limited participation in labour migration and the enhanced dependence on migrant earnings increases male authority in the household, increasing dependence on men for cash and health care. In this case, migration may amplify gender inequalities.

EQUITY & DEVELOPMENT World Development Report 2006 Background Papers
Discussion of the emancipatory effect for women in West Bengal is more complicated and reveals a number of different changes that vastly depends on the region within the state, and the socio-cultural background. The West Bengal study focused on four different migration streams, three of which involved family groups with at least as many women migrating as men.
The fourth, though, in Murshidabad district, differs from much of West Bengal in that it is a Muslim majority district; this stream consisted almost exclusively of men. Although here women face extreme dependency on kin networks and hardship in coping with limited rice and often no money, some also reported that they were left to make expenditure decisions on remittance money. 60 The case of West Bengal reveals that seasonal labour migration is neither necessarily emancipatory nor oppressive for the women involved and that changes to intra-household relations are often mediated through the distribution of remittances, which varies between and within migration streams. Through the migration process, gender identities and ideologies of work are variously (re)constructed, reinforced and changed and in some cases intra-household power relations along lines of gender and age have been successfully contested. 61 Although some women in employer households in West Bengal are involved in distributing in-kind payments, it is men who are responsible for almost all the recruitment in these households. In turn, women in both employer and employee households often also have to take on a number of additional, unwelcome chores.

Migration and gender equality in South Asia
Our final case study concerns the impact of migration on gender equality in South Asia in the context of movements that have been more heavily dominated by women. In India, the ruralurban migration of Darana women in Jaipur has allowed them greater physical freedom and an opportunity to escape indebtedness and excessive demands within extended families, but also also brought greater workloads. 62 In contrast, in Calcutta there is evidence that women experience greater physical restrictions in cramped urban slums. 63 Yet it is the cases of Bangladesh and Sri Lanka where the effects of female migration on gender inequality are perhaps best documented.
In the case of Bangladesh, there has been significant internal migration of women from poor rural areas to towns and cities in search of work, and more recently, especially into the exportorientated garment industry both within and outside two 'Export Processing Zones', one in Dhaka and one in Chittagong. According to Kabeer and Mahmud, EPZ workers tend to come more from the relatively prosperous end of the economic spectrum, whilst other wage workers come from the poorer end. 64 EPZ workers were also less likely to be landless, fewer had experienced food shortages in the past and they had higher levels of education. This suggests that between the migrant women themselves, economic inequality remains constant or is increased, as returns to labour mirror the initial economic ranking of the workers. 65 Moreover, employment in the garment industry, even in the EPZs, is not necessarily secure, with many garment employers regarding their female labour force as dispensable in the face of an apparently limitless supply of new workers from the countryside. 66 However, the Dhaka garment factories have recruited workers from poorer sections of the rural population, where women had previously been 'marginalized from mainstream forms of employment and confined to casualized and badly paid work in a limited number of occupations or to unpaid family labour'. 67 As a result, there has been an improvement in working conditions

EQUITY & DEVELOPMENT World Development Report 2006
Background Papers for these women, even if the benefits have been greater for more educated women. Moreover, the very existence of this labour migration can be held to have challenged the widely-held myth of the male breadwinner in Bangladesh, offering women the chance to support themselves and improve their economic condition within the household. Indeed, Kabeer writes of the 'apparent ease with which women appear to have abandoned old norms in response to new opportunities in what has traditionally been presented as one of the least negotiable patriarchies in the world. ' 68 In Sri Lanka, the dynamics of labour migration are somewhat different, as since the late 1980s, women have been encouraged to migrate internationally. By the late 1990s, women made up some 84 per cent of Sri Lankan migrants to the Middle East 69 , with the majority working as domestic workers in Kuwait and Saudi Arabia, whereas their male counterparts were employed as skilled workers. A study by IOM/INSTRAW suggests that although migrant men earned considerably more than women in the Gulf, saved more, and had greater control over their earnings, migrant women had gained more control over money, and were able to invest in small enterprises such as grocery stalls and animal husbandry. 70 The study also notes that female migration had altered gender roles within the family, with an increase in the number of decisions made jointly by husband and wife, a decrease in the husband's involvement in housing decisions, and a moderate increase in men's participation in decisions concerning food, clothing and health. Some migrant women were reported to have gained greater levels of responsibility in disciplining children, enhanced social status, or even had become the main providers for their families -although this was not the case for all women, and these changes were not always viewed as positive. In contrast, in non-migrant households, women's reproductive role remained the defining characteristic of their identity and status.
Meanwhile, an ethnographic study of the migration of Muslim women from Eastern Sri Lanka to the Gulf also highlights the enhanced economic and social status of migrant women, in particular their ability to access consumer goods and material wealth. 71 However, he argues that in part, women's increased freedom of movement and the respect and authority accorded to them has gone hand in hand with a deepening of orthodox Islamic views on women's seclusion. Indeed, he argues that these migrant women are 'successfully negotiating and recasting their roles by utilizing those very same religious discourses and practices which attempt to regulate them', noting that Islamic orthodoxy for these women becomes associated with sophistication, social respect, mobility and modernity. 72

Migration, inequality and institutions
A major thrust of the WDR 2006 approach to inequality is that institutions are critical in influencing the extent to which development either increases or decreases inequality. Yet we would argue that it is increasingly possible to view migration itself as an 'institution', in the sense of constituting a set of norms and practices that structure the way in which individuals and households gain access to economic advancement. Whether in Mexico or Albania, South Asia or West Africa, migration of different kinds has become increasingly common as a potential strategy to diversify rural household income, and for individuals to seek wealth and economic advancement. In this context, this paper has argued that across different geographical, economic and social environments, the key questions about how the institution of migration affects inequality relate to access -who gets to migrate where -and the different opportunities that different types of migration stream offer.

EQUITY & DEVELOPMENT World Development Report 2006 Background Papers
Bearing this in mind, the most obvious -although perhaps not completely politically palatablepolicy response is to suggest that the key to ensuring that migration has a positive impact on inequality is to ensure the widest possible access to migration opportunities for poor people. A number of the studies reviewed here show how an inequality-reinforcing effect of migration is enhanced where the poor are effectively restricted to shorter distance, less remunerative or less safe migration streams. In Albania, a major reason why the poor are restricted to less beneficial migration within the country or to neighboring Greece is because of strong restrictions on movement to elsewhere in Europe. Those who attempt to migrate anyway are forced into undocumented positions where they often experience exploitative conditions and find it difficult or impossible to contribute to households back home. 73 Siddiqui has pointed out how restrictions on international unskilled female migration from Bangladesh effectively drives the migration of poorer women into illegality, making them significantly more vulnerable to exploitation and reducing the potential for their migration to help reduce either gender or income inequality. 74 Although not reviewed here, studies in China have also suggested that the imposition of restrictions on migration operate to enhance both spatial and inter-household inequality, whereas migration acts to reduce such inequality. 75 The Chinese example is useful in this context, since it represents an attempt to constrain internal migration, something that is arguably more likely to be amenable to policy change. In turn, whilst relatively few countries internationally explicitly seek to control internal migration in the way that China does, and China itself has an increasingly ambivalent attitude towards migration, negative attitudes to internal migration -especially rural-urban migration -remain commonplace amongst governments and many development agencies alike. In turn, these attitudes often translate into petty restrictions, harassment and denial of rights to internal migrants in ways that contribute to negative effects on the poor -and hence on inequality. For example, rural-rural migrants in India who cross state boundaries are generally left without economic, social and indeed political rights, significantly impoverishing a group of migrants that are draw substantially from poorer social classes and castes.

Migration volumes and inequality
If it is true that migration as an institution helps to improve equity provided that the poor have access to it, then the implication is that significant expansion of migration flows is likely to improve both equity and growth. Some support for the view that increased migration would stimulate growth is provided by Philip Martin, in his study of migration as part of the so-called 'Copenhagen Consensus'. 76 Martin cites work in the 1980s by Hamilton and Whalley which suggested that world GDP could more than double if free migration were allowed, whilst making his own estimate that the movement of an additional 10 million migrants would raise world GDP by $260 billion, or one per cent -about five times annual ODA. However, he also notes that whilst the economic impacts for receiving countries are 'clearly positive', for sending countries, the benefits are less clear-cut. In part, this is because the gains accrue primarily to individual migrants, so that the consequences for global equity depend on what he calls the 'Three R's': recruitment, remittances and return.
Martin suggests three policy options for increasing and spreading the benefits of increased migration -active selection of migrant workers most likely to succeed; widening entry channels for unskilled guest workers; and promoting the flow of remittances and returnees to redistribute gains back to home countries. Yet others within the Copenhagen Consensus were less EQUITY & DEVELOPMENT World Development Report 2006 Background Papers optimistic that gains could be spread. One commentator questioned the feasibility of increasing temporary mobility 77 , whilst another questioned the magnitude of the gains calculated by Martin. 78 Indeed, where there is most consensus on the potential for gains -through the recruitment of skilled workers, and return of emigrants to their home countries -there is arguably the greatest potential for increased inequality within sending societies.
In contrast, a rather different conclusion emerges from recent econometric analysis of the potential impact of liberalisation of temporary mobility under GATS Mode IV by Winters et al. 79 , which suggests more scope for a beneficial impact of increased migration on both growth and equity. Their model estimates that increased mobility equivalent to three per cent of developed countries' skilled and unskilled work forces would generate an estimated increase in world welfare of over US $150 billion. Moreover, its distribution -which is fairly equal between developing and developed countries -suggests it would be at least benign for global inequality. However, this analysis is based on data aggregated at the national level, and so tells us nothing about the distribution of gains within migrant-sending countries.

A world migration institution?
In addition to seeing migration itself as an institution or process that can enhance growth and equity in certain conditions, it is also useful to consider how other institutions influence the extent to which migration might have such an effect. For example, at a macro-level, Jagdish Bhagwati has argued for over a decade for a World Migration Organisation (WMO) to oversee and monitor migration in the way that the WTO does for world trade, and to codify and spread best practice. 80 The recent establishment of a Global Commission for International Migration goes some way towards this end, in its aim of analysing current approaches to migration management by different states and in different regions, and its recommendations could well include the establishment of a WMO. Böhning argues that such an organisation could be the basis on which a consensus could be built in favour of 'desirable migration'. His slightly different formulation to Bhagwati envisages a WMO playing a role to combat unlawful crossborder flows ('undesirable' migration), to operate a technical cooperation facility to help countries build capacity to manage migration, and involving a principle of the sharing of income tax of migrant workers between sending and receiving countries.
However, whether a WMO would make migration more equitable is a moot point. First, it is not unreasonable to suggest that the political will needed even to establish a WMO is currently conspicuously absent, and that it is this hostility towards migrants and migration on the part of states that represents the real problem, rather than the lack of a specific organisation to tackle migration issues. 81  One alternative way of moving forward institutionally is to think of migration management within the context of other forms of international negotiation, so that sending nations seeking concessions -whether increased access to labour markets for their workers, tax sharing arrangements, or any other kind of redistribution of benefits -can potentially gain these by making concessions in other areas. This is the basis on which there has been some interest in GATS Mode IV as an area of policy dialogue, since this places the issue of more equitable migration management squarely in the arena of trade talks, where developed nations are themselves seeking liberalisation. Some countries have also sought to develop bilateral agreements on labour migration, although systematic analysis of their outcomes remains to be conducted. 82

Conclusion
This review does not pretend to be a comprehensive analysis of the relationship between migration and inequality, but instead seeks to draw lessons from particular case studies that are relevant to policy debate. It demonstrates how the mutual causality between migration and inequality varies across space both between and within these regions. It illustrates the necessity of defining which kind of migration, and which kind of inequality are being analysed, and highlights the issue of access to a hierarchy of different migration possibilities and opportunities.
There is an inevitable unevenness in the treatment of the different regional cases, partly as a result of the contrasting methodologies -both quantitative and ethnographic -employed in the studies we review. In some regions, it is much more possible to follow through on socioeconomic inequalities, but in others, we have chosen to focus on evidence on economic inequality. However, this relates to a general observation that can be drawn from across the case studies -that the extent of knowledge on the effects of migration on inequality -and vice versa -remains limited.
This is not necessarily for lack of data. Significant and potentially useful large datasets exist in the form of LSMS surveys in several countries, with the possibility of future inclusion of 'migration modules' in future surveys where this is not covered in depth in existing surveys. Large-scale sample surveys of migrant households have been noted here for Kerala and Mexico, and exist for some other countries too. There is also a wealth of information in ethnographic accounts of migration from particular areas, of which this review only scratches the surface. But the specific issue of how inequalities of various kinds are created, reproduced or eroded by different forms of migration remains hidden within broader studies that have tended to focus on macro-economic outcomes, or outcomes for individual migrants and their families, rather than the distribution of these outcomes. There is also a critical lack of time series data in most cases that would allow us to consider how inequality changes over time.
Nonetheless, a key theme of this review has been the role of institutions in mediating the impact of migration on inequality. At one level, migration itself can be seen as an institution, which has potential to reduce inequality where the poor have access to it. In turn, given wide variation in outcomes of migration for equity, as well as the fears of states in opening up EQUITY & DEVELOPMENT World Development Report 2006 Background Papers migration opportunities, there have also been calls for the establishment of a global institution to regulate migration flows. Yet a key element across most, if not all of the case studies reviewed here, is the important role played by meso-level institutions -notably social networks -in both facilitating migration in the first place, and influencing how the income earned by migrants is, or is not channeled back to help reduce poverty and inequality. In this sense, the institution of migration can be seen as embedded in numerous other institutions, which themselves become key to understand.
The lessons for public policy in relation to this observation are not always completely obvious. For example, in Albania, it is clear that access to social capital or social networks is critically important for successful migration. The ability of marginalised groups such as the Roma to move even to Greece is a function of either having a family member who has already migrated, or their ability to join a group of migrants who cross the border collectively, thus sharing the risks and costs of a difficult (because illegal) journey. Yet it is difficult to see how public policy can act to strengthen family networks, particularly if the objective is to increase migration opportunities. Nonetheless, development finance is -at least in principle -partly focused in Albania on strengthening social capital 83 , and if effective, this might be expected to bring benefits not only in general development terms, but also specifically in influencing the impact of migration on inequality.
Meanwhile, the South Asian case studies presented above show how the role of social networks in influencing the migration process can go well beyond family and kin ties, providing a rather wider arena for public policy to make a difference. In India, the existence of stable networks that link migrant workers to labour recruiters is seen as crucial in influencing outcomes for migrant workers, suggesting that there is scope for regulation of the recruitment process to ensure that contracts are fair, accommodation for migrant workers is affordable, and abuses are avoided. There is also potential for supporting initiatives by NGOs and local associations that seek to reduce the risks and costs faced by seasonal migrant workers, such as initiatives to support health care for migrant workers, whether in destination areas or on return, and attempts to make bus routes used by migrant workers safer. 84 In West Africa, there are also a wide range of social institutions, including hometown, religious, ethnic, village and alumni associations that not only help to facilitate migration in the first place, but can play a crucial role in channeling remittances not only to individuals and families, but also to community-level investments and initiatives. Perhaps the best-documented case involves migration from Senegal to Western Europe, where mouride religious associations have ensured that migrant remittances are invested in a wide range of community development projects. 85 However, these associations -and indeed more formal NGOs -may also play a crucial role in limiting the exploitation of, or harm to poorer migrants. In this context, it is instructive to note that attention and funding appears to have shifted recently away from NGOs that support poor and vulnerable migrants, and towards 'anti-trafficking' measures that may increase rather than decrease the vulnerability of poor migrants. 86 Another important set of institutions that influence the inequality-reducing or reinforcing effect of migrant remittances is in the financial sector. Migrants use a wide range of mechanisms to send money back to families and home communities, ranging from formal banks to money transfer agencies such as Western Union and MoneyGram, more informal hawala or hundi systems of transfer, and physically carrying or sending home cash during return visits. Of

EQUITY & DEVELOPMENT World Development Report 2006
Background Papers these transfer mechanisms, the more formal kinds tend to be heavily stacked against poorer migrants who wish to send smaller amounts of money, as exchange rates may be poor, and commission rates crippling. Here, there is potential for policy to promote competition as well as regulating the activities of money transfer agencies in order to ensure transparency, trust and competitiveness of the services offered.
More generally, macro-economic stabilization in countries of origin may also be critical in enabling investment of remittances in anything other than personal consumption. Whilst personal consumption may itself involve investment in social terms, and may contribute to reducing inequality, substantial and sustainable investment in economic and social activities at a community level is clearly a higher goal for most governments interested in reducing inequality. Migrants cannot, and should not, be expected to make such a contribution on their own, but the creation of economic conditions in which they perceive it as in their interests to do so is clearly a worthwhile objective.