Introduction

1The issue of brain drain has returned to the forefront of public debate in recent years, if indeed it had ever left it. Although this debate has tended to revolve around losses of highly qualified persons from developing countries to developed countries, any migration represents a transfer of human capital from the sending to the receiving country, the “value” of which depends on its productive potential. That potential is generally viewed as the product of education. If the education has been financed by the sending country, then migration represents a net loss to the sending country, because the returns to the investment in education will accrue, in the first instance, to the migrant, but also to the society and economy of the receiving country. And because the investment in and returns to tertiary qualifications tend to be high, the focus has naturally been on the migration of persons holding such qualifications.

2None of this is terribly new. However, there have been a number of developments which have redirected attention to the brain drain but also have nuanced the outcry about its impact. The first of these is the fact that in recent years the concern over brain drain spread to developed countries in the wake of the dot com bubble, when a certain number of oecd countries feared the loss of young ict professionals to countries such as the United States where the informatics and telecommunications boom seemed to be acting as a magnet for talent (e.g. Hugo et al. [2003], Ferrand [2001], Saint-Paul [2004]).

3At the same time, the contribution of innovation to technological development and economic growth focused attention on the research and development which was at the heart of innovation, and at the small coterie of researchers and engineers deemed to be responsible for this. Here again, it was feared that the attraction of countries where technological innovation and research infrastructure were strong would draw away national talent seen to be vital in maintaining national competitivity and dynamism.

4In developing countries, the concern over brain drain, although still present, was somewhat muted by the growing realisation of the contribution of migrants’ remittances to the welfare of populations in sending countries, and to bringing in needed foreign exchange to help in the payment of external debt (Ratha [2003]). The fact that the propensity to remit seemed to be higher among low qualified migrants (Faini [2003]), and that sending countries have an obvious surplus of these, tended to nuance the formerly negative image of migration associated with brain drain. A better understanding of the contribution of national diasporas and of returnees in some countries to home country development has also had an influence on the negative view of brain drain (Barre et al. [2003], Lowell and Gerova [2004]), discussions of which now tend to be reformulated in terms of “brain circulation”. [1]

5On the other hand, the issue of brain drain has returned, not in general, but as a result of what appear to be growing shortages of medical and educational personnel in developed countries and the inevitable pull this has exerted on the labour force in these occupational domains in sending countries. [2] This has been viewed as a particularly pernicious development, because of the potential for undermining the capacity of developing countries to develop their own human capital.

6Despite the longstanding and renewed interest in the brain drain however, there has been little in the way of solid statistics to inform the debate. The relative scale of movements within oecd countries and from developing countries to oecd countries has remained a matter for speculation. Carrington and Detriagiache [1998] produced estimates based on United States census data on the foreign-born population by country of birth and nationality and oecd data on stocks of the foreign population, but the estimates were made using a number of rather heroic assumptions and the data have become somewhat dated. Although they have been updated by Adams [2003] on the basis of 2000 United Sates census data, the assumptions remain, along with the uncertainty of their impact on estimates of expatriation of the highly qualified to oecd countries.

7Now however, for the first time, a database of the foreign-born population by educational attainment has been compiled for almost all oecd countries, making it possible to examine the destination and educational attainment of expatriates from countries across the globe to the oecd area. This has been made possible by the fact that with the 2000 round of censuses, virtually all oecd countries have incorporated in their census a question on the country of birth of persons enumerated, as well as on their nationality. With this information, it is possible to provide, for the first time, a detailed, comparable and reliable picture of immigrant populations within oecd countries, reflecting the cumulative effect of movements within and to the oecd area over the past decades. Not only can immigrant populations be compared on a common basis across countries, but the extent of migration from a single source country to each oecd country, as well as to the oecd as a whole, can be determined. And with additional information on the educational attainment of migrants, the cumulative impact of flows of human capital can be depicted and, in particular, the conventional wisdom on the brain drain confronted with actual data. As will be seen, the picture which emerges will confirm a certain number of commonly held views about international migration movements, but also reveal quite a few surprises.

8This paper is divided into four main sections. The first section describes the new database on immigrants and expatriates in oecd countries and non member countries that is the basic source of the information in this paper. The second section presents the key results derived from this database. The third and fourth sections respectively discuss briefly the recent literature on brain gain and analyse the basic covariates of highly skilled emigration in the light of the new empirical evidence compiled. A summary and conclusions follow.

A new comparable database on international migrants

9The new database on immigrants and expatriates in oecd countries (see Box 1) is the first internationally comparable data set with detailed information on the foreign-born population for almost all member countries of the oecd. In addition, using the data base, it is possible to calculate “emigration rates” to oecd countries by level of qualification and country of origin for approximately 100 countries. This provides a broad view of the significance of highly skilled emigration, for both oecd and less developed countries.

10Until the constitution of the data set described in this paper, there was only very limited data on the extent of international mobility of the highly skilled. One study by Carrington and Detragiache [1998], which has recently been updated by Adams [2003], relies on United States census data on the foreign-born and oecd immigrant stock data from the annual publication Trends in International Migration (oecd) to construct a data base for emigration by level of education and by country of origin. The authors use the United States 1990 census data to determine the educational profile of immigrants by country of birth and apply it to immigrants (in many cases, foreigners) living in other oecd countries to estimate the total stocks of migrants by level of education and country of origin. The Barro and Lee [1993] database on educational attainment levels is the source for the stock of the population by level of education in countries of origin. This then becomes the denominator of reference to estimate the emigration rates.

11The estimates based on this methodology are subject to a number of limitations. One significant problem concerns the assumptions made because of data availability limitations. In particular, the foreign-born population in eu countries is assumed to be the foreign population and foreigners of a particular nationality are considered to have the same educational profile as the foreign-born of the United States. As a result, the estimates tend to be problematical for small source countries and countries whose citizens tend to migrate to countries other than the United States. In addition, Cohen and Soto [2001] have shown that the Barro and Lee [1993] database on educational attainment is of uneven quality.

12The database on immigrants and expatriates has direct measures of the educational attainment of immigrants for all oecd receiving countries, and thus can avoid the assumptions made in previous studies. “Emigration rates” can be produced by level of qualification and country of origin. The “emigration rate” for country i and education level l (“Emigration rate_{i, l}”) is calculated by dividing the expatriate population from the country of origin i and level of education l (Expatriates_{i, l}) by the total native-born population of the same country and level of education (Native Born_{i, l} = Expatriates_{i, l} + Resident Native born_{i, l}). Three levels of qualification are considered. Highly skilled persons correspond to those with a tertiary level of education (isced levels 5 or 6).

13A set of estimates of the Resident Native born_{i, l}, using a reference data base for the structure of education of the population 15+ in origin countries, has been compiled, from an updated version of Barro and Lee [1993] for the year 2000 which covers 113 countries (Barro and Lee 2000). [3]

14Before some basic results are presented, a number of qualifying remarks are in order. First of all, the term “expatriate” as used here means no more than that the person concerned was born in a country different from the one in which he/ she is residing. In practice, the departure from the country of birth may have occurred at any age and there may or may not be plans to return to the latter. The person may also have been born abroad as a citizen of the current country of residence, a phenomenon common in former colonial powers, such as France and Portugal. [4]

15The term “highly qualified” or “highly skilled” as used throughout this article means “holding a tertiary qualification”. The education in question may have been obtained in the country of origin, in the country of residence or in some other country. It may have been paid for by public funds in either the home country or the country of residence, by the emigrant him/herself and his/her family, by the institution attended or a combination of some or all of these. The expatriate stocks described in this article concern only those to oecd countries. Emigration to the rest of the world is not covered [5]. Finally, the data presented refer to the current stock of immigrants in oecd countries. They represent the cumulative results of decades of migration movements and may not be entirely representative of recent flows.

16As a result of these limitations, the number of persons born in a particular country, holding a tertiary qualification and living in an oecd country, constitutes only an approximate measure of the actual extent of brain drain (both past and current) in general and from developing countries in particular. However, by and large, it is likely the case that the picture which emerges from the dataset described above is a reasonably accurate one for most sending countries as far as expatriation to the oecd area is concerned. Even when sending countries have not defrayed the costs of expatriates’ tertiary education, the emigration of persons apt to complete tertiary schooling remains a loss to their economies.

First results on expatriates from OECD member countries and non-member countries

Expatriates Residing in oecd Member Countries

17Around the year 2000, there were over 18 million persons living in an oecd country who were born abroad and held a tertiary qualification. Of these over 10 million came from outside the oecd area. oecd countries themselves were the source of 40% of highly qualified foreign-born, the Asian continent about 30% and other continents or regions (Africa, non-oecd Europe, South and Central America and the Caribbean) between 5 and 10% each (Table 1).

Table 1

Percent of tertiary graduates among foreign-born persons 15+ in oecd countries by area/continent of birth, ca 2000

Place of birth Percent with tertiary Non-OECD Asia 39.0 Africa 25.2 OECD zone 21.9 OECD excl. MEX and TUR 29.9 Non-OECD Europe 21.0 South and Central America 21.0 Non-OECD Oceania 22.2 Caribbean 19.9 All countries 24.2

Percent of tertiary graduates among foreign-born persons 15+ in oecd countries by area/continent of birth, ca 2000

18The immigrant population in oecd countries on average has higher qualifications than the native-born population of these countries and this is the case whether one looks at immigrants from other oecd countries or from the rest of the world (Table 2). With a few exceptions (in particular, Finland, Germany and the Netherlands), this holds for individual oecd receiving countries as well, that is, their foreign-born populations tend to be more highly educated than the native-born. Curiously, however, it is in countries that are most associated in the public mind with emigration (Ireland, Mexico, Portugal, Turkey) and in the United Kingdom that the foreign-born have significantly higher qualifications than the native-born (more than 10 percentage points higher incidence).

Table 2

Percent of tertiary graduates among native- and foreign-born persons 15+ in oecd countries, ca 2000

All persons Native-born Foreign-born Total 20.6 20.2 24.2 AUS 39.7 38.6 42.9 AUT 11.0 10.9 11.3 BEL 22.8 22.9 21.6 CAN 32.9 31.5 38.0 CHE 19.4 18.1 23.7 CZE 10.3 10.2 12.8 DEU 18.9 19.5 15.5 DNK 18.8 18.8 19.5 ESP 19.5 19.4 21.8 FIN 23.3 23.4 18.9 FRA 17.0 16.9 18.1 GBR 21.6 20.1 34.8 GRC 13.6 13.4 15.3 HUN 11.0 10.7 19.8 IRL 24.7 22.7 41.0 ITA 8.3 8.1 12.2 JPN 27.4 27.4 29.9 KOR 26.7 26.7 32.2 LUX 16.0 12.8 21.7 MEX 11.0 10.9 37.8 NLD 19.3 19.5 17.6 NOR 22.9 22.4 30.6 NZL 28.0 27.2 31.0 POL 10.5 10.4 11.9 PRT 8.5 7.7 19.3 SVK 10.2 10.0 14.6 SWE 23.0 22.8 24.2 TUR 7.0 6.4 15.2 USA 26.8 26.9 25.9

Percent of tertiary graduates among native- and foreign-born persons 15+ in oecd countries, ca 2000

19The situation from an individual sending country perspective shows considerable variability. The percentage of the highly qualified among expatriates from sending countries ranges from the low (less than 10%) to the very high (more than 50%). Table 3 gives the percentage of expatriates with tertiary qualifications for the 20 countries having the highest percentages and the 20 having the lowest. The top seven countries or economies from the point of view of the qualifications of their expatriates are all from outside the oecd area and originate in Asia (Taiwan Province of China, India and Malaysia) and Africa (Nigeria, Egypt, Zambia and South Africa). The United States, Japan and Australia are the only oecd countries among the top twenty. All others are Asian or African countries.

Table 3

Countries with highest/lowest percentage of tertiary graduates among expatriates in oecd countries (minimum 25K expatriates 15+), ca 2000

Twenty highest Twenty lowest Percent with tertiary Country Percent with tertiary Country 62 Taiwan Province of China 14 Slovakia 57 Nigeria 13 Macedonia 54 India 13 Serbia-Montenegro 53 Egypt 13 Somalia 53 Malaysia 13 Italy 51 Zambia 13 Guinea-Bissau 51 South Africa 13 Mali 51 United States of America 13 Tonga 50 Japan 12 Samoa 48 Singapore 12 Dominican Republic 48 Lesotho 12 Bosnia 48 Philippines 10 Honduras 48 Namibia 10 American Samoa 47 Iran 9 Albania 47 Australia 8 Guatemala 46 Myanmar 8 El Salvador 45 Zimbabwe 7 Portugal 45 Mongolia 7 Turkey 45 Kuwait 6 Cape Verde 45 Libya 6 Mexico

Countries with highest/lowest percentage of tertiary graduates among expatriates in oecd countries (minimum 25K expatriates 15+), ca 2000

20At the other extreme are the sending countries which have less than 10% of persons with tertiary attainment among their expatriates. These include Mexico, Turkey and Portugal (oecd countries), Cape Verde, Guatemala and El Salvador, Albania. The bottom twenty list includes a number of other central and eastern European countries (Slovak Republic, Macedonia, Serbia and Montenegro, and Bosnia and Herzegovina) plus a few African (Mali, Guinea-Bissau and Somalia) and Oceanian countries (Samoa and Tonga), but also Italy (largely because of older migration waves).

21Highly qualified expatriates from sending countries tend to be numerically concentrated in few receiving countries (Table 4). Not surprisingly, these are the countries with the largest numbers of immigrants in general, which are the traditional settlement countries of the United States, Canada and Australia, plus France, Germany and the United Kingdom. In some cases, the main country of destination clearly reflects historical ties (France for Algeria, Portugal for Mozambique, the United Kingdom for Kenya) or geographic proximity (the United States for Mexico, Australia for Papua New Guinea, Finland for Sweden). The average sending country has about 90% of its expatriates concentrated in just four countries, with a minimum value of about 60%, which is what is observed for Denmark and Sweden. The sending countries whose qualified expatriates are least concentrated in this respect are, with the exception of the United States and Iraq, all European and include Bosnia and Herzegovina, Bulgaria, Romania and Serbia and Montenegro, as well as many eu countries.

Table 4

Concentration of highly qualified expatriates in oecd destination countries, ca 2000

Number of receiving countries One Two Three Four All Average percentage of highly qualified expatriates found in : 58 76 85 90 100 Number of sending countries with > 75 % of highly qualified expatriates in : 53 129 174 204 224

Concentration of highly qualified expatriates in oecd destination countries, ca 2000

Net Gainers from International Migration Among oecd Countries

22Within the oecd area, only the United States, Australia, Canada, Switzerland, Spain, Sweden, Luxembourg and Norway (in this order) are net beneficiaries of highly skilled migration from other oecd countries (Chart 1). The United Kingdom has 700,000 more highly skilled expatriates in oecd countries than it has highly skilled immigrants from other oecd countries. Comparable figures exceed 500,000 for Germany, 400,000 for Mexico and 300,000 for Poland. France and Belgium have almost as many highly skilled immigrants from, as expatriates, to oecd countries. This of course gives only a partial picture of brain drain/brain exchange, because it does not include movements of the highly skilled between non-oecd and oecd countries. When movements from all countries to the oecd are included, the picture changes significantly.

Chart 1

Net migration stock, population aged 15+ with tertiary education within oecd, ca 2000 (Thousands)

Net migration stock, population aged 15+ with tertiary education within oecd, ca 2000 (Thousands)

23With the notable exceptions of some central and eastern European countries as well as Mexico, Ireland, [6] Korea and Finland, highly skilled immigration towards oecd countries from the rest of the world systematically exceeds highly skilled emigration from oecd countries to other oecd countries (Chart 2). [7] On this measure (and provided that expatriation of the highly skilled to non-oecd countries can be assumed to be relatively uncommon), most oecd countries would seem to benefit from the international mobility of the highly skilled.

Chart 2

Immigrant and emigrant population aged 15+ with tertiary education in oecd countries, ca 2000 (Thousands)

Immigrant and emigrant population aged 15+ with tertiary education in oecd countries, ca 2000 (Thousands)

24The difference between the number of highly skilled emigrants to oecd countries and highly skilled immigrants from all countries is largely positive in the United States (+ 8.2 million), Canada and Australia, but also in France and Germany, even though these countries have a significant number of highly skilled expatriates in other oecd countries. Highly skilled immigration expressed as a percentage of the total highly skilled workforce is particularly significant (over 20%) in Australia, Luxembourg, Switzerland, Canada and New Zealand. The percentage of the highly skilled who are expatriates is below 10% for most oecd countries and particularly low in Japan, the United States, Spain and Australia. Conversely, more than 10% of the highly skilled born in Switzerland, Portugal, Austria, or the United Kingdom are living in other oecd countries. This percentage is over 20% for three countries : Luxembourg (22.2%), Ireland (24.2%) and New Zealand (24.2%).

Chart 3

Emigration rate of highly skilled 15+ and emigration rate of population 15+, ca 2000 (Percentages)

Emigration rate of highly skilled 15+ and emigration rate of population 15+, ca 2000 (Percentages)

25Moreover, there is a strong selectivity of migration in favour of the highly skilled. For almost all countries reviewed, the “emigration rate” of the highly skilled exceeds that of persons aged 15 and over as a whole (Chart 3). This phenomenon is the result of pull factors attributable to selective migration policies in receiving countries, but also to other factors such as the fact that highly qualified persons are more tuned into the international labour market (because of social capital, language skills, access to information…) and have more resources to finance a move.

Highly Skilled Migration from Non-Member Countries Towards oecd Countries : New Evidence on the “Brain Drain”

26Among non-member countries, the biggest expatriate community (15+) is that originating in the former USSR with 4.3 million people, followed by the former Yugoslavia (2.5 million), India (2 million), the Philippines (1.9 million), China (1.9 million), Vietnam (1.5 million), Morocco (1.5 million) and Puerto Rico (1.3 million). [8] Among persons with tertiary education, the former ussr still ranks first (1.3 million) with India having the second largest expatriate community (1 million), and the Philippines (0.9 million), China (0.7 million) and Vietnam (about 350,000) next in line.

27To estimate “emigration rates” by level of qualification for non-member countries, information on the level of education of the relevant population in the country of origin is required. The results are presented in Table 5 for the 20 countries with the lowest “emigration rates” for the highly qualified aged 15 and over, as well as for the 20 countries with the highest rates. Most oecd countries, some of which figure among the bottom twenty in Table 5, would tend to fall among countries having lower rates.

Table 5

Emigration rates of the highly qualified to oecd countries, ca 2000

Top 20 Bottom 20 Guyana 76.9 United States of America 0.4 Jamaica 72.6 Japan 1.2 Guinea-Bissau 70.3 Brazil 1.5 Haiti 68.0 Thailand 1.5 Trinidad and Tobago 66.1 Indonesia 1.5 Mozambique 52.3 Paraguay 1.8 Mauritius 50.3 Argentina 1.8 Barbados 47.1 Australia 2.4 Fiji 42.9 Spain 2.4 Gambia 42.3 Myanmar 2.5 Sierra Leone 32.4 China 2.6 Ghana 31.7 Peru 2.9 Kenya 27.8 Turkey 3.0 Cyprus 26.0 Canada 3.0 Hong Kong SAR 25.3 Bangladesh 3.0 Uganda 24.9 Nepal 3.2 Congo 24.9 Bolivia 3.2 Liberia 24.4 India 3.4 Ireland 22.6 Egypt 3.6 Sri Lanka 20.4 Venezuela 3.7

Emigration rates of the highly qualified to oecd countries, ca 2000

(Highly qualified expatriates as a percentage of all native-born tertiary graduates)

28Among countries with low “emigration rates” of highly qualified persons (i.e. inferior to 5%), are found most of the large countries included in the database (i.e. Brazil, Indonesia, Bangladesh, India and China) and in particular, a number of countries among those whose expatriates are the most highly qualified, such as Egypt and India. At the other end of the spectrum, smaller countries, a number of which are islands such as Jamaica, Haiti, Trinidad and Tobago, Barbados, Mauritius or Fiji, have more than 40% of their highly skilled populations abroad and sometimes as much as 70%.

29The world map (see Map 1) presents “emigration rates” of the highly skilled for all countries, with African countries standing out as those having particularly high “emigration rates”. Anglophone African countries as well as Portuguese-speaking countries (e.g. Mozambique and Angola, but also Cape Verde) record the highest brain drain rates. Emigration of the highly skilled is also quite significant in Central America but more moderate in Asia, with the relative exceptions of Hong Kong, China and Singapore. The former ussr faces intensive migration from former soviet republics towards Russia, which unfortunately it is not possible to illustrate here. [9] However, emigration of the highly skilled from countries of the former ussr, considered as a whole, towards oecd countries remains moderate relative to the total stock of qualified persons in these countries.

Map 1

Percentage of expatriates to oecd countries among all highly skilled born in the country

Percentage of expatriates to oecd countries among all highly skilled born in the country

Overview of the recent theoretical literature on brain gain in the light of the new statistical evidence

30This section does not intend to settle the question of the impact of the international mobility of the highly skilled on origin countries but rather to recall the argument, put forward in the recent literature on brain gain, namely that international migration increases the availability of human capital in origin countries, and to consider this argument in the light of the new data collected and presented here.

31The literature on the economic impact of the international mobility of the highly skilled has its roots in the 1960s when important migration flows of scientists and medical professionals from Europe to the United States were observed. [10] The term of “brain drain” itself was apparently used for the first time in this context in 1963 in a publication of the British Royal Society but the debate was also very much alive in the United States during this period (Cervantes and Guellec [2002]). [11]

32Although views concerning the brain drain in the 70s and 80s were initially mixed, [12] following the seminal work of Bhagwati on brain drain and taxation (e.g. Bhagwati and Hamada [1974], Bhagwati and Partington [1976], Bhagwati [1976], Bhagwati and Wilson [1989]), the possible detrimental effects of the international mobility of the highly skilled on less developed countries tended to be emphasized. [13] Recent studies, which focused on increasing returns to scale associated with human capital and on the long-term impact of highly skilled migration, have come to similar conclusions (Miyagiwa [1991], Haque and Kim [1995], Reichlin and Rustichini [1998], Wong and Yip [1999]). In the late 90s, however, alternative views re-surfaced. Three types of mechanism have been advanced to explain how brain drain can be transformed into brain gain.

33The first type of models concentrate on return migration, which can, under certain circumstances such as imperfect information on the level of skills (Starck, Helmerstein and Prskawetz [1997]) [14] or associated technology transfers (Dos Santos and Postel-Vinay [2003]), [15] be sufficient to induce a positive effect of highly skilled expatriation. These effects depend however, on incentives to return to the country of origin, on who actually returns and on the opportunities available to make the best use of newly acquired skills.

34In practice, migration movements which start as temporary do not always remain so (Martin [2001]), since some people overstay and others change their status to a permanent one. Many oecd countries have recently made policy changes to increase the possibility for some categories of temporary migrants to extent the validity period of their initial residence permit, to renew it or to shift to a permanent one (oecd [2004]). Most countries, for example, now allow students to remain after the completion of their studies to take up or seek jobs. The United States has progressively loosened the so-called “dual intent constraint” which prevented temporary migrants from asking for a permanent “green card” while on United States territory. The other settlement countries (Australia, Canada, New Zealand) tend to value more and more the work experience acquired in their country in their point systems, bridging de facto their temporary and permanent programmes.

35Some kinds of return migration, however, do occur and eventually benefit the country of origin. India, Chinese Taipei or to a less extent China [16] are generally mentioned in this context, but there is less evidence of this kind from less developed or smaller countries, which happen to be those experiencing higher expatriation of their highly skilled citizens. In the oecd context, Korea and Ireland have clearly benefited from highly skilled return migration, but this was observed in both cases after intensive investment in education and sustained economic growth. [17] This echoes the recent literature on fdi which emphasises the role of human capital and infrastructures in attracting foreign investments (Mouhoub [1998]).

36In the second set of models, which make up the bulk of the “revisionist theory on brain drain”, brain gain arises because of the fact that education choices are made under uncertainty with regard to migration prospects (e.g. Mountford [1997], Starck, Helmerstein and Prskawetz [1998], Reichling [2001], Beine, Docquier and Rapoport [2001], Lundborg [2004]). Agents are heterogeneous with regard to skills but all qualified workers face the same probability of emigration. Educational costs are generally decreasing with ability and returns to education are assumed to be higher abroad. [18] Under these hypotheses, without international migration, the education investment in the country of origin tends to be lower than if migration is allowed and under certain parametric conditions (which tend to be quite restrictive) brain gain may occur (i.e. the stock of highly qualified people remaining in the source country is higher than if the probability of emigrating were nil). Beine et al. [2002] synthesise this result by identifying the cases when the incentive effect (brain effect) dominates the detrimental effect (drain effect).

37Commander, Kangasniemi and Winters [2004] however, have pointed out that if receiving countries are selecting migrants according to their skills (i.e. the probability of emigration depends on skill) there would be no special incentive for people at the lower end of the skill ladder to invest in education and thus the brain effect would vanish. Main receiving countries, indeed, are more and more selective vis-à-vis immigrants. Settlement countries (Australia, Canada and New Zealand) have adapted their point systems on the basis of empirical research which has identified the socio-economic determinants of long-term integration in receiving countries. Some of them also carry out direct evaluation of language proficiency and motivation interviews. Wages of job offers in receiving countries (France, Germany) or prior wages (United Kingdom) are now used to identify highly skilled workers. Some countries (e.g. the United States) also have specific entry programmes for outstanding persons, identified on the basis of their academic record. [19]

38The correlation across countries between tertiary enrolment rates (a proxy for investment in tertiary education) and expatriation rates of the highly skilled should provide a simple, although imperfect, assessment of this theory. Table 6 indicates indeed a significant and negative correlation between these two variables, which means that countries where there appear to be lower incentives to invest in tertiary education also face higher emigration rates. In other words, emigration opportunities are not sufficient to influence the economic opportunities in the source country, either because emigration rates are too low or because the brain effect is not significant.

Table 6

Gross enrolment rate in tertiary education and Emigration rate of the highly skilled

Coefficient Standard Err. T stat Intercept 18.20 2.30 7.92 Expatriation rate of the highly skilled 0.22 0.07 3.20 Number of observations 88.00 Source : Unesco, our calculations.

Gross enrolment rate in tertiary education and Emigration rate of the highly skilled

39Another group of models posits a positive effect of the international migration of the highly skilled by assuming that homogenous workforce and human capital accumulation is subject to increasing returns to scale (Vidal [1998], Stark [2004]). Indeed, there would seem to be more arguments for assuming that emigration of the highly skilled reduces the supply of teachers increases the cost of training and alters the quality of education (i.e. emigration affects negatively the capacity to accumulate human capital).

40One of the shortcomings of the recent economic literature on brain gain comes from its implicit assumption that demand for tertiary education is mainly limited by lack of incentives to train. Indeed, in most developing countries, as pointed out by Commander et al. [2004], the cost of education is born by governments, partly because the social return is higher than the private one. As a matter of fact, international mobility of the highly skilled may offer an argument for liberalising the tertiary education system (World Bank [2000]). Indian IT schools and nursing schools in the Philippines certainly provide straightforward examples, although most people in developing countries face strong liquidity constraints and imperfect credit markets which would prevent them from taking advantage of these new opportunities.

41To a large extent the terms of the recent debate are the same as those of the 60s and 70s. [20] Some researchers emphasise the potential loss of talent and the risks of jeopardising long-term economic growth, whereas others argue that there is little to worry about and that, at least in the long run, the international mobility of the highly skilled will enhance global welfare and even foster economic convergence. The lack of reliable and comprehensive data, but also the complexity of the phenomenon which concerns, inter alia, return migration, technology transfer, as well as remittances, [21] has made an empirical assessment of the impact of expatriation a difficult one. The availability of new data will at least assist researchers who wish to test empirically the relevance and the robustness of the theoretical models presented above, if not to reduce the analytical complexity of the task.

Covariates of highly skilled expatriation

42The database described above allows an assessment of the importance of certain covariates in “explaining” variations in emigration rates of highly qualified persons across countries. Even a cursory scrutiny of the data cannot help picking up the fact that large countries tend to have low expatriation rates and island countries high rates. Already this suggests that the importance of the availability of opportunities in the educational system and labour market of a country for the expatriation decision, availability that can be limited by geographic isolation but enhanced by the sheer size of the educational system or labour market. Countries such as China and India may have large numbers of highly qualified native-born persons abroad, but they account for only a relatively small fraction of the tertiary stock in those countries. Conversely, persons from small or island states (the two tend to be correlated) may need to expatriate for education reasons and the move abroad for educational reasons may often be a stepping stone to eventual settlement (see Tremblay [2001]). In some cases certain small countries (Nepal or Paraguay) show low expatriation rates, but this may reflect more the fact that the database does not cover emigration to countries such as India and Argentina, which may be more “natural” destinations for persons from these countries than oecd countries.

43Another significant covariate of expatriation is the proportion of persons with tertiary qualifications (even after controlling for the gdp per capita) ; the lower this proportion, the higher the emigration rate of the highly qualified. This again tends to reinforce the notion that expatriation may often be associated with more limited opportunities in sending countries. In short, if there are few opportunities for persons with high-level qualifications, few persons invest in such education and many of those that do may tend to leave or not return (if educated abroad). This, however, begs the question of why there are fewer opportunities in the first place, a more complex issue.

44These three covariates of emigration rates explain approximately 40% of the variation in such rates across countries (Table 7) and would appear to be highly robust in the presence of other possible covariates of emigration rates. The fact of having English as an official language [22] is also a significant covariate, illustrating the importance of English-speaking countries, in particular the United States, as significant destination countries [23]. On the other hand, neither gdp per capita (averaged over 20 years) [24], trade openness (also averaged over twenty years) [25], unemployment (also averaged over twenty years) or continent indicator variables (except the Caribbean which specificity would need to be further analysed) show much explanatory power or have much association with the incidence of tertiary graduates from sending countries living in oecd countries.

Table 7

Main covariates for the expatriation rate of the highly skilled

All countries Non oecd countries (Eq1) (Eq2) (Eq3) (Eq4) (Eq5) (Eq6) Intercept 77.1*** 70.6*** 69.7*** 72.3*** 49.7*** 84.8*** (13.2) [5.9] (14.6) [4.8] (14.1) [4.9] (17.6) [4.1] (11.7) [4.3] (20.7) [4.1] Log (Population) 8.7*** 7.6*** 8.1*** 8.5*** 6.2*** 9.6*** (1.9) [4.6] (2.0) [3.8] (1.9) [4.2] (2.2) [3.8] (1.5) [4.2] (2.9) [3.3] % highly skilled in 15+ 0.51*** 1.0*** (0.13) [3.9] (0.35) [2.9] % highly skilled in 15+ controled for gdpi 0.47*** 0.52*** 0.53*** 0.49*** (0.18) [2.6] (0.18) [2.9] (0.18) [2.9] (0.12) [4.1] Island 8.5*** 8.7*** 4.8 (4.0) [2.1] (4.1) [2.1] (4.2) [1.1] Carribbean 17.4*** 23.2*** 23.6*** 27.2*** 30.9*** 27.6*** (6.5) [2.7] (7.0) [3.3] (6.7) [3.5] (6.1) [4.5] (4.2) [7.3] (8.4) [3.3] English speaking country 7.6*** 9.0*** 13.7*** 4.9*** (3.1) [2.5] (2.9) [3.1] (2.2) [6.3] (4.3) [1.1] gdp per capita ppaii 0.0019*** 0.001 7.1E-04 0.001 3.4E-03 (8.7E-04) [2.2] (9.1E-04)[1.1] (8.9E-04)[0.8] (6.7E-4) [1.5] (2.5E-03)[1.3] Square gdp per capita ppaii 7.2 E-08 2.9E-08 1.5E-08 6.3E-8** 2.7E-07 (4.4E-08) [1.6] (4.6E-08) [0.6] (4.5E-08) [0.3] (3.3E-08)[1.9] (2.5E-07)[1.1] Trade opennessiii 0.01 (0.07)[0.21] Unemployment rateiii 0.3 (0.2) [1.6] Political stabilityv 0.9 (2.5)[0.34] Adjusted R2 43.3 49.6 53.1 52.2 74.5 47.3 Number of observations 100 74 74 74 62 50 Source : For data on expatriation rate and the percentage of highly skilled we use our data based on Barro and Lee [2003]. Data on Population, gdp per capita ppa, unemployement and Trade openness are from World Development indicators 2004, our calculations. Data on governance are from Kaufmann D., A. Kraay and M. Mastruzzi [2005]. Note : Standard Error is indicated in parenthesis and t-Stat in brackets. i) residual of the regression of the percentage of highly skilled 15+ on gdp and its square, ii) gdp is the average when data are available over the last 20 years and iii) trade openness defined by (Import + Exports) as a percentage of gdp is gdp is the average when data are available over the last 20 years, iv) unemployment rate is the average unemployment over the last 20 years, v) the political stability indicator is for 1996.

Main covariates for the expatriation rate of the highly skilled

Conclusions

45Results presented in this paper based on the new database on immigrants and expatriates in oecd countries, show that :

46• The stock figures shown here reflect migration waves over a long period. Although recent migration to oecd countries tends to come largely from non-oecd countries, migration between oecd countries continues to have a significant impact. This migration is quite selective towards highly skilled migrants, underlining the effects of the current competition between oecd countries to attract “the best and the brightest” from other countries, both inside and outside the oecd area. It is also quite concentrated in a limited number of receiving country.

47• In most oecd countries, the number of immigrants with tertiary education exceeds the number of highly qualified expatriates to other oecd countries. On this measure, most oecd countries would appear to benefit from the international mobility of the highly skilled. This conclusion, however, must be considered as tentative, because the database described here does not cover expatriates to oecd non-member countries.

48• Among non-member countries the impact of the international mobility of the highly skilled is diverse. The largest developing countries seem not be significantly affected and indeed may benefit from indirect effects associated with this mobility (return migration, technology transfers, remittances…). At the other end of the spectrum, some of the smallest countries, especially in the Caribbean and in Africa, face significant “emigration rates” of their elites. Further analysis is needed to better understand the determinants, the dynamics and the impact of the international mobility of the highly skilled on these countries.

49If receiving countries and migrants are generally believed to profit from the opening up of borders to international migration of highly skilled human capital, the impact on sending countries is not so clear. For instance, some observers have claimed that the increase in the expected return on human capital as a result of expatriation increases incentives to invest in human capital in sending countries and that this increase is sufficient to off set the depletion effect of emigration on human resources in these countries. The new evidence on the international mobility of the highly skilled presented in this paper seem not to fully support this theory.

50Nonetheless, the potential negative impact of emigration on the supply of human capital needs to be seen in the context of the employment situation in the origin country (the extent of participation and unemployment, the productivity of human capital). In many cases, expatriated professionals would have had few opportunities to work at home in their field. This is, at least, what seems to result from the simple analysis of the covariates of the emigration of the highly skilled, presented above.