UNCTAD The Least Developed Countries Report 2007 Background Paper

Transcription

1 UNCTAD The Least Developed Countries Report 2007 Background Paper Brain Drain and Brain Gain: A Survey of Issues, Outcomes and Policies in the Least Developed Countries (LDCs) Chris Manning The Australian National University Background Paper No. 6 This paper was prepared as a background paper for UNCTAD's Least Developed Countries Report 2007: Knowledge, Technological Learning and Innovation for Development. The views in this paper are those of the author and not necessarily those of UNCTAD. The designations and terminology employed are also those of the author.

2 Table of Contents A. Introduction 3 B. Brain drain and brain gain: International experience.3 1. The incidence and rates of brain drain Debates on the impact of emigration..6 (i) The optimists..7 (ii) The realists...8 (iii) Remittances Implications for the LDCs 10 C. Estimates of brain drain and remittances in LDCs: OECD countries More recent developments in LDCs Brain drain and remittances in LDCs...16 D. South-South migration.17 E. Regional trends and country case studies Africa Asia The South Pacific Islands.20 F. The case of health.21 G. Receiving and sending country policies Countries of destination 24 (i) The United Kingdom and Europe...24 (ii) The United States.26 (iii) International programs.26 (iv) Sending country policies..28 References 31 Annex

3 A. Introduction This paper examines brain drain and brain gain through international migration of skilled and professional workers from the least developed countries (LDCs). It deals mainly, although not exclusively, with permanent emigration to developed OECD countries. The paper is set in the more general context of international experience with brain drain, and debates over costs and benefits of outmigration, from developing countries (Section II). In the third section, we present data on migration of skilled workers for individual LDCs and regions. This provides an overview of the situation, although detailed country-level data for LDCs, as a whole, are now somewhat dated (the most comprehensive information is available only for the year 2000). The analysis is based on lifetime migration data for OECD countries and uses tertiary educated residents as a proxy for skilled and professional migrants. We also provide an estimate of more recent trends for all LDCs. Next the discussion moves to major patterns of South-South migration among the skilled in Section IV. Although these latter movements are of considerable significance, quantitative data is fragmentary at best. Subsequently, in section V, the discussion focuses on regions and case studies of brain drain and brain gain. The exposition is organised around three major groups of LDCs: Africa (with a focus on sub-saharan Africa, and within sub-sahara, East and West Africa), which account for the large majority of LDCs; second, the more densely populated LDCs in Asia; and, third, the smaller, island states of the South Pacific. We give examples of case studies of skilled migration experience in each of these groups. Section VI takes a sectoral approach, giving special attention to the case of health care professionals (doctors and nurses), who has been a focus of much of the literature on brain drain. Finally, in section VII, the paper turns to policies towards skilled migration in both receiving and sending countries. Besides looking separately at the evolution of policies in the UK and the rest of Europe, and USA, we also track international initiatives and deal very briefly with regional arrangements of importance to some LDCs. The section closes with a short examination of LDC policies, especially with regard to professionals in health care. B. Brain drain and brain gain: International experience Brain drain has long been an important issue in the development literature as underutilised skilled and professional manpower from newly independent countries Africa and Asia sought jobs in Europe, North America and Oceania from the 1950s (Lucas, 2004). Slow economic growth and political instability, especially in parts of Africa, saw an increase in cross border movements of professionals, during the 1970s and 1980s, both to developed countries and more rapidly growing neighbouring states (Russel, Jacobsen and Stanley, 1990). While supply pressures from developing countries, including many LDCs, have remained strong in the past two decades, demand pressures for increased deployment of migrants from developing countries have increased in industrialised countries, despite their rapidly rising numbers of tertiary graduates. This has in turn contributed to increased skilled migration to developed countries. Such movements have been underpinned by large and increasing absolute wage differentials for skilled manpower between developed and developing 3

4 countries. More open policies towards skilled migration in developed countries have also been important, especially in North America, Australia and New Zealand. 1 In this section we document some of these trends. It examines the impacts of emigration and temporary migration of skilled and professional workers on countries of origin and draw some conclusions for LDCs. Most the data and discussion are based on literature dealing with permanent emigration of skilled migrants to developed countries (mainly North America, Europe and Oceania, specifically Australia and New Zealand). In particular, two important groups of skilled migrants are not covered in most of the literature. The first are the increasing proportion of skilled workers migrating on temporary contract to developed and other developing countries. The numbers are large, and are not incomparable with permanent movements. Several of the issues for temporary migrants are somewhat different from those with respect to permanent out-migration, which was the dominant mode of recruitment in most developed countries through to the 1980s and 1990s. Return migration is more predictable for many contract workers, although contracts are renewed in many cases, and highly valued contract workers may well become permanent. Brain gain is more immediate, but probably less substantial for those who do move, and remittances are probably larger. Second, the discussion ignores, for the most part, South-South migration, which is estimated to account for around half of all migration flows (Ratha and Shaw, 2006). South-North flows are relatively more important for skilled and professional manpower than movements of the unskilled. But nevertheless, South-South movements of skilled manpower are very important in certain parts of the world, for example Southern Africa and Southeast Asia. We will come back to some of these issues in the discussion of specific LDC experience in later sections. 1. The incidence and rates of brain drain The latest estimates of skilled out-migration rates are for the year 2000 when population censuses were undertaken in most OECD countries. The data suggest that skilled out-migration (proxied by movements of tertiary educated people) from developing countries, increased sharply in the 1990s (Docquier and Marfouk, 2006). 2 While the total OECD population increased by less than 20 per cent in the 1990s, migration rose by some 40 per cent overall (from 42 to 59 million) and skilled migration rose by some two-thirds (12 to 20 million). The patterns are documented by Docquier and Marfouk (2006, Table 5.3, pages ), based on OECD data. 3 Table 1 summarises several of the main findings of this study: 1 See especially Freeman (2006) and Pritchett (2006). 2 Recent studies showing similar patterns include Adams (2003). 3 See Table 5.3, pages Migration data analysed by Docquier and Marfook are only for OECD countries (based on the birth place of resident populations from national censuses) and hence they understate the extent of intra-regional migration across national borders within Africa, South and East Asia and the South Pacific. A significant proportion of skilled migrants to South Africa and Singapore, for example, were born in other Southern African and Asian countries, respectively. However, Docquier and Marfook point out that the bias is less important for skilled than for unskilled migrants, 4

5 Skilled out-migration rates were inversely related to country size; they were: o seven times greater in countries with <2.5 million people compared with those with 25 million and above o higher in middle than in low or high income countries o very high among UN-defined small island developing countries By geographical region, skilled migration rates in 2000 were: o were highest in Africa among regional groups (compared with Asia, America, Europe, and Oceania), even though Africa only accounted for eight per cent of all out-migrants, see last column in Table 1). 4 o among sub-regional groups, they were approximately more than ten times higher in Polynesia, 4 6 times higher in the Caribbean, Melanesia and Micronesia, and 2 3 times higher in Eastern, Western and Central Africa, and Central America than in all lower middle and low income countries. However even though the small Island states recorded very high rates of out-migration, they accounted for a tiny proportion (less than one per cent) of all skilled emigrants. o even though rates were relatively high in most of Africa and the South Pacific, a much higher proportion of all educated migrants living in OECD countries in 2000 came from East, Southeast and South Asia (mainly China, India and the Philippines) Rates of skilled out-migration were highest (13 per cent) in LDCs, compared with all other country income groups. Nevertheless, LDCs only accounted for less than five per cent of all skilled migrants, compared with close to 30 per cent each among lower middle income and higher income countries. 5 As might be expected the stock of skilled manpower was positively related to level of economic development. However, the share of skilled migrants was negatively correlated with the level of development, except for the high income countries (that is, skilled migration relative to the stock of educated manpower at home was highest in low-income countries and especially in the LDCs, but lower in middle income countries: see column 4 in Table 1) These data on skilled (tertiary educated) migration flows provide no breakdown by industry/occupation and level of schooling. Thus outmigration is very much higher in certain professions that are skill intensive and where skills are relatively uniform internationally, such as medicine. The greatest concerns in many countries are related to the out-migration of health professionals. Moreover, migration of highly educated persons with more than basic tertiary training tends to be much greater than for the since a high proportion of international skilled migration flows (and estimated 90 per cent) were to OECD countries in the 1990s. 4 In terms of selectivity, one study suggests that over 75 per cent of all African out-migrants to the OECD were tertiary educated, compared with just over, and just under, 50 per cent in the Asia Pacific and Latin America, respectively, around 1990 (Lowell and Findlay, 2002) 5 See section III below for further detail on LDCs. 5

6 tertiary educated population as a whole. For example Lowell et al. (2004: 9) cite studies which suggest that as many as per cent of the developing world s population trained in science and technology live in the developed world. Thus, most of the skilled migrants have not come from least developed countries but rather from large, lower middle income and middle income countries China, India, Mexico, Brazil and the Philippines. A large repository of skilled manpower related mainly to absolute population size, a well established diaspora in host countries and policies targeting greater migration of skilled manpower have all played an important role in these movements. But the incidence of out-migration was high in several LDCs Ethiopia, Sudan, United Republic of Tanzania (Tanzania), Bangladesh, Nepal and Myanmar (Burma), as well as some of the smaller LDC, island states in the South Pacific such as Samoa. It is understandable that much of the discussion of brain drain has been focused on the larger countries of out-migration such as India. However, since per capita incomes are only marginally higher than in many of the LDCs, there are important lessons to be learned for the LDCs from the experience of brain drain and brain gain in the larger countries. On the demand side, opportunities for work among skilled immigrants in developed countries accelerated from the 1990s and increasingly into the 21 st century. More open policies were related to increasing shortages of skilled manpower, as a result of demographic and structural change. While skill shortages have been experienced across the board in many increasingly technologically advanced developed countries, three sets of factors have been especially important in influencing renewed demand for skilled manpower. They have stimulated renewed attention to the issue of brain drain, as well as potential brain gain benefits to poorer origin countries from deployment of former migrants (or their knowledge) back home as well as to potential worldwide benefits from international migration (World Bank, 2006: Chapter 2). 6 First, ageing of developed country populations, especially in Europe and later Japan, has contributed rising demand for skilled manpower in non-tradable service industries, particularly in health and aged care. 7 Second, the information technology revolution has greatly increased the demand for skilled manpower in the production of computer software and the demand for computer and ICT engineers. Third, shortages of lower to middle-level skilled manpower technicians, electricians, plumbers, nurses and teachers has been especially marked, as developed country workers shun difficult and dirty blue collar and related jobs, and the output of their educational institutions have failed to keep pace with demand. 2. Debates on the impact of emigration 6 Modelling exercises have paid most attention to the potential benefits from liberalising the movement of unskilled workers, although they also note modest gains to developed and developing countries from increased mobility of skilled manpower. See, for example Winters, et al. (2003). 7 Much of the increased demand has also been for unskilled workers in non-tradable services, the most rapidly growing sector of employment in the United States and other developed economies (Pritchett, 2006). 6

7 The literature on brain drain and brain gain is ambivalent as to whether sending countries gain from migration of skilled manpower. A range of factors have been identified as important: rate of economic growth and utilisation of skilled manpower back home, especially in certain skilled occupations (particularly relevant to the LDCs); the size of the brain drain relative to the domestic supply of skilled manpower; the role of remittances; and the extent to which migration stimulates development of human capital in countries of origin (partly determined by the scale of out-migration and the role of the diasporas). Early theoretical studies tended to emphasize the negative effects of brain drain, which supported much of the public debate focusing on the negative effects of out-migration of skilled manpower. 8 The studies focused on the short-run impact of a loss of human capital, the cost of which is mostly born by domestic taxpayers, and the impact of the fall in the supply of educated manpower on national output. Subsequent research regarding the impact of out-migration of skilled manpower on countries of origin can be divided into two groups, for convenience: the findings of the migration optimists and the migration realists. (i) The optimists Early research gave way to more complex and optimistic models that stressed the dynamic effects of migration. It highlights the positive impacts of remittances on consumption and investment at home, and the impacts on human capital development in sending countries. The scope broadened to include technology and knowledge transfer and other benefits of brain circulation, and the potential benefits from diaspora links. Theoretical models developed by migration optimists such as Stark (2004) 9 examined the impact of skilled migration in an open economy framework, and posited the potential for brain gain and a higher level of human capital per worker as a result of increased demand for and access to education among those left behind. Mountford (1997) argued that such general improvements in education could contribute to greater equality and the possibility of a permanent increase in average productivity of workers. Docquier and Rapoport (2004: 27) summarise the main effects in response to the successful experience of migrants abroad: successive cohorts adapt their eduction decisions, and the economy-wide average level of education partly...or totally catches up, with a possible net gain in the long run.. and...the creation of migrants networks that facilitate the movement of goods, factors and ideas between migrants host and home countries. 10 The diaspora reduces the costs of migration and risks in countries of destination, providing greater incentive and demand for migration linked education at home (Kanbur and Rapoport, 2004 cited in Docquier and Rapoport, 2004: 28). 8 See Grubel and Scott (1966) and Bhagwati and Hamada (1974). While Grubel and Scott argued the loss to developing countries was likely to be relatively small (equal to the direct loss of output from withdrawal of skilled manpower), Bhagwati and Hamada developed a theoretical model predicting larger losses as a result of to upward pressure on skilled and unskilled urban wages, and increased unemployment, from migration, owing to institutionally set minimum wages in developing countries. 9 In this paper, Stark draws on earlier research published in 1997 and Thus trade and capital, and labour flows, are no longer viewed as substitutes but complements as a result of such networks. See Rauch and Trindale (2002) on the links between migration and trade networks and Javorcik, et al. (2006) who find evidence for a positive relationship between migration and foreign investment (FDI) flows from the United States to countries of origin. 7

8 At the same time, dynamic effects associated with brain circulation have received increasing attention. More attention in the empirical literature was given to the role of return migrants in raising skill levels, and promoting technology transfer and capital accumulation, especially in the successful growth cases of Taiwan and South Korea by the late 1990s, and in India following economic reforms in the early 1990s (Saxenian, 2002). In the case of both Taiwan and South Korea a range of factors contributed to a rising number of returnees from the mid to late 1980s and into the 1990s ( Yoon, 1992; Luo and Wang, 2002; the World Bank, 2006). Government policies targeted the return of skilled manpower that had stayed abroad after studying in developed countries, especially the United States. The emphasis on industrial upgrading and technological change was accompanied by financial and other inducements for engineers and managers to return to take up high level jobs in government and research institutes. 11 At the same time, as the economies expanded, a more sophisticated industrial base meant a greater demand for highly skilled manpower from the private sector, in a range of industries, such as the emerging computer, electronics and automobile industries. Complementary efforts involved government efforts to develop links between diaspora groups overseas, and business and government in source countries to help compensate for the loss of skilled manpower and promote investment in skills and productive activities back home (Saxenian, 2002). As in India, overseas professionals contributed to economic development through exchange of ideas, shortterm and temporary visits to countries of origin, as well as permanent relocation. (ii) The realists The above-mentioned relationships are complicated, however, especially since theoretical models fail to take account of a number of factors: migration realists have focused on differences in the quality of out-migrants, and return migrants, compared with their (potential) replacements back home and on the extent to which skilled migrants are employed in skilled occupations abroad (see especially Docquier and Rapaport, 2004 for a survey). 12 Several of these factors have been identified as reducing potential gains from brain circulation and remittances from skilled and professional manpower in many LDCs. Many studies have focused on the migration premium a range of 2 10 time higher earnings among migrants compared with non-migrants in the same occupations, according to Docquier and Rapoport (2004: 16) while paying less attention to the costs of migration, both psychic and social, as newcomers seek to assimilate in new environments. One important finding on the jobs undertaken by educated migrants suggests that many work in less skilled jobs, and thus experience 11 The encouragement of emigrants to take positions in technology parks, such as the well known Hsinchu Science-Based Industrial Park in Taiwan, is one example. 12 For a summary of some of the debates and empirical analysis in recent literature, see especially Lucas s major survey of the literature and empirical evidence on international migration and remittances (2004), the International Organisation of Migration Global Migration report (2005), the World Bank Global Economic Prospects (2006, Chapters 3 5), and papers included in Ozden and Schiff (2006). 8

9 brain waste. Such patterns have been noted among migrants from countries where educational standards (the quality of education) are low relative to those more developed countries. It also seems to have been more prevalent among emigrants from non-english speaking backgrounds and those from countries that experience political or other conflicts (Ozden, 2006). Doctors working as nurses are one example, especially in the case of the Philippines. In such cases, migration of educated persons is not necessarily a stimulus for education in countries of origin, or may be a stimulus for learning skills which do not replace those that are lost (for example, doctors retraining to become nurses in the Philippines). These impacts may be inter-generational. Mackenzie (2006), for example, found that the children of tertiary educated migrants who had moved to the United States from Mexico, had lower enrolment rates at ages compared with children of non-migrant compatriots. He suggests that potential returns to schooling were lower for children of emigrants, compared with children in non-migrant households. Mackenzie argues that those in non-migrant households were less responsive to the potential gains from migration, and hoped that a tertiary education would provide an opening for better jobs at home in Mexico. Impacts on human capital in places of origin are likely to be varied and larger (understandably) in low human capital and low migration contexts, either through return migration or remittances, than where there is already an abundant supply of educated manpower, and substantial out-migration (Docquier and Rapoport, 2004). 13 Nevertheless, while short run brain drain effects are likely to be greater in countries with a narrow human capital base, 14 brain drain has been found to be negatively related to economic growth in countries where migration rates for the highly educated are high (above 20 per cent) and where the proportion of more educated persons in the total population is also relatively high (Beine, Docquier and Rapoport, 2003). 15 One possible explanation is that rates of return migration are low among skilled migrants, in general, and a large diaspora discourages return migration. This begins to have a drag on growth in countries of high out-migration. Heterogeneity among migrants and non-migrants is also an important issue. Schiff (2006) has drawn attention to the fact that the more optimistic models of migration tend to ignore self-selection which results in higher quality (both more clever and more experienced) manpower going abroad. For these migrants there are not near-perfect substitutes among the remaining stock of skilled or potential manpower. It has also been noted that the less successful skilled migrants tend to return home, and hence the brain gain is smaller than some of the theoretical models predict Nevertheless some studies suggest even in country cases of high migration and abundant human capital in source countries, the effects of remittances on skill formation may be large, such as in the case of Indian doctors who remit substantial shares of their earnings in the United Kingdom back home (Kangasniemi, et al., 2004 [cited in Docquier and Rapoport, 2004: 26) 14 Although Docquier and Rapoport note that brain drain has only had a significant short run effect on per capita income in extreme cases such as Guyana and Jamaica. 15 Beine, et al. (2004: 23 24) did find however that the migration rate among educated persons exerted a strong positive influence on human capital formation in a sample of 50 countries. 16 Borjas (cited in Beine, et al., 2004) showed, for example, that less successful foreign scientists in the United States were more likely to return home than those who fared better in their professions abroad. 9

10 (iii) Remittances Finally to remittances. From the period of the oil boom in the 1970s, when large numbers of temporary migrants left their home countries for the Middle East, there was increasing attention to the impact of remittances on incomes of those left behind (Amjad, 1989). In more recent times, several studies have pointed to the dramatic increase in remittances abroad totalling an estimated $167 billion in 2005 according to official estimates. Their major role has been highlighted both at micro and macro levels in stimulating consumption and investment, and helping relax foreign exchange restraints. 17 The absolute flows of remittances back home have been dominated by nationals of the large emigration and temporary out-migration countries Mexico, the Philippines, India and Brazil. They have nevertheless been significant in many LDC economies. For example, in a range of LDCs (Cape Verde, Haiti, Nepal and in the micro states of Kiribati and Samoa), remittances are greater than total exports, and are a major source of foreign exchange in many others (the World Bank, 2006: 88). 18 However, even though incomes earned abroad are higher among the skilled than the unskilled, we have already mentioned the possibility of lower levels of remittances from more permanent skilled migrants. This group are more likely to move with families, or are either more likely to inter-marry with residents, or to be joined later by family members). They also tend to migrate longer distances than unskilled migrants, and hence have less intensive contacts with their country of origin (Faini, 2006). Faini (p. 3) notes that although The direct effect on skills might be positive, the overall effect, that controls for the longer propensity to stay abroad of skilled migrants, may well be negative. The less skilled are more likely to migrate to nearby locations for shorter periods, and if they are employed on temporary contracts are more likely to send remittances back home on a regular basis. Evidence from the Philippines, Mexico and Egypt all provide some support for this proposition. Faini s own analysis of data on skilled and unskilled overseas migrants in OECD countries finds that a higher proportion of skilled migrants (and an increasing proportion over time, tends to be negatively related to the size of remittances (pp. 8 12, Table 2 and 3). 19 Remittances may well have a positive impact on consumption and investment back home, and contribute to poverty alleviation (Adams, 2007). But these effects appear to have been stronger in cases where unskilled migration predominates, than for migration of more highly educated workers. 3. Implications for the LDCs 17 These figures are based on official IMF data, indicating remittances grew faster than foreign direct investment and official development assistance (ODA) over the past decade, doubling in several countries and rising by close to 10 per cent per annum between 2001 and 2005 (the World Bank, 2006: 87 89). Remittances accounted for around seven per cent of domestic investment and imports in all developing countries in A high proportion of the flows were between developing countries (40-50 per cent) and it is estimated that an additional 50 per cent of all remittances flow through informal channels. 18 See Section III below for further details on LDCs. 19 See also Niimi and Ozden (2006). 10

11 Many of the findings of this literature survey need to interpreted with care since they pertain mainly to larger, lower income and lower middle income countries which account for a large share of international migration of skilled manpower. Nevertheless some important lessons can be learned. First, following Docquier and Rapoport (2004: 34), while the optimal rate of skilled and professional out-migration...is likely to be positive, whether the current rate is greater or lower than this optimum is an empirical question that must be addressed country by country. In short, there appears to be huge variation in individual country experience with respect to brain drain, brain circulation and brain gain. One important factor is the size of the brain drain, which has both positive and negative effects: a large diaspora provides a cushion and a support for would be skilled migrants, but (as noted) at the same time may reduce the potential benefits to countries of origin over time. More settled migrants tend to have more tenuous links with home countries, especially net of the impact of remittances. Second, industries which employ emigrants also play a part in determining the benefits. Benefits to the home country from out-migration of doctors and nurses in a largely non-tradable, and heavily regulated industry (despite the internationalisation of health care service provision in some countries) might be expected to have few benefits in terms of technology transfer, investment from abroad and, of course, trade. They can be expected to be much more positive in a highly open, tradable industry such as ICT, where economic benefits from nationals working for private investors abroad can be substantial for technology, employment and investment in countries of origin. It is perhaps thus no surprise that Indian policy makers tend to be much more enthusiastic about brain drain into ICT industries than Filipinos are about loss of doctors and nurses to hospitals in countries across the globe. Finally, home country policies and growth prospects can play a major role in increasing brain gain, and reducing the costs of brain drain. Rapidly growing, middle income, countries, that have passed the migration hump, are likely to be in a better position to utilise skilled manpower from abroad, and to invest in the human capital that is necessary to plug the gaps created by emigrants (Lucas, 2004). 20 But even at lower levels of per capita income, domestic policies appear to be important. For example, while the Philippines shows no sign of being over the migration hump (and certain sectors appear to suffered negative effects from brain drain), the government has been proactive in seeking to reduce the costs, encourage return migration and increase the social and economic benefits from emigration. As we shall see, LDCs have much to learn from the experience of some of the middle income countries with regard to increasing the net gains from emigration. 21 C. Estimates of brain drain and remittances in LDCs: OECD countries We now turn to some quantitative estimates of skilled out-migration for the LDCs. The latest data on the total number of skilled out-migrants are from the round of censuses conducted in 1990 and 2000 in OECD countries, which are host to a high 20 The migration hump refers to a process whereby the rate of (net) out-migration increases in the early stages of economic development until it reaches a peak, somewhere in the middle income range of national GDP per capita, and then begins to decline. 21 See Section V below for a discussion of some of these policies. 11

12 proportion of all skilled migrants. This is the basis for our discussion of brain drain in terms of absolute numbers, and shares in the sending populations. Hence the data presented in this section are only a rough proxy of total skilled migration stocks and flows, and of current conditions with respect to LDCs. One major, potential bias in the data is the under-enumeration of illegal or undocumented/unregistered migrants in censuses in receiving countries, especially among individuals who have overstayed their visas. This problem is typically much greater for unskilled than for skilled migrants. Nevertheless, overstays among skilled migrants emerged as a significant problem in the United States in particular from the late 1990s, after policies towards skilled migrants were liberalised (Martin, et al., 2006). Our guestimate is that perhaps as many as five per cent of all skilled migrants may be missed from the OECD censuses. However, we don t have any basis for estimating the number or share of illegals by country of origin. More recent data suggest that total migration has probably accelerated in the period (see below for some rough estimates), although there is no breakdown by country of origin and for skilled (more educated) workers separately. 22 Despite these data limitations, later we provide an estimate of the share of skilled migrants to total population for LDCs in 2004, based on past trends in skilled migration to major developed (OECD) countries. Case studies provide some additional information on more recent trends. The data set used for much of the quantitative analysis in this paper is the only comprehensive one available internationally, as an indicator of skilled and professional out-migration from the LDCs. While it is a little out of date, it is nevertheless a remarkable collection of information: the first quantitative estimate ever made of the incidence of brain drain from all developing countries and LDCs, based on the census and related labour force data of some 30 OECD countries. Skilled emigrants are proxied by the share of tertiary educated overseas population in all OECD countries as a share of the stock of tertiary educated in source countries in the years 1990 and Table 2, and Annex Table 1, provide information on the main rates for all emigrants and tertiary educated emigrants, as well as rates of change for the period for all LDCs for which there are data. The main data source is Docquier and Marfouk (2004), which was updated with relatively small changes in Docquier and Marfouk (2006). 24 To help interpretation the data are organised by region: the main regions of Africa, where the bulk of LDCs are located; the Island economies (in the Pacific and 22 More up-to-date analysis awaits publication and analysis of census data for 2005 in receiving countries, which provide information of birthplace and previous residence of the total population. 23 In the discussion of these data, we use the terms skilled, skilled and professional and tertiary educated interchangeably, as the emigration rates of tertiary educated are used as a proxy for skilled out-migration. As noted in Section II, rates for tertiary educated persons are only a rough proxy for skilled out-migration, bearing in mind that numbers of tertiary educated persons are likely to be employed in less skilled occupations. From the perspective of brain drain from sending countries, however, the bias is not too significant. But it obviously impacts on rates of skilled in-migration into OECD countries, and policy implications drawn from these movements. 24 Docquier and Marfouk (2006: ). 12

13 elsewhere mainly in Africa); and Asia. 25 Within regions, countries are ranked by total population size in Annex Table 1, which is correlated with the absolute number of emigrants, although not necessarily with migration rates. Annex Table 1 also includes data on per capita income and the human capital index for all countries. What are the main patterns of skilled emigration and changes in emigration rates in the period among the LDCs. Three stand out. Emigration rates were high; they demonstrated some important regional effects even though variations were quite large within the main geographical regions; and rates increased in the 1990s, again with some significant differences among regions and countries. First, as noted in Section II, emigration rates were generally high among tertiary educated persons by international standards (unweighted mean by country of 21 per cent in 2000). There was considerable variation in the (unweighted) total rates of emigration among tertiary educated persons within and by country group among the LDCs. They were close to 25 per cent in the Pacific and other island states, West Africa and East Africa, and lowest in the generally more populated countries of Asia (5 per cent), with Central Africa falling in between (15 per cent). The overall unweighted mean was much higher than for all lower middle and low income countries (7.6 and 6.1 per cent respectively in Table 1 above), although the latter figure (weighted) is heavily influenced by quite low out-migration rates in the giant developing countries of China and India in particular. As mentioned in Section II, these rates compare with overall out-migration rates among tertiary educated persons that were very high among all Caribbean countries (39 per cent) (see Docquier and Marfouk, 2004: Table 2). This was followed by West and East Africa (27 and 18 per cent respectively most of which consist of LDCs). They were moderately high among several of the better endowed, developed countries in Europe (in particular Northern Europe), Central America and Southeast Asia (close to 10 per cent), but moderately low (around 5 per cent) in Western and South-Central Asia, Latin America and all of Oceania (the latter dominated by Australia), and very low in the North America (the United States less than one per cent, much less populated Canada recorded a rate of around five per cent). Second, these average rates of emigration of skilled persons across the main LDC regions hide very substantial intra-regional variations, with coefficients of variation close to one in all regions except for East Africa (Table 3). Figures 1 6 summarise the data by country for the major regions of skilled out-migration among the LDCs. All regions, especially West and East Africa, show substantial variations in rates across countries, both in 1990 and Out-migration rates were especially high in several of the very small island countries, both in the South Pacific and elsewhere (Sao Tome and Principe, Cape Verde and Samoa), in countries that had experienced political instability in the 1980s 25 The main sub-regions among LDCs in Africa are East, Central and West Africa. Following UNCTAD practice, we have included one country from North Africa (Sudan) and one from Southern Africa (Lesotho) with Central and Southern Africa respectively (hereafter we refer to East, Central and West Africa as the main groups on the African sub-continent). For ease of exposition, Haiti, the only South American country, is included with West Africa in the regional grouping of the data, also following the current convention in UNCTAD publications on LDCs. 13

14 and 1990s (Sudan, Liberia, Mozambique, Somalia, Eritrea) and in some of the poorest countries (eg., Sierra Leone) (see Figure 7). They were lowest in some of the larger countries (the Congo, Sudan, Niger and Malawi, and in all the more populous Asian countries (especially land locked Nepal, and Myanmar; see Figure 8). Third, increases in out-migration among the tertiary educated to OECD countries were quite substantial: they increased by over 70 per cent in the 1990s, a figure only slightly less than the 96 per cent rise for all migrants (see the last two columns in Table 2). As Figures 1 6 indicate, rates of emigration were relatively stable across countries and regions over the ten year period. In general, the same countries recorded high emigration rates in 2000 and 1990, with one major exceptions: They increased very substantially in Madagascar and war torn Mozambique in the 1990s. Nevertheless, despite sustained high rates of migration from several countries, there was also some variation across countries in out-migration rates recorded in 1990 compared with By region, the greatest increase was for West Africa, followed by Central Africa, whereas out-migration declined among the tertiary educated in the Pacific Island economies in the 1990s. Very large increases in rates of out-migration were recorded in a handful of countries with several of the same characteristics noted above for high emigration countries. Thus, out-migration rates are estimated to have more than tripled in Angola, Mauritius, Guinea-Bissau and Burundi and were close to double rates in 1990 in Senegal and Guinea (in West Africa), Rwanda and Mozambique (in East Africa) and in the largest LDC economy, Bangladesh, in South Asia. At the other extreme, eight African, two Asian (Lao PDR and Bhutan) and all South Pacific countries registered a decline in out-migration rates from Declines were smaller than increases, but were nevertheless quite large. In East Africa, for example, Lesotho and Uganda both registered a falling stock of tertiary educated persons abroad: by 61 and 28 per cent respectively. Compared with Samoa and Kiribati in the Pacific, where declines were relatively small, the declines were substantial (40 and 47 per cent respectively) among the very small island states of Solomon Islands and Vanuatu. This suggests a return of some of the tertiary educated to these countries, despite the disappointing development record of the 1990s. What factors contributed to these high emigration rates? We have not undertaken a multivariate analysis of the determinants of out-migration. Nevertheless, correlation coefficients between tertiary levels of out-migration and several key variables confirm some of the observations made in the description of patterns and trends above (Table 4). Population size and the human capital index were (weakly) inversely correlated with out-migration rates (especially among the West African countries), while GDP was positively correlated with out-migration among educated people (again the strongest relationship was for West Africa). 26 The coefficient of variation for increases/decreases in migration rates for the entire sample was more than two for the period Data reliability probably contributed to this wide range in changes in migration rates for countries and regions. Docquier and Marfouk (2006) indicate that one needs to be especially careful in interpreting apparent trends in out-migration, given data problems. The data for 1990 are probably less reliable than for 2000 and in both years are based on estimates using data from neighbouring countries with similar characteristics. 14

15 Finally, to put these movements in perspective, we have assembled data which compare emigration rates in 2000 among the countries with the largest absolute number of outmigrants, and LDCs with the highest rates of emigration. The latter organised into two groups and ranked by the magnitude of emigration in each (Table 5). Two points stand out: First, the absolute number of tertiary educated out-migrants was relatively small among all LDCs, viewed on a global scale. Thus whereas several of the large exporters (the Philippines, India, China and Mexico) had around a million educated people living abroad in 2000, only Haiti among LDCs recorded close to 100,000 skilled emigrants. Most of the rest of the larger LDC exporters recorded a stock of around 20 40,000 tertiary educated people living overseas in The differences between the two groups of countries are partly a function of population size. It is also partly associated with low enrolment rates at tertiary level in the LDCs, compared with some of the globally large exporters of skilled manpower, such as the Philippines. Second, emigration rates among the educated were indeed very high by international standards in a number of LDCs. Table 5 indicates that among the large emigration countries only Jamaica recorded higher out-migration rates than Haiti (82 per cent), Cape Verde (69 per cent), Samoa (67 per cent), Somalia (59 per cent), Eritrea (46 per cent) and Mozambique (42 per cent). This was not simply a matter of scale. Although we noted above that emigration rates were high in some of the smallest countries, four LDCs with populations of five million or more ranked among the top ten countries in the world, in terms of emigration rates in 2000 (Docquier and Marfook, 2006: ). 27 Thus even for a sample of larger countries, high emigration rates were a feature of economic and social life among the LDCs. 1. More recent developments in LDCs The available data do not enable us to update the figures on rates of emigration among more educated and skilled people by country. However, we have attempted to estimate the magnitude of emigration to OECD countries for all LDCs based on the baseline data calculated by Docquier and Marfouk (2006). Assuming that outmigration rates in were at least similar to those in the previous decade, we 27 The four were Haiti, Somalia, Mozambique and Sierra Leone. Data in Table 5 for individual countries are taken from Docquier and Marfouk (2004), which has a relatively complete country list of emigration rates. In the revised version of their paper (which only reports the data for selected countries), the same two authors reported higher migration rates for the Lao PDR (37 compared with 14 per cent reported in the 2004 paper) and Uganda (36 per cent compared with 22 per cent in 2004) and Angola (33 per cent compared with 26 per cent in 2004); the later publication records a decline in emigration rates among the educated for Somalia (33 per cent recorded versus 59 per cent in 2004). For consistency, we have only used the data from the 2004 publication for the discussion of country trends in this paper. While the absolute rates do differ between the two studies, only Lao PDR (which was already the highest out-migration country among LDCs in Asia according to the 2004 study) changes significantly in ranking among the high emigration countries. 15

16 find that total emigration of skilled manpower from the LDCs probably amounted to around 15 per cent of their total stock of tertiary educated people aged 25 or above (Table 6), a share which probably increased, albeit only slightly, compared with the year In total this would have amounted to around one million tertiary educated people working abroad from LDCs, whose total stock of educated manpower amounted to around 6.6 million (including over one million in Bangladesh alone) in The composition of migrants to OECD countries from Africa, in particular, has been heavily dominated by educated persons. Thus OECD (2002) reported that 75 per cent of migrants from African countries to the OECD countries had completed tertiary education, many among them medical personnel doctors and nurses who had received subsidised training in their countries of origin. Many of these migrants came from low and middle income countries such as Ghana, Nigeria and South Africa, however, rather than the least developed countries in the region. 28 At the same time, there has been some evidence of substitution effects in LDCs. Outmigrants were being replaced by other nationals in several locations. Adedpoju (1988, cited in Gubert: 24) found early evidence of rural migrants or migrants from neighbouring countries replacing skilled international migrants in Burkina Faso, Senegal and Mali among the least developed countries Brain drain and remittances in LDCs How large were the remittances associated with these patterns of brain drain? International data sets do not breakdown of remittances by skill group, and hence the evidence is indirect with regard to the importance of remittances from educated persons abroad. 30 Table 7 presents data on remittances in selected years from the World Bank website for a collection of LDCs for which data appear to be plausible. 31 On average, excluding a number of extreme values in the calculation of changes over time, remittances per capita appear to increased quite significantly in LDCs in the 1990s and even more in the first half of the 21 st century. The mean value rose three fold from around $200 million in 1990 to just under an estimated $600 million in Remittances are highly correlated with both total rates of emigration to OECD countries and out-migration rates among skilled workers (for both a correlation coefficient of 0.79 between the value of remittances and migration rates in 2000). 28 For example, Adebusoye (2006, cited in Gubert, n.d.: 23) reported that close to one-quarter of Ghanaian health workers trained in migrated, including around two-thirds of all general practitioners. 29 Presumably the rural-urban migrants tended to replace less skilled emigrants (although they may also consist of rural doctors or nurses moving to higher paid jobs in the cities a kind of internal brain drain that is not uncommon in many countries). Gubert suggests that international migration may be a final step after skilled migrants move from villages to cities, to other cities and then abroad. 30 It should also be borne in mind that an estimated per cent of all remittances occur through informal channels and are not recorded in official balance of payments data (the World Bank 2006: Chapter 4). 31 The data need to be interpreted with care, given that the reliability of coverage appears to differ significantly for individual countries from year to year. 16