EMPLOYMENT PAPER 2002/33. Trade and international. labour mobility

EMPLOYMENT PAPER 2002/33 Trade and international labour mobility Ajit K. Ghose

Copyright International Labour Organization 2002 ISBN 92-2-113079-7 Publications of the International Labour Office enjoy copyright under Protocol 2 of the Universal Copyright Convention. Nevertheless, short excerpts from them may be reproduced without authorization, on condition that the source is indicated. For rights or reproduction, or translation, application should be made to the ILO Publications Bureau (Rights and Permissions), International Labour Office, CH-1211 Geneva 22, Switzerland. The International Labour Office welcomes such applications. Libraries, institutions and other users registered in the United Kingdom with the Copyright Licensing Agency, 90 Tottenham Court road, London W1P 9HE (Fax:+44 171 436 3986), in the United States with the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923 (Fax:+ 1 508 750 4470), or in other countries with associated Reproduction Rights Organizations, may make photocopies in accordance with the licences issued to them for this purpose. The designations employed in ILO publications, which are in conformity with United Nations practice, and the presentation of material therein do not imply the expression of any opinion whatsoever on the part of the International Labour Office concerning the legal status of any country, area or territory or of its authorities, or concerning the delimitation of its frontiers. The responsibility for opinions expressed in signed articles, studies and other contributions rests solely with their authors, and publication does not constitute an endorsement by the International Labour Office of the opinions expressed in them. Reference to names of firms and commercial products and processes does not imply their endorsement by the International Labour Office, and any failure to mention a particular firm, commercial product or process is not a sign of disapproval. ILO publications can be obtained through major booksellers or ILO local offices in many countries, or direct from ILO Publications, International Labour Office, CH-1211 Geneva 22, Switzerland. Catalogues or lists of new publications are available free of charge from the above address.

Foreword This paper is an output of the research project on global economic integration and employment policy carried out by the Employment Strategy Department. Globalisation has so far meant increasingly freer cross-border movement of goods and capital but not of labour. This appears anomalous to economists, who generally expect globalisation to increase migratory pressures and regard the severely restrictive immigration policies, currently enforced by industrialised countries, as inappropriate. Yet, the linkage between the growth of international trade (the central feature of the globalisation process) and cross-border movement of labour remains an under-researched subject. This context defines the point of departure for this paper. It seeks to answer, in theoretical and empirical terms, two questions of much current interest. Are there reasons to think that the growth of trade is adding to the pressures for increased international migration? Is freer trade encouraging brain drain from developing countries? The answers suggested by the analysis presented in the paper are as follows. Theoretical reasoning does not suggest that growth of trade either adds to or reduces migratory pressures. It does suggest, however, that the emerging pattern of North-South trade encourages emigration of skilled labour from South. The available empirical evidence, though admittedly inadequate, is consistent with the theoretical conjectures. In the 1990s, a period of rapidly expanding North- South trade, immigration (legal and non-legal) into the industrialised world actually showed a declining trend. On the other hand, the ratio of the skilled in the immigrant labour force has been on the rise. These findings, together with the fact that the current immigration policies of the industrialised countries encourage clandestine immigration of low-skilled labour, underline the need for international cooperation in defining a global regulatory framework for international migration. Rashid Amjad Director a.i. Employment Strategy Department

Acknowledgement Thanks are due to Makiko Matsumoto for her invaluable assistance in carrying out statistical exercises and to Manolo Abella, Azfar Khan and Rolph van der Hoeven for their helpful comments and suggestions. The author alone is responsible for the views expressed.

Contents Page Foreword Acknowledgement 1. Introduction 1 2. Trade and international migration: insights from theory. 2 3. International migration: the empirical evidence. 6 Some preliminaries. 6 International migration: some general facts. 6 Immigration into industrialised countries 8 The phenomenon of brain drain.. 16 International migration: then and now. 23 4. Conclusions.. 24 Appendix: Statistical Tables.. 27 References 41

1. Introduction It is by now well-known that the rapid growth of cross-border flows of commodities and capital since the mid-eighties has not been accompanied by any appreciable growth in cross-border labour flows. To most observers, this lack of growth in international labour mobility appears surprising and economists generally attribute it to the policy stance of governments in receiving countries. A widely shared view is that globalization would have substantially increased international migration but for the restrictive immigration policies prevailing in the industrialised countries. 1 Historical experience seems to lend plausibility to this view. Massive international migration was a feature - perhaps even the most important feature - of the globalization process of 1870-1910. 2 The fact that international migration does not figure prominently in the current period of globalization, therefore, does appear rather puzzling and calls for an explanation. The search for such an explanation inevitably brings into focus the restrictive immigration policies of the industrialised countries of Europe, North America and Oceania, particularly since these stand in sharp contrast with their increasingly liberal policies with respect to trade and capital flows. It is not obvious, however, why we should expect globalization to increase international migration. The central feature of the current process of globalization is expansion of manufactured trade between industrialised and developing countries. The standard model of trade, based on the idea of comparative advantage arising from differences in factor endowments between countries, in fact suggests that increased cross-border flows of goods serve to reduce cross-border labour mobility. There are, of course, other trade theories which predict a positive correlation between trade and migration, but these theories are not generally regarded as relevant in explaining trade between industrialised and developing countries in the current period. It is also far from obvious that the historical experience provides a good basis for thinking about international migration today; both trade and labour flows in today s world are in fact qualitatively different from those observed during the earlier period of globalization. Mainly unskilled workers, for example, migrated during the earlier globalization period while brain drain is a key feature of international migration today. Evidently, the linkage between trade and international migration remains inadequately explored. This makes it difficult to develop a coherent view of the role of international migration in a globalizing world. Meanwhile, there are other developments that make international migration an increasingly important area of policy concern. There is a strong inverse relation between the rate of population growth and the level of development across countries and it is fairly obvious that, for decades to come, the incremental world labour force will be located almost entirely outside the industrialised world. These demographic trends, by themselves, point to a 1 See, for example, the papers in Faini, de Melo and Zimmermann, eds. (1999) and World Bank (2001). 2 See O Rourke and Williamson (1999). It should be noted that a longer period, 1850-1913, is often referred to as the globalization period. The process is thought to have begun with the repeal of the Corn Laws in Britain in 1846 and ended with the commencement of the First World War.

2 need for increased migratory flows. 3 They also imply a widening of the already huge gap in skill endowments between developed and developing countries. It is obviously important to know, in thinking about policy, whether and how growth of trade might alter these possibilities. Against this backdrop, this paper addresses two main questions. First, are there reasons to believe that the widespread trade liberalisation and the consequent growth of trade since the early eighties has added to the pressures for increased international migration? Second, what are the actual or likely effects of the growth of trade on brain drain from developing countries? The existing literature on international migration, though large, do not provide answers to these questions. Much of this literature has been concerned with assessment of the effects of immigration on employment and wages of native workers in host countries, the extent to which immigrants are a net burden on the economy of host countries and the extent to which emigration benefits the source countries by generating inflows of remittances. 4 Issues of social protection of migrant workers have also figured quite prominently. 5 The problem of brain drain received some attention in the seventies 6 but has not figured prominently in recent discussions. Most importantly, the issues of international migration have most often been discussed in abstraction from developments in the area of trade. 2. Trade and international migration: insights from theory Economic analysis of international migration focuses on the movement of workers from less developed to more developed countries (from developing to industrialised countries in schematic terms) in search of higher wages and better working conditions. 7 In the absence of legal restrictions on immigration, cross-border labour mobility could thus be assumed to depend on the size of the gap in labour-income (wages, working conditions and social security arrangements) that exists between industrialised and developing countries and on the extent of information on this gap that is available to potential migrants. Migration increases when either the gap widens or more information on the gap becomes available or both happen. However, migration is not costless; the process of relocation involves substantial transportation, communication and search costs for the individuals concerned. Trends in international migration, therefore, are also influenced by the trends in these costs; migration increases when costs decline. 8 3 These facts are well-documented in UN (2000). (1993). 4 See Borjas (1994, 1995, 1999), Borjas and Freeman, eds. (1992), Zimmermann (1995) and Freeman 5 See, for example, Stalker (1994). 6 See Adams (1968), Bhagwati and Hamada (1974) and Bhagwati and Partington (1976). 7 As we shall see below, migration for economic reasons is not the only, perhaps not even the most important, form of migration in the real world. But migration for non-economic reasons is random in character and does not lend itself to systematic analysis. 8 There are also substantial disutility costs associated with the relocation from one s own social-culturallinguistic context into an alien one. These costs, though generally ignored in the literature (see, however, Faini, 1999), are in fact among the most important factors that set natural limits to international migration.

A well-recognised implication of the mainstream trade theory - the Heckscher-Ohlin model - is that product flows and factor flows (i.e., flows of capital and labour) are substitutes. 9 Free trade eliminates cross-border movement of factors of production; it is autarky that calls for factor mobility. These propositions follow fairly obviously from the idea that comparative advantage derives from differences in relative factor endowments. Free cross-border movement of factors eliminates these differences and hence eliminates trade. On the other hand, free trade eliminates the incentive for factor mobility because it leads to factor price equalisation (i.e., eliminates the gap in labour-income between the trading countries). Simultaneous growth of trade and capital flows, however, is as much a feature of the current process of globalization as it was of the earlier one; the two types of flows seem to be complements rather than substitutes in the real world. This appears to contradict the prediction of the Heckscher-Ohlin model which (in its original formulation) defines comparative advantage in terms of differences in endowments of capital and labour - the two basic factors of production. Recognition of this problem has led to a reformulation of the notion of comparative advantage. With free movement of capital, which is now regarded as a standard assumption, comparative advantage cannot arise from differences in relative endowments of capital and labour; these differences simply disappear. So a new source of comparative advantage has to be identified and the widely accepted new source is the difference in relative endowments of skilled and unskilled labour. 10 A typical industrialised country is relatively abundant in skilled labour and hence has comparative advantage in the production of skill-intensive goods; a typical developing economy is relatively abundant in unskilled labour and hence has comparative advantage in the production of labour-intensive goods. Trade between these two types of countries, therefore, involves exchange of skill-intensive goods for labour-intensive goods. On this view, growth of trade reduces the demand for skilled labour in skill-scarce developing countries and increases the demand for skilled labour in skill-abundant industrialised countries. This leads to a widening of the gap in labour-income of skilled workers between industrialised and developing countries; wages of skilled labour increase in industrialised countries (because of increased demand) and decline in developing countries (because of reduced demand). In the case of unskilled labour, however, growth of trade increases the demand in labour-abundant developing countries and reduces the demand in labour-scarce industrialised countries so that there is a narrowing of the gap in labour-income. 11 Increased trade, therefore, simultaneously encourages migration of skilled labour and discourages migration of unskilled labour from developing to industrialised countries. Thus, in a world without restrictions on labour mobility, increased trade increases brain drain from developing countries but has uncertain effects on overall migration. Trade and flows of skilled labour are complements while trade and flows of unskilled labour are substitutes. But would increased mobility of skilled workers not undermine the basis for trade between industrialised and developing countries? The answer is no. There is a basic difference between capital flows and skill flows: while capital flows are (at least are expected to be) from capitalabundant to capital-scarce countries, skill flows are from skill-scarce to skill-abundant countries. Skill flows, therefore, leads to a strengthening of the initial comparative advantage of the trading 3 9 The first formal demonstration of the result is to be found in Mundell (1957). 10 See Wood (1994) for an elaborate exposition. 11 Ghose (2000) observes that, in developing countries, trade generally increases the employment of unskilled labour but does not always increase wages. Since wages of unskilled labour in industrialised countries tend to decline, however, the gap in labour-income still narrows.

4 countries; skill-abundant countries become even more skill-abundant and labour-abundant countries become even more labour-abundant. Thus growth of trade sets in motion a process in which comparative advantage and mobility of skilled labour reinforce each other in turn. 12 These implications of trade theory put the problem of brain drain in a fresh perspective. The existing literature basically focuses on the short-run costs of brain drain to source countries. These costs, to be sure, are substantial. Higher education is heavily subsidised in developing countries and skilled migrants carry away scarce human capital built through public investment. 13 Skilled migrants are also less likely than unskilled migrants to send back remittances for the simple reason that the former usually are from relatively well-off households that do not require income support on a regular basis. 14 However, so long as brain drain is not a self-reinforcing problem, it is sensible to think in terms of suitable compensation schemes. 15 But growth of trade seems to actually make the problem self-reinforcing. To the extent that trade between industrialised and developing countries arises from differences in relative endowments of skilled and unskilled labour, trade expansion combined with unrestrained mobility of skilled workers could conceivably make accumulation of human capital an unattainable objective for many developing countries. 16 This possibility must be viewed with concern because, as modern growth theory emphasises, failure to accumulate human capital will adversely affect the long-term growth prospects of developing countries; they are likely to find it difficult to move beyond specialisation in simple labour-intensive products. 17 The relation between product flows and factor flows, of course, turns out to be quite different when a sector-specific and inherently immobile factor (eg. land) is required to produce 12 The possibility of trade widening the initial gap in skill endowments has been noted by Wood and Ridao-Cano (1999), but they emphasise a different mechanism. In their model, the usual assumption of labour immobility across countries is retained but skill supply in a country is assumed to be responsive to the skillpremium. Since, as a result of trade, this premium rises in industrialised countries and declines in developing countries, skill supply expands in industrialised countries and contracts in developing countries. The initial comparative advantage of each of the two types of countries is strengthened as a result. Thus there is a sustained process of widening of the gap in skill endowments. 13 See UNDP (2001), pp. 91-93 for some relevant information on subsidies on higher education. 14 There are examples of skilled migrants contribution to development of their home countries through investment and knowledge-sharing. There are also examples of return migration of skilled workers. But all this typically happens when the home countries have already acquired economic dynamism. 15 Such schemes were suggested, for example, by Bhagwati and Partington (1976). 16 The argument developed above considers trade in final products only. However, a good deal of trade now occurs in intermediate goods as a result of outsourcing by firms in industrialised countries. It has been argued by Feenstra and Hanson (1997, 2001) that the intermediate goods outsourced from developing countries are more skill-intensive than the final products exported by them. In that case, the effect of trade on the skill premium in developing countries depends on the mix of final products and intermediate goods in their export basket; the higher the share of intermediate goods, the lower is the decline in the skill premium. Even if this is true, however, our argument is not much affected. For, if outsourcing moderates the decline in the skill premium in developing countries, it also simultaneously accentuates the rise in the skill premium in industrialised countries (since the least skill-intensive intermediate goods are outsourced). Thus the rise in the gap in labour income of skilled labour between the two types of countries is not prevented by outsourcing. Rodrik (2000). 17 The argument is developed in Grossman and Helpman (1991), Matsuyama (1992) and Rodriguez and

a tradeable good. In the specific-factors model 18, the basis for trade is also traced to differences in relative factor endowments, but one of the trading countries (country 1, say) is assumed to be abundant in a specific factor (which needs to be combined with the non-specific factors such as capital and labour for purposes of production). The other trading country (country 2) is assumed to be abundant in the non-specific factors. Trade between these two countries will then entail flows of the non-specific factors from country 2 to country 1; trade flows and factor flows are complements. Intuitively, the reasons are easy to see. Growth of trade will require each of the countries to expand production of the good in which it has comparative advantage. This will require increased availability of the non-specific factors in country 1 and the required increase in availability can be achieved through flows from country 2. 19 The incentives necessary for such flows to occur exist because the relative prices of the non-specific factors are higher in country 1 (where they are scarce) than in country 2 (where they are abundant). This framework, as we shall see below, is useful for analysing the observed patterns of trade and factor flows during the earlier globalization period when manufactures were traded for agricultural and mineral products. But it is inappropriate for studying trade and factor flows in today s world. As already noted, the key feature of the current process of globalization is expansion of trade in manufactures between industrialised and developing countries 20 and manufacturing production does not require specific factors. There is also empirical evidence to support the idea that the basis for manufactured trade between industrialised and developing countries lies in differences in relative endowments of skilled and unskilled labour between the two types of countries. 21 There is a variety of other models of trade which also imply complementarity between product flows and factor flows. In these models, trade arises not because of differences in factor endowments between countries but because of differences in technology or market structure, or because of the existence of increasing returns to scale in the production of traded goods. 22 However, since these models assume away differences in factor endowments, they are more appropriate for analysing migration between countries with similar economic structures and levels of development (eg. migratory flows within the European Union) than for analysing migration between developing and industrialised countries. 5 18 Formulations can be found in Jones (1971), Samuelson (1971) and Jones and Neary (1984). As in the case of the Heckscher-Ohlin model, there are several variants of the specific-factors model. Here we consider the variant that seems most appropriate in the context. 19 In the beginning, non-specific factors will flow from other sectors within country 1, but the assumption of relative scarcity of the non-specific factors implies that there are strict limits to this. 20 See Ghose (2000) for an elaboration of the point. 21 See Wood (1994), Mayer and Wood (2001) and Ghose (2000) for such evidence. 22 See Krugman (1980. 1991) and Helpman and Krugman (1985). A useful comparative analysis of the implications of different trade models for migration is available in Venables (1999).

6 Some preliminaries 3. International migration: the empirical evidence Empirical analysis of the effects of trade on international migration is rendered extremely difficult by the paucity and limitations of statistical data. The available data are rather sketchy. Moreover, despite the efforts made by the United Nations to standardise definitions and methods of collection and compilation of statistics on migration 23, these still vary widely across countries. Thus inter-country comparisons are hazardous. Besides, the available statistics already incorporate the effects of the restrictions on immigration prevailing in receiving countries; they do not tell us what the trends and patterns of migratory flows would have been in the absence of the restrictions. Finally, very little firm data on clandestine migration are available, though this type of migration is known to be significant. Methodological difficulties are also considerable. There clearly are economic forces other than trade which influence international migration in today s world and it is not easy to empirically isolate the effects of trade in such a context. The current pattern of population growth across countries, as already noted, creates a need for increased international migration. It also generates the necessary incentives. Ceteris paribus, the high growth of population in poorer countries has the effect of widening the gap in labour-income between low-income and high-income countries because absorption of the incremental labour force in low-income countries tends to be associated with declining productivity. Another factor is the rapid development of transport and communication facilities, which is expected to stimulate migration since it reduces the material costs of migration (arguably, this also reduces the disutility costs of living in an alien environment) and increases the availability of information on the gap in labour-income between industrialised and developing countries. These difficulties mean that we must combine a priori reasoning and intuitive understanding with the sketchy empirical evidence that is available so as to be able to draw even tentative conclusions. In what follows, the focus is on studying the trends in migration flows and in brain drain in the nineties - a period of rapid trade expansion. In so far as the other factors - the demographic changes and the improvements in transport and communication facilities - have been at work for a longer period, the deviations of the trends in international migration the nineties from the long-term trends could be plausibly attributed to trade expansion. It is also useful to bear in mind that the demographic factors and the declining costs of migration are expected to increase migratory flows while the growth of trade is only expected to increase the average skill level of the migrants. International migration: some general facts According to UN estimates, the stock of migrant population in the world in the year 2000 was approximately 150 million; this constituted just 2.5 percent of the world population. It is wellknown that the majority of the migrants comes from developing countries (this is hardly surprising since a large majority of the world s population lives in developing countries). A less well-known fact is that the majority of the migrants also lives in developing countries; in 1998, for example, an estimated 58 percent of the migrants lived in developing countries. In fact, migratory flows most often are between neighbouring or geographically proximate countries; inter-continental migration accounts for only a small proportion of total migration. One study shows, for example, 23 See United Nations (1998).

that well under 10 percent of the migrants from Asian countries (excluding China) left Asia during 1974-1994. 24 These patterns are linked to the fact that causes of migration are often noneconomic. Though reliable statistical figures are difficult to establish, it is well-recognised in the literature that much of the observed migration is provoked by wars, civil conflicts, natural disasters, systemic transitions (as in the case of Eastern Europe and erstwhile Soviet Union) or political instability. 25 Many of the migrants are refugees or asylum seekers. Table 1: International migration, 1965-2000 7 1965 1975 1985 1990 1998 2000 Stock of migrant population (millions) World Developed countries Western Europe North America Oceania Developing countries 75 27 12 13 2 48 84 35 17 15 3 49 105 45 20 20 4 60 120 52 23 24 4 68 144 58 20 31 4 86 150 Migrants as % of population World Developed countries Western Europe North America Oceania Developing countries 2.3 4.1 2.7 5.9 17.6 1.9 2.1 4.9 3.6 6.2 19.4 1.5 2.2 5.9 4.2 7.7 21.5 1.5 2.3 6.5 4.6 8.6 22.8 1.6 2.4 6.9 3.9 10.4 19.8 1.7 2.5 Note: Source: The data refer to long-term migrants defined as those residing in foreign countries for more than a year. Estimates of migrant population are from Zlotnik (1998) and IOM (2000), except for those for the year 1998. The figures for 1998 are estimated by the author by using the relevant data from the above two sources as also from OECD ( 2001). Population estimates are from World Bank, World Development Indicators, CD-ROM, 2000. The stock of migrant population doubled in absolute terms between 1965 and 2000, but, as percentage of world population, it showed only a minor increase from 2.3% to 2.5%. The increase, moreover, was not steady. The percentage declined during 1965-75, regained the level of 1965 in 1990 and surpassed that level only in later years. The rise in the nineties, however, resulted from a deceleration in the growth of world population rather than from an acceleration in migration. The rate of growth in the stock of migrant population accelerated till 1990 but then decelerated; it was 1.1 percent during 1965-75, 2.2 percent during 1975-85, 2.7 percent during 1985-90 and 2.3 percent during 1990-2000. 24 See IOM (2000), p. 6. 25 See Zlotnik (1998), Stalker (2000) and IOM (2000).

8 The time-trends were noticeably different in industrialised and developing countries. The percentage share of the stock of migrant population in total population showed a steady rise in industrialised countries throughout the period 1965-98. In developing countries, it fell during 1965-75 and, though it showed a steady rise thereafter, it was still lower in 1998 than it had been in 1965. These differences, however, are explained more by the differences in the population growth than by the differences in the growth of migration. For the entire period, the stock of migrant population grew by 2.3 percent per annum in industrialised countries and by 1.8 percent per annum in developing countries, while population increased at annual rates of 0.8 percent in industrialised countries and 2.1 percent in developing countries. In the nineties, there were fairly dramatic shifts in the trends. The rate of growth in the stock of migrant population fell from 2.6 percent during 1965-90 to 1.4 percent during 1990-98 in industrialised countries while in developing countries it rose from 1.4 percent to 3.0 percent. This rather sharp deceleration of immigration into industrialised countries is explained by the fact that, in Western Europe and Oceania, the stock of migrant population actually declined in absolute terms in the nineties. Comparable estimates of the stock of migrant workers in the world are not directly available. However, we can use the available data for developed countries to estimate the workerpopulation ratio for the year 1998. Assuming this ratio - which works out to 0.515 - to be valid for the entire migrant population of the world, the number of migrant workers in the world in 1998 can be estimated as 74 million (2.6 percent of the world labour force). 26 Given our assumption, the distribution of the migrant workers across country-groups would obviously be exactly the same as that of the migrant population. The same would be true of the time-trends. Immigration into industrialised countries The observations above relate to overall migration and serve to define a general backdrop. They, however, also suggest that the pace of immigration into industrialised countries was decelerating in the nineties. It is this trend that is further explored in this section through an analysis of the available data on immigration of workers into OECD countries during the decade 1988-98, a decade of rapid growth of trade. 27 The stock of migrant labour force in OECD countries increased from around 22 million in 1988 to around 30 million in 1998; the share of migrant workers in total labour force increased from 5.9 percent to 7.3 percent during the same period (Table 2). 28 While these statistics indicate growing importance of migrant workers in these economies, they do not in fact mean rising immigration flows. Much of the observed increase is accounted for by the increase in the United 26 The figure for 2000 is 80 million. 27 The OECD countries, for the purpose of this paper, exclude Republic of Korea, Mexico and Turkey which are treated as developing countries. For a few countries, only incomplete data are available and these are also excluded. The basic data used here cover the following countries: Australia, Austria, Belgium, Canada, Denmark, Finland, Germany, Italy, Japan, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States of America. Even these data, it must be noted, suffer from quite serious limitations. For discussions of these limitations, see OECD (2000, 2001) and Coppel, Dumont and Visco (2001). 28 There are two other regions - the petroleum producing countries of the Middle East and the East Asian emerging economies - which are important destinations for migrant workers. Rough estimates show that, in 1997, the OECD countries accounted for 78 percent of all migrant workers based in the three regions. From the data available from OECD (1999, 2000), the stock of migrant workers in OECD countries in 1997 works out to 28.8 million. The available statistics for the Middle East and East Asia are presented in Appendix Tables 1 and 2.

States, which absorbed 68 percent of the additional migrant workers. Outside the United States, the share of migrant workers in the labour force showed generally insignificant growth and actually declined in several countries. More importantly, annual immigration flows were declining everywhere, even in the United States, during much of the nineties. As the Figures 1 and 2 show, following a short-lived surge in the late-eighties and the early-nineties (attributable to the change in political and economic regimes in the erstwhile Soviet Union and Eastern European countries) annual immigration actually declined in absolute terms during the rest of the period. 29 The increase in the relative importance of migrant labour was due not to increased immigration but to the extremely slow growth of the native labour force in these countries. Table 2: Immigration into OECD, 1988-98 9 Stock of migrant workers (in thousands) Labour force (in thousands) Migrant workers as % of labour force 1988 1998 1988 1998 1988 1998 Australia Austria Belgium Canada Denmark Finland France Germany Italy Japan Luxembourg Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom United States 2 066 161 285 2 517 65 8 1 557 1 911 262 417 69 176 49 46 58 220 608 871 10 758 2 294 327 375 2 896 94 35 1 587 2 522 352 670 135 210 67 89 191 219 691 1 039 16 100 8 101 3 494 3 961 13 985 2 873 2 525 24 589 39 072 23 784 62 533 160 6 642 2 104 4 785 15 476 4 471 3 416 28 007 122 250 9 563 3 797 4 184 16 363 2 969 2 628 26 481 41 024 25 339 68 261 183 7 378 2 305 4 984 17 323 4 780 3 837 29 528 137 850 25.5 4.6 7.2 18.0 2.3 0.3 6.3 4.9 1.1 0.7 43.1 2.6 2.3 1.0 0.4 4.9 17.8 3.1 8.8 23.8 8.6 9.0 17.7 3.2 1.3 6.0 6.1 1.4 1.0 73.8 2.8 2.9 1.8 1.1 4.6 18.0 3.5 11.7 Total 22 104 29 893 372 228 408 777 5.9 7.3 Western Europe 6 346 7 933 165 359 176 740 3.8 4.5 Source: Data on the stock of migrant workers are from OECD (2001) and the data on labour force are from World Bank, World Development Indicators, CD-ROM, 2000. 29 Figure 2 only serves to show that annual net immigration, data on which were available for only a few countries, closely followed the annual inflows of migrant population. Appendix Table 3 gives data on annual inflows for some other countries.

10 Figure 1. Annual inflows of migrant population (Index number, 1998 = 100) 300 250 200 WE7 USA 150 100 50 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 0 Note: The relevant data are presented in Appendix Table 3. 800 Figure 2. Annual net migration and inflows to WE6 (Index number, 1998 = 100) 700 600 500 Net migration Inflows 400 300 200 Note: The relevant data are presented in Appendix Table 4. 100 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 0 Note: The relevant data are presented in Appendix Table 4.

Indeed, the variation in the rate of immigration across countries had much to do with the variation in the rate of growth of native labour force. A simple regression exercise (Table 3) shows that the rate of growth of the migrant labour force across OECD countries is in large part explained by two variables: the initial share of migrant workers in total labour force and the rate of growth of native labour force during the period. Both of these variables varied inversely with the rate of growth of migrant labour force. A plausible interpretation of these results is as follows. Given the popular concerns about social consequences of immigration, there was a political compulsion to restrict immigration in countries with a high initial share of migrant workers in total labour force. On the other hand, there was an economic compulsion to allow substantial fresh immigration in countries with a low rate of growth of native labour force. The observed immigration flows resulted from the interaction between these political and economic compulsions. 11 Table 3: Regression results, dependent variable: annual rate of growth (%) labour force of migrant (1) (2) Constant 9.326*** (4.655) The ratio of migrant workers in total labour force, 1988-12.102 (-1.149) 11.381*** (5.449) -86.799** (-2.273) Annual rate of growth (%) of native labour force, 1988-98 Interaction of the two independent variables -7.846** (-2.184) 17.252* (1.779) -8.829** (-2.669) 63.255** (2.591) Adjusted R-square 0.28 0.42 F-statistic 3.34 5.15 Note: The data are presented in Appendix Table 5. For equation (1), the full sample of 19 countries were used; for equation (2), Luxembourg (a special case) was excluded from the sample. Figures in parentheses are t- statistics. * statistically significant at 10% ** statistically significant at 5% *** statistically significant at 1% In the OECD countries as a group, the relative importance of migration from developing countries increased; the share of migrant workers from developing countries in all migrant workers as well as that in total labour force showed substantial rises. Indeed, the entire increase in the migrant labour force during the period is accounted for by immigrants from developing countries (Table 4). Yet, these trends were not general across countries; they basically reflected the trends in a small number of countries outside Western Europe. More than 81 percent of the new migrant workers from developing countries went to the United States; another 11 percent went to Canada and Australia. In Western Europe, the relative importance of migrant workers

12 from developing countries showed no significant growth. At any rate, a more significant fact, as Figures 3 and 4 show, is that annual immigration from developing countries was declining in absolute terms virtually everywhere, even in the United States, for much of the nineties. 30 Table 4: Stock of migrant workers from developing countries, 1988-98 Stock (in thousands) As % of labour force As % of all migrant workers M 1988 1998 1988 1998 1988 1998 1988-98 Australia Austria Belgium Canada Denmark Finland France Germany Italy Japan Luxembourg Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom United States 397 34 43 179 14 1 564 604 138 246 0 61 8 24 18 10 30 215 5 777 579 53 84 1 019 16 5 635 744 172 389 0 65 6 49 155 13 33 234 13 041 4.9 1.0 1.1 1.3 0.5 0.04 2.3 1.5 0.6 0.4 0.0 0.9 0.04 0.5 0.1 0.2 0.9 0.8 4.7 6.0 1.4 2.0 6.2 0.5 0.2 2.4 1.8 0.7 0.6 0.0 0.9 0.03 1.0 0.9 0.3 0.9 0.8 9.5 19.2 21.1 15.1 7.1 21.5 12.5 36.2 31.6 52.7 59.0 0.0 34.6 16.3 52.2 31.0 4.5 4.9 24.7 53.7 25.2 16.2 22.4 35.2 17.0 14.3 40.0 29.5 48.9 58.0 0.0 30.9 8.9 55.0 81.1 5.9 4.8 22.5 81.0 79.8 11.4 45.5 221.6 6.9 14.8 236.7 22.9 37.8 56.5 0.0 11.8-11.1 58.1 103.0 # 3.6 11.3 136.0 Total 8 363 17 292 2.2 4.2 37.8 57.8 114.6 Western Europe 1 764 2 264 1.1 1.3 27.8 28.5 31.5 Note: M = (increase in the stock of migrant workers from developing countries/increase in the stock of all migrant workers) * 100. Developing countries include Turkey but exclude the transition countries of Easter Europe. Given the nature of the data available, the figures slightly underestimate the number of migrants from developing countries. # While the stock of migrants from developing countries increased by three thousand, the stock of all migrants declined by one thousand. Source: Estimated from the data in Table 2 and other data available in OECD (1999) and (2000). 30 Appendix Table 6 gives data for some other countries.

13 Figure 3. Annual inflows of migrant population from developing countries (Index number, 1998 = 100) 300 250 200 WE7 USA 150 100 50 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 0 Note: The relevant data are presented in Appendix Table 6. 400 300 Figure 4. Annual net migration and inflows from developing countries into West Europe 6 (Index number, 1998 = 100) Net migration Inflows 200 100 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 0 Note: The relevant data are presented in Appendix Table 4.

14 The evidence clearly suggests that legal immigration into industrialised countries, from developing countries as well as from others, was declining in a period when trade flows and capital flows increased rapidly. Immigration policies, it would seem, had something to do with this, particularly since there is evidence to show (as we have seen) that identifiable economic and political compulsions influenced legal immigration. But this explanation looks inadequate once we consider what the trend in immigration into OECD countries would have been had immigration policies been liberalised. To see this, we need to focus on the trends in the number of asylum seekers and in clandestine immigration on the assumption that a liberalised immigration regime would have meant granting legal immigrant status to all of them. Some data on annual inflows of asylum seekers for the period 1990-98 are available. Figures 5 and 6 show, for Western Europe and the United States, what the trends in immigration would have been had all asylum seekers been granted legal immigrant status. A comparison with Figures 1 and 3 makes it obvious that the trends would not have been very different; migratory flows would still have declined in much the same way. Unfortunately, no proper analysis of trends in clandestine migration is feasible. We know that this type of immigration is significant but have no precise idea of the magnitudes involved. However, it is possible argue a priori that only a very high growth of clandestine immigration could have substantially modified the observed trends in annual immigration flows (legal immigration + inflows of asylum seekers). In the case of a group of seven West European countries, for example, even to sustain the annual immigration flow at the level reached in 1992, the stock of clandestine immigrant workers would have had to increase by about 5.2 million during the period 1992-98 (or by nearly 0.9 million per annum). In the case of the United States, to sustain the annual immigration flow at the level reached in 1991, the stock of clandestine immigrant workers would have had to increase by 5.8 million during the period 1991-97 (or by nearly 1 million per annum). These numbers seem improbably large 31 and would have to be even larger if annual immigration flows under a liberalised regime were to show a rising trend. 31 Latest estimates suggest the annual inflow of clandestine migrants to be currently around 0.5 million to Europe and around 0.3 million to the United States. See Coppel, Dumont and Visco (2001), p.10. Between 1992 and 1998, the number of clandestine immigrants to the European Union countries is estimated to have increased by around 1.3 million. See World Bank (2001), p. 80. According to estimates made by the Immigration and Naturalization Service of the USA, clandestine immigration into that country increased by 1.6 million between 1992 and 1996 (i.e., by 0.4 million per annum). See Solimano (2001), Table 2.It should be noted too that these figures relate to total clandestine immigration rather than clandestine immigration from developing countries.

15 140 Figure 5. Gross inflows of foreign population, legal immigrants and asylum seekers (Index number, 1991 = 100) 120 100 80 1990 1991 1992 1993 1994 1995 1996 1997 1998 60 40 20 0 Note: The relevant data are presented in Appendix Table 7. WE7 USA 100 90 80 70 60 50 40 30 20 Figure 6. Gross inflows of foreign population, legal immigrants and asylum seekers, from developing countries (Index number, 1991 = 100) Note: 10 The relevant data are presented in Appendix Table 7. 0 WE7 USA 1990 1991 1992 1993 1994 1995 1996 1997 1998 Note: The relevant data are presented in Appendix Table 7.

16 The conclusion that emerges from these observations is that there was a downward trend in total immigration, not just in legal immigration, from developing to industrialised countries in the nineties. Since it cannot be convincingly argued that immigration policies determined non-legal labour inflows, 32 the possibility that the downward trend in legal immigration was a mere reflection of the downward trend in migratory flows in general cannot be ruled out. At any rate, the fact that immigration into industrialised countries was declining certainly rules out the possibility that trade expansion created new pressures for increased international migration. What appears puzzling is that the growing income-gap between rich and poor countries and the declining costs of transport and communication (one could add the rising incidence of political crisis and civil conflict) do not seem to have had the expected effects on international migration. The explanation probably lies in the fact that such factors only influence migration of low- or medium-skilled workers, which tends to be confined to geographically contiguous or proximate countries (see below). Thus Mexican migration to the United States is most likely to increase if the gap in per capita GDP between the two countries increases or if the costs of migration to the United States decline. But similar developments can hardly be expected to increase migration from Tanzania or Senegal to the United States. Being geographically remote, the poorest countries of the world have in fact never been important suppliers of migrant workers to the richest countries. It is simply implausible to suppose that a rise in the income-gap between the richest and the poorest countries of the world or a general decline in transport and communication costs have global effects on international migration. To study the effects of rising income inequality or declining migration costs, we need to focus attention on particular cases such as Turkish migration to West Europe or Mexican migration to the United States. 33 The phenomenon of brain drain Brain drain refers to the fact that the average skill level of the migrants from developing to industrialised countries tends to be much higher than that of the population of their home countries. As such, international migration has the effect of augmenting the stock of human capital in industrialised countries and of depleting it in developing countries. Moreover, the cost of augmenting the stock of human capital in industrialised countries is in effect borne by developing countries. These tendencies seem rather perverse and worrying. Modern growth theories identify human capital as a major source of economic growth. 34 On this view, the brain drain associated with international migration improves the growth prospect of industrialised countries and worsens that of developing countries. 35 It thus serves to accentuate global inequality in the long run. Although the problem of brain drain has been a subject of debate among economists since 32 It can perhaps be argued that increasingly restrictive immigration policies serve to increase migration costs and hence reduce clandestine immigration and discourage would-be asylum seekers. But empirical support for such an argument is missing. On the other hand, there is some evidence to suggest that clandestine immigration is often officially tolerated since it is a source of cheap low-skilled labour for labour-intensive industries. See, for example, Hanson and Spilimbergo (1999a). 33 Hanson and Spilimbergo (1999) find that attempted illegal migration from Mexico to the United States is very sensitive to changes in the gap between US and Mexican wages. 34 See, for example, Romer (1986) and Lucas (1988). The empirical works of Barro (1991), and Barro and Lee (1993 and 1994) are also relevant in this context. 35 For a formal demonstration of this result, see Haque and Kim (1995).

the 1960s 36, empirical evidence on the seriousness of the problem remains scant. Even general information on international migration, as we have seen, is scarce. Information on characteristics of the migrant population is even scarcer. Some newly available data, however, do allow us to get an idea of the ubiquity and the quantitative significance of brain drain. Table 5 presents estimates of average years of education of the population of a fairly large number of developing countries as also of the emigrants from those countries to the United States in the year 1990. It reveals that the number of years of education received by an average emigrant was, in most cases, more than twice as many as that received by an average person of his country of origin. Indeed, the average emigrant from most of the countries was better educated than the average native American (who had 11.9 years of education). The estimates refer to emigration to the United States alone, but there are no good reasons to think that similar estimates for other OECD countries, if these could be derived, would show radically different patterns. Brain drain clearly is a very general feature of migration from developing to industrialised countries. 37 17 36 See, for example, Adams (1968), Bhagwati and Hamada (1974) and Bhagwati and Partington (1976). 37 More than 36 percent of the migrants from developing countries to the United States in 1990 had a tertiary education. The percentage rises to more than 62 for the migrants from Asia-Pacific and Africa.