The views expressed in this report do not reflect those of the World Bank or its Board. Shaping the Future A Long-Term Perspective of People and Job Mobility for the Middle East and North Africa Background Papers TABLE OF CONTENTS Migration from MENA to OCED Countries: Trends, Determinants, and Prospects By Flore Gubert and Christophe J. Nordman Prospects for Management of Migration between Europe and the Middle East and North Africa By Johannes Koettl Labor Migration in the Middle East and North Africa: A View from the Region By Georges Corm Migration from Egypt, Morocco, and Tunisia: Synthesis of Three Case Studies By Christophe Schramm PREFACE This volume contains four background papers commissioned from various authors by the World Bank as the initial step following the decision to embark, jointly with the European Commission, on a study of migration in the Middle East and North Africa (MENA) Region. The papers thus constitute singly the building blocks and, together, the foundation of the main study contained in the main volume. The four papers have many commonalities. They deal—in different depth and breadth—with the main migration parameters (stocks, flows, migrant characteristics) and their determinants: (i) the role of education and skills; (ii) the economic impact of migration, especially remittance flows; (iii) the prospects for offshoring and outsourcing; and (iv) the role of migration institutions, policies, and management. The first two papers—by Gubert and Nordman, and Koettl—focus on labor migration from the MENA Region to member countries of the Organisation for Economic Co-operation and Development (OECD) (Europe especially) and the convergence of interest between the two. These papers draw heavily on population and labor force projections to ask “Can migrants from the MENA Region help filling labor market gaps in OECD countries?” (Gubert and Nordman) and to point out good migration policy practices for Europe (Koettl). If the first two papers deal largely with the demand side, the last two address the supply (MENA) side. These papers are strong on institutions and point to good migration policy practices as well as shortcomings among migrant sending countries. The first of these two papers (Corm) looks at the region as a whole, whereas the second (Schramm) is a synthesis of three country case studies (the Arab Republic of Egypt, Morocco, Tunisia) carried out in 2006 by Tahar Abdessalem, Ahmed Basti, Mohamed Khachani, Fouzi Mourj, and Ayman Zohry, and the International Organization for Migration (IOM) country branches of Tunis and Cairo. The caveats about the data noted in the Introduction to the main volume apply here, too, and are reiterated in the individual background papers. The papers that follow are based on the best and most up-to-date data available as of the time of their preparation. World Bank staff commented extensively on the papers in this volume, but the views expressed in them are the authors’ and do not necessarily represent those of the Bank. Migration from MENA to OECD Countries: Trends, Determinants, and Prospects Flore Gubert and Christophe J. Nordman ABBREVIATIONS AND ACRONYMS ANPE Agence nationale pour l'Emploi DARES Direction de l’Animation de la Recherche, des Etudes et des Statistiques DL Dumont and Lemaître DM Docquier and Marfouk DGTPE Direction Générale du Trésor et de la Politique Économique EECA Eastern Europe and Central Asia EU European Union FAP Nomenclature des familles professionnelles GCC Gulf Cooperation Council GDP Gross Domestic Product ILO International Labor Organization ISCED International Standard Classification of Education JVR Job Vacancy Rate LDCs Least Developed Countries NACE General Industrial Classification of Economic Activities MENA Middle East and North Africa MPI Migration Policy Institute OECD Organisation for Economic Co-operation and Development OLDC Other Least Developed Countries OLS Ordinary least squares PPP Purchasing power parity UAE United Arab Emirates UNEDIC Union Nationale pour l'Emploi dans l'industrie et le Commerce TABLE OF CONTENTS INTRODUCTION AND SUMMARY............................................................................................................ 1 I. OVERVIEW OF REGIONAL MIGRATION TRENDS .................................................................... 4 1. Migration from MENA to OECD Countries: A Host Countries’ Perspective...........................4 2. Migration from MENA to OECD Countries: A Sending Countries’ Perspective .....................9 3. Country Facts ...........................................................................................................................12 II. GRAVITY MODEL ANALYSES OF MIGRATION TRENDS FROM MENA TO OECD COUNTRIES .......................................................................................................................................... 17 1. A Review of Migration Theories and Determinants ................................................................17 2. Data Collection ........................................................................................................................20 3. The Determinants of Migration: A Gravity Approach.............................................................22 3.1. Cross-Sectional Data Analysis ..........................................................................................23 3.2 Panel Data Analysis...........................................................................................................29 III. CAN MIGRANTS FROM MENA COUNTRIES BE A SOLUTION TO LABOR SHORTAGES IN OECD COUNTRIES ............................................................................................................ 33 1. Labor Shortages in OECD Countries.......................................................................................33 1.1. Overview ...........................................................................................................................33 1.2. Recruitment Difficulties and Skill Shortages in France ....................................................38 1.3 Domestic Labor Market Reforms or Increased Migration?...............................................41 2. Can Migrants from the MENA Region Help Filling Labor Market Gaps in OECD Countries? ...............................................................................................................................42 2.1 Demographic Prospects.....................................................................................................42 2.2 Europe’s Skills Requirements and MENA’s Skill Supply: Is There a Match? .................44 STATISTICAL ANNEX ...........................................................................................................................54 Annex A. Descriptive Statistics......................................................................................................54 Annex B. Estimation Results..........................................................................................................77 REFERENCES ........................................................................................................................................84 TABLES Table 1 Emigration from Algeria since the 19th Century ............................................................12 Table 2 Major Foreign Communities in the GCC States, 2002 ..................................................14 Table 3 The Determinants of Migration Using Gravity Models .................................................32 Table 4 The 20 Occupations with the Strongest Labor Shortages in France (July 04-July 05) ..40 Table 5 Unemployment Rates by MENA Country in Recent Years...........................................43 Table 6 Immigrants from Maghreb and Mashreq Countries by Type of Activity in Spain, 2001 ...............................................................................................................45 Table 7 Business Environment in the MENA Region ................................................................51 FIGURES Figure 1 Distribution of Migrants by Region of Origin and Sub-Group of OECD Countries .......5 Figure 2 Distribution of Migrants from the MENA Region by Sub-Region of Origin and Sub-Group of OECD Countries, 2000 ............................................................................6 Figure 3 Share of Highly Educated Migrants from MENA Countires in OECD countries ...........7 Figure 4a Share of Migrants from MENA Countries in the Migratory Flows to Some Selected Continental European Countries, 1995-2002 ..................................................................8 Figure 4b Share of Migrants from MENA Countries in the Migratory Flows to Some Selected Northern European Countries, 1995-2002 ......................................................................8 Figure 4c Share of Migrants from MENA Countries in the Migratory Flows to Some Selected Anglo-Saxon Countries, 1995-2002................................................................................8 Figure 5 Expatriation Rates to OECD Countries of Population Aged 25 and Over, By MENA Country .........................................................................................................9 Figure 6 Distribution of Migrants from the MENA Region by Sub-Region of Destination and Sub-Group of MENA Countries, 2000 .........................................................................10 Figure 7 Highly Educated Expatriation Rate by MENA Country, 2000 ......................................11 Figure 8a Composition of Migrant Stocks by Level of Education and Sending Country, 1990....11 Figure 8b Composition of Migrant Stocks by Level of Education and Sending Country, 2000....12 Figure 9a Number of Job Vacancies in Some European Countries, 2001-2004 ............................34 Figure 9b Number of Occupied Jobs in Some European Countries, 2001-2004............................34 Figure 10 Job Vacancy Rate in Some European Countries, 2001-2004.........................................35 Figure 11 Annual Changes in Job Vacancy Rate, 2002-2004 ........................................................35 Figure 12 Number of Job Vacancies by Sector in 2005 .................................................................36 Figure 13 Number of Job Vacancies by Sector in 2005 .................................................................38 Figure 15 Unemployment Numbers in Algeria by Sector, 2000-2004...........................................46 Figure 16 Unemployment Numbers in Egypt by Sector, 2000-2004 .............................................48 Figure 17 Unemployment Numbers in Morocco by Sector, 2000-2004 ........................................48 Figure 18 Unemployment Numbers in Turkey by Sector, 2000-2004 ...........................................49 Figure 19 Approximate Value of Offshore Services in Countries that Supply Them ....................50 Figure 20a Wages in the Manufacturing Sector for a List of Selected Countries ............................51 Figure 20b Wages in the “Real Estate, Renting and Business Activities” Sector for a List Of Selected Countries ...................................................................................................52 Figure 20c Wages in the “Financial Intermediation” Sector for a List of Selected Countries.........52 BOX Box 1 Nurse Shortages in OECD Countries .............................................................................37 INTRODUCTION AND SUMMARY Migration from the Least Developed Countries (LDCs) to member countries of the Organisation of Economic Co-operation and Development (OECD) has been and continues to be one of the most controversial subjects of concern over the past few years because of the increasing perceptions among governments and political observers in the latter countries that (i) migration from LDCs to more developed countries has to be regulated on a selective basis; and (ii) the aging population process in developed countries entails demographic transitions and potential economic upheavals in their labor markets. Hence, it is expected not only that more jobs will become available in the near future, but also that medium- and high-skilled workers will be sorely lacking. The rising deficit of medium- and high-skilled workers is often presented as a European challenge of increasing importance. As Constant and Zimmermann note, “This is a matter of size and intensity. Even in the long-term, it will be difficult for European firms to hire the appropriate quantities on their local labor markets. Supply is not likely to keep pace with demand. A permanent effort will be needed to participate in the rising world market for flexible high-skilled workers. This international effort is a prerequisite for keeping the own talents and the hired migrants of the European Union member countries. Appropriate policy instruments have to be found to enable companies to deal with this challenge. 1 ” Little is known about the specific economic, demographic, and labor market determinants of migration from certain regions of the developing world. In this report, we shed light on the trends, determinants, and prospects of migration from the Middle East and North Africa (MENA) Region to OECD countries. 2 We exclude intraregional migration from the analysis even though it has important economic and demographic impact on both host and origin economies. The report is in three parts. Part I provides a picture of the levels and trends of migration from the MENA Region to OECD countries. The overall discussion is based on two complementary databases made available only recently on the stocks of international migrants in OECD countries: that of Docquier and Marfouk (2005) and that of Dumont and Lemaître (2005). Based on the 2000 round of censuses held in each OECD country, these two databases provide a detailed, comparable, and reliable picture of immigrant populations within OECD countries. Based on statistical analyses using these data sets, our findings are as follows. First, migrants from the MENA Region represent a small share of the migrant population in most OECD countries. Second, their skill composition strongly varies between receiving countries because of 1 Constant and Zimmermann 2005, p. 1. 2 According to the World Bank’s country classification, the MENA Region includes 19 countries: Algeria, Bahrain, Djibouti, the Arab Republic of Egypt, the Islamic Republic of Iran, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Occupied Palestinian Territories, Oman, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, the United Arab Emirates (UAE), and the Republic of Yemen. All MENA countries except Bahrain, Kuwait, Qatar, Saudi Arabia, and the UAE are low- or middle-income countries. -2- differing migration and labor market policies. Migrants from the MENA Region are much more educated on average in Anglo-Saxon destination countries than in the traditional destination countries of Continental Europe. Given the skill composition of the migrants, “brain drain” appears to be nonnegligible in Algeria, the Islamic Republic of Iran, Lebanon, Morocco, and Tunisia although estimates strongly vary. Furthermore, for all MENA countries, the overall level of education of the migrants increased between 1990 and 2000. Third, destination also strongly varies with origin. Emigration from the Maghreb countries is strongly concentrated toward Continental Europe, while that from the other MENA countries is focused on the group of Anglo-Saxon countries. Part II provides econometric analyses of the determinants of migration from MENA to the OECD countries. In particular, the influence of economic, demographic, and political factors on the size and composition of migration flows from MENA to OECD countries over the period 1990–2002 is investigated. We first provide a brief survey of the economic literature on international migration and discuss the role of potential key determinants of migration (“push” and “pull” factors 3 ), such as economic and demographic pressures, network effects, or the so-called welfare magnet effect. Two complementary econometric analyses are then carried out. First, a cross-sectional data analysis builds on Docquier and Marfouk’s database on international migration, which provides data on the stocks of MENA immigrants in all OECD countries by education attainment in 1990 and 2000. Second, a panel data approach makes use of annual migration flows made available online by the Migration Policy Institute (MPI). The two alternative data sets are then combined with economic and noneconomic data on both the origin and the destination countries to estimate gravity models. In its most basic form, the gravity model explains migration flows (or stocks) from one country to another by each country’s economic characteristics and bilateral geographic characteristics. In our approach, we examine the impact of other determinants of migratory flows or stocks, such as demand factors in the host country, and their potential pull effects. We focus on the following questions: How much do the “pure” economic factors like differences in income and unemployment levels, returns to education, or labor productivity growth explain migration behavior? How much is explained by other factors such as immigration policies, social networks, cultural and linguistic distance, threat to own freedom such as the level of political rights, and civil liberties in home country? The emerging findings emphasize the relevance of pull factors to explain the magnitude of expatriation rates from developing to OECD countries despite restrictive immigration policies in most destination countries. Next to economic factors (income levels, returns to education, labor productivity growth), demographic determinants in particular appear to be strong predictors of migration flows. These findings are consistent with the premise that migration is partly demand- driven and used by some countries to compensate for population aging. The results support the idea that different forces are at work concerning the migration of high-educated and low- educated individuals. Push and pull factors have a notably different impact according to the type 3 A push factor is a feature or event that encourages a person to leave his or her country (high unemployment, poverty, famine, drought, natural disasters, political oppression or persecution, and so on). A pull factor is a feature or event that attracts a person to move to another country. -3- of migration considered. This should be taken into account by governments that wish to implement selective immigration policies. Part III of the report examines whether migration from MENA countries can be a solution to labor shortages in OECD countries. We first present a general overview of job vacancies in some European countries. We then focus on one OECD country, France, and present a more detailed picture of the recruitment difficulties and skill shortages faced by this country. Despite methodological problems, all the available data confirm that labor markets are tight in several OECD countries. Labor shortages are experienced in all sectors, particularly in the Wholesale and Retail Trade, Hotels and Restaurants, and Transport and Communication subsectors. The number of job vacancies is also high in the Public Administration Education and Health sector, partly due to the nurse shortages that affect most OECD countries. In some European countries, namely France and Germany, substantial unsatisfied labor requirements are observed despite persistently high rates of unemployment. In these countries, labor shortages could be attenuated through domestic reforms aimed at mobilizing the unused local labor supply. However, such reforms are unlikely to have an immediate impact on the population’s education, occupation, or location choices. Under these conditions, turning to labor migration programs could be a more rapid and effective means of addressing shortages. We then investigate whether the MENA Region can fill labor gaps in OECD countries. Compared with Eastern European and former Socialist countries, MENA economies have the demographic potential for large-scale emigration to OECD countries. From a strict quantitative standpoint, the diverging demographic trends and structural differences between most European countries and MENA economies thus add credence to the idea that potential synergies need to be developed between the two regions. In particular, increased labor mobility from the MENA Region could compensate for demographic trends in European labor markets in the next two decades, while constituting a response to the lack of employment in the home countries. Evidence from the estimated gravity models (see part II) does suggest that migration flows from one country to another are strongly driven by demographic features. The central question of interest is not so much whether the MENA region can provide high numbers of working-age individuals but, rather, what kind of migrants the region can provide. We thus scrutinize whether there would be a match between Europe’s skill requirements and MENA’s skill supply by first looking at the current labor market situation of migrants originating from the MENA Region in OECD countries. Then, in a more prospective approach, we examine available data on unemployment rates by sector within MENA countries. High unemployment numbers in some sectors are likely to be due to skill mismatch or to lack of available jobs in these sectors. If the latter is true, migration might be an appropriate means for job seekers to find a suitable job in the same sector abroad. As noted, the analysis confirms that increased labor mobility from the MENA Region could compensate for labor shortages in European labor markets in the coming decades, while responding to the lack of employment in some home countries. This finding needs to be taken with caution, however, because of the lack of more disaggregated data on both labor shortages in OECD countries and labor surpluses in MENA countries as well as on skill requirements. -4- Finally, we explore the question of whether offshoring could gain ground in the MENA Region. By comparing wages in some MENA countries with those prevailing in countries where offshoring has been gaining ground for the last 10 years, we find that some MENA countries offer significant cost advantages. However, offshoring business process or services is not simply cost driven. In some other regards, MENA countries are disadvantaged compared with other countries. So far, the region has not taken advantage of the opportunities for the outsourcing of business service jobs, offered mainly by U.S. or British firms. However, given the conditions prevailing in some of the French-speaking MENA countries (namely, Morocco and Tunisia), these countries could appear in the future offshoring plans of French and other continental countries. I. OVERVIEW OF REGIONAL MIGRATION TRENDS This section provides a picture of the levels and trends of migration from the MENA Region to OECD countries. All MENA countries except Bahrain, Kuwait, Qatar, Saudi Arabia, and the UAE are low- or middle- income countries. The overall discussion is based on two databases made available only recently on the stocks of international migrants in OECD countries: that of Docquier and Marfouk (2005) and that of Dumont and Lemaître (2005). Based on the 2000 round of censuses held in each OECD country, these two databases provide a detailed, comparable, and reliable picture of immigrant populations within OECD countries. 4 The two databases differ in the way they define a migrant. Docquier and Marfouk (2005) count as migrants all working-age (25 and over) foreign-born individuals living in an OECD country, where foreign born is defined as an individual born abroad with foreign citizenship at birth. By contrast, Dumont and Lemaître (2005) count as migrants foreign-born individuals age 15 and over living in an OECD country. Given this definition, people born, say, with French nationality outside of France (in Algeria, for example) are counted as foreign born. Another difference between the two databases lies in the reference year: while Docquier and Marfouk’s database covers both 1990 and 2000, Dumont and Lemaître’s database delivers information on migrant stocks in 2000 only. In what follows, we use one database or the other alternatively, depending on the question at hand. Migration from MENA to OECD Countries: A Host Countries’ Perspective Stocks The two complementary databases of Docquier and Marfouk and Dumont and Lemaître provide a rich picture of emigration from the MENA Region, the salient features of which are as follows. First, the share of migrants from MENA countries in the total stock of migrants is rather low in most OECD countries (see annex A, table A1). This stock is higher than 10 percent in only eight OECD countries, all of them being located in the European Economic Area (Belgium, Denmark, France, Italy, Netherlands, Spain, Sweden, and Norway). In North America, the share of migrants from MENA countries is still lower: 5.3 percent in Canada, and 2.6 percent in the 4 The picture provided, however, focuses on recorded migrants only. -5- United States. At a more aggregated level, the same pattern emerges. In figure I.1, OECD countries are classified into one of five groups according to their welfare state regime and their migration policy general orientation. Following Esping-Andersen (1990), the group of Continental European countries consists of Austria, Belgium, France, Luxemburg, the Netherlands, and Switzerland; the group of Northern European countries consists of Denmark, Finland, Norway, and Sweden; the group of southern European countries includes Spain, Portugal, Italy, and Greece; and the group of Anglo-Saxon countries consists of Australia, Canada, Ireland, New Zealand, the United Kingdom, and the United States, plus Japan. The Other category includes Republic of Korea, Mexico, and Turkey. Former socialist countries (Hungary, Poland, the Czech Republic, and the Slovak Republic) and Germany are excluded here because of the unavailability of data on the country of origin of a large share of their migrants. As shown by figure I.1, the MENA region ranks second in Continental European countries behind the group of high-income countries, but only third and fourth in Northern and Southern European countries, respectively. In the two other groups of OECD countries, migrants from the MENA region account for a marginal share of the migrant population. Figure I.1 Distribution of Migrants by Region of Origin and Subgroup of OECD Countries, 2000 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Continental Europe Northern Europe Southern Europe Anglo-Saxon Other countries High income countries Europe and Central Asia Latin America and the Caribbean Asia MENA Sub-Saharan Africa Unknow n Source: See annex A, table A1. Note: The High-Income Countries category includes both OECD and non-OECD countries according to the World Bank’s classification. However, one needs to be cautious with regard to migration statistics. Quoting CARIM’s (the Euro-Mediterranean Consortium for Applied Research on International Migration) Mediterranean Migration 2005 Report, “With no exception in the case of Med-MENA countries, migrants counted by country of origin are in excess compared with those counted by host countries. In the European Union, there is an aggregated difference of 2.5 million migrants for five Med-MENA countries of origin for which calculation is possible (Algeria, Egypt, Morocco, Tunisia -6- and Turkey). 5 This figure reflects a variety of situations, including dual citizens, former migrants who have now left the country, and irregular migrants.” 6 See annex A, tables A2 and A3 for detailed figures on Algeria and Egypt. Second, destination strongly varies with origin (see figure I.2). In Continental and Southern Europe, respectively, 90 percent and 77 percent of migrants from the MENA Region come from a Maghreb country, while this share is less than 10 percent in Northern Europe and in the group of Anglo-Saxon countries. The sudden rapid growth of North African immigration to Southern European countries such as Italy and Spain is the result of a number of factors among which the interruption of immigration in the countries that traditionally received North Africans. 7 Quoting Giubilaro, “Initially, the absence of legal requirements defining conditions of entry and residence in the Southern European countries, together with the manpower requirements of certain sectors, facilitated this movement, which could subsequently be kept going by means of migration networks. However, under pressure from their European partners, Italy and Spain have adopted a legal system in respect of immigration, which is comparable to that of the rest of Europe. With this new situation, the flows would appear to be stabilizing.” In the group of Anglo-Saxon countries, as in the group of former socialist countries, more than 40 percent of migrants from the MENA region come from the Mashreq. In Northern Europe and in the Republic Korea, Mexico, and Turkey, nearly two-thirds of the migrants from the MENA Region are either Iraqis or Iranians. (A more detailed picture of the distribution of migrants from the MENA Region by country of origin is provided in annex A, tables A4, A5, and A6 for some selected OECD countries.) Figure I.2 Distribution of Migrants from the MENA Region by Subregion of Origin and Subgroup of OECD Countries, 2000 100% 80% 60% 40% 20% 0% Continental Europe Northern Europe Southern Europe Anglo-Saxon Eastern Europe Other countries GCC States Maghreb Mashreq Other MENA Sources: Dumont and Lemaître 2005, authors’ calculations. 5 In all CARIM’s publications, Turkey is included in MENA, unlike in the standard classifications followed in World Bank publications. 6 CARIM 2005, p.8 7 Labib 1996, cited in Giubilaro 1997. -7- Third, the share of highly educated migrants from MENA countries in the total stock of highly educated migrants is rather low in most OECD countries (see figure I.3 and annex A, table A7, column 4). Migrants from MENA countries appear to be less educated on average than other migrants in seven out of the eight aforementioned countries in which they represent more than 10 percent of the total stock of migrants (Belgium, Denmark, France, Italy, the Netherlands, Spain, Sweden, and Norway). In these countries, indeed, the share of highly educated migrants from MENA countries in the total stock of highly educated migrants is lower than the share of migrants from MENA countries in the total stock of migrants. In Belgium, for example, individuals born in MENA countries represent 14.6 percent of the foreign born, but among them, the highly educated represent only 10.1 percent of the highly educated foreign born. Fourth, among migrants from MENA countries, the share of the highly educated strongly varies by country of destination and reflects differences in migration policy regimes (see figure I.3 and annex A, table A7, column 3). It is rather low in most countries of the European Economic Area except for the former Socialist countries (Hungary, Poland, the Slovak Republic, and the Czech Republic) and the Anglo-Saxon countries (Ireland and the United Kingdom). It is rather high in North America (Canada and the United States). Thus, countries where migrants from the MENA Region account for a marginal share of the migrant population are also those where migrants from the MENA region are the most educated. Figure I.3 Share of Highly Educated Migrants from MENA Countries in OECD Countries, 2000 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Continental Northern Europe Southern Europe Anglo-Saxon Eastern Europe Other Europe countries % of highly qualified migrants in the total stock of migrants from MENA countries % of highly qualified migrants from MENA countries in the total stock of highly qualified migrants Source: See annex A, table A7. Note: See figure I.1. Flows Even though data on migratory flows are scant, the available evidence suggests that the share of migrants from MENA countries in the migratory flows to OECD countries has been rather low and rather stable for the last 10 years, except for a few European countries—that is, Sweden, the Netherlands, and Norway (see figure I.4 and annex A, table A8). -8- Figure I.4 Share of Migrants from MENA Countries in the Migratory Flows to Selected Countries Panel A. To Selected Continental European Countries, 1995–2002 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 1995 1996 1997 1998 1999 2000 2001 2002 Austria Germany Netherlands Source: See annex A, table A8. Note: Countries were selected on the basis of data availability. Panel B. To Selected Northern European Countries, 1995–2002 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 1995 1996 1997 1998 1999 2000 2001 2002 Denmark Finland Norw ay Sw eden Source: See annex A, table A8. Note: Countries were selected on the basis of data availability. Panel C. To Selected Anglo-Saxon Countries, 1995–2002 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 1995 1996 1997 1998 1999 2000 2001 2002 Australia United Kingdom United States Source: See annex A, table A8. Note: Countries were selected on the basis of data availability. -9- Migration from MENA to OECD Countries: A Sending Countries’ Perspective General Stylized Facts Adopting the sending countries’ perspective, the four following features emerge from the data. First, as shown by figure 1.5, expatriation rates from MENA to OECD countries are generally rather low, either because emigration flows are directed to non-OECD countries (Egypt is one such example, with migration flows mostly directed to Arab oil countries) or because emigration flows are rather low on the whole (this is particularly the case of labor-importing countries such as Libya, Oman, Saudi Arabia, the UAE, and Republic of Yemen). 8 Exceptions include the Maghreb countries that have strong historical links to Europe and Lebanon. Figure I.5 Expatriation Rates to OECD Countries of Population Age 25 and Over, by MENA Country Morocco Mashreq Maghreb Tunisia Algeria Lebanon Jordan Syria Egypt Kuw ait GCC States Bahrain Qatar United Arab Emirates Saudi Arabia Oman Occupied Palestinian Territory Other MENA Iraq countries Iran Libya Djibouti Yemen 0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0% 16.0% Source: Docquier and Marfouk 2005. Note: GCC = Gulf Cooperation Council; MENA = Middle East and North Africa. Second, emigration from the Maghreb to OECD countries is strongly concentrated toward Continental Europe, while emigration from the other MENA countries is focused on Anglo- Saxon countries (see figure I.6). This second stylized fact suggests that past colonial links and common language are strong pull factors, as further developed in part III. 8 Expatriation rate from country of origin i to OECD countries is calculated by dividing the expatriate population age 25 and over from that country by the native-born population age 25 and over of the same country. - 10 - Figure I.6 Distribution of Migrants from the MENA Region by Subregion of Destination and Subgroup of MENA Countries, 2000 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Maghreb Mashreq GCC States Other MENA Other OECD Eastern Europe Anglo-Saxon countries Northern Europe Southern Europe Continental Europe Source: Docquier and Marfouk 2005, authors’ computations. Note: GCC = Gulf Cooperation Council; MENA = Middle East and North Africa; OECD = Organisation for Economic Co-operation and Development. Third, available data on the size of the brain drain by MENA country exhibit stronger disparities. In figure I.7, brain drain is measured by the highly educated expatriation rate, that is, the number of individuals with tertiary education (13 years and above) born in country j and living in an OECD country, divided by the total number of individuals with tertiary education born in country j. 9 Three sets of estimates have been reported, based on different data sources. As suggested by the figures, emigration of highly educated workers particularly affects the Islamic Republic of Iran, Lebanon, and the Maghreb countries. In contrast, the brain drain is rather low in the Gulf States. Compared with other developing regions belonging to the middle-income group, based on the World Bank classification, the size of the brain drain is higher in the MENA Region (10.5 percent) than in Latin America (7.5 percent), East Asia and the Pacific (6.1 percent), and Eastern Europe and Central Asia (3.9 percent). 10 Fourth, for all MENA countries, the share of the highly educated in the total stock of migrants has increased since 1990 (see figure I.8). This evolution is due to both the adoption of selective migration policies biased toward the highly-skilled in receiving countries and to the increasing proportion of educated individuals in the sending ones. 9 In what follows, the diagnosis on the size of the brain drain by country of origin is biased, because several major destinations such as the Persian Gulf countries are ignored. In particular, given the importance of migration flows from Mashreq to Gulf countries, the size of the brain drain in these countries of origin is likely to be much higher. 10 Because of the significant size of their brain drain, very small countries and islands were removed from the different subgroups (Guyana and Suriname, for example, were dropped). - 11 - Figure I.7 Highly Educated Expatriation Rate by MENA Country, 2000 Tunisia Maghreb Morocco Algeria Syria Mashreq Lebanon Jordan Egypt Saudi Arabia States GCC Oman Yemen Occupied Palestinian Territory Other MENA Libya Iraq Iran Djibouti 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% 40.0% 45.0% Estimate 1 Estimate 2 Estimate 3 Sources: Dumont and Lemaître 2005; Docquier and Marfouk 2005. Note: GCC = Gulf Cooperation Council; MENA = Middle East and North Africa. Estimate 1 is the highly educated expatriation rate according to the Cohen and Soto database (population 15+), while estimate 2 is the highly educated expatriation rate according to the Barro and Lee database (population 15+). Both estimates come from Dumont and Lemaître (2005). Estimate 3 comes from Docquier and Marfouk (2005) (population 25+). Figure I.8 Composition of Migrant Stocks by Level of Education and Sending Country Panel A. 1990 Mashreq Maghreb Algeria Morocco Tunisia Egypt Jordan Lebanon Syria Bahrain GCC States Kuwait Oman Qatar Saudi Arabia United Arab Emirates Djibouti Other MENA Iran Iraq Libya Occupied Palestinian Territory Yemen 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Low Medium High Sources: Docquier and Marfouk 2005, authors’ calculations. Note: GCC = Gulf Cooperation Council; MENA = Middle East and North Africa. - 12 - Panel B. 2000 Mashreq Maghreb Algeria Morocco Tunisia Egypt Jordan Lebanon Syria Bahrain 2000 GCC States Kuwait Oman Qatar Saudi Arabia United Arab Emirates Djibouti Other MENA Iran Iraq Libya Occupied Palestinian Territory Yemen 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Low Medium High Sources: Docquier and Marfouk 2005, authors’ calculations. Note: GCC = Gulf Cooperation Council; MENA = Middle East and North Africa. Country Facts Maghreb (see annex A, tables A9–A14) Algeria Emigration from Algeria is concentrated toward France: 84.2 percent of Algerians residing in an OECD country live in France, 11 despite a diversification of migration flows from Algeria in the last 30 years (see table I.1). Table I.1. Emigration from Algeria since the Nineteenth Century Period Main Destination Countries From 1830 Middle East Beginning of the twentieth France, Arab countries century 1945–1962 France, Morocco, Tunisia 1962–1973 France 1974–1991 France, East Germany, Western Europe, North America 1992– France, Western Europe, North America, Gulf States, Australia, and East Asia Source: Table extracted from CARIM’s Mediterranean Migration 2005 Report. In terms of qualifications, the composition of Algerian migrant stocks is highly varied between Europe and North America: respectively, 84.1 percent and 72.8 percent of Algerian migrants are highly educated in Canada and the United States, but only 10 percent in Europe’s main destination countries (Belgium, France, and Spain). Two factors explain this diversity. First, the 11 Docquier and Marfouk 2005. - 13 - earlier the migrating cohort, the less educated it tends to be. Second, migration and labor-market policies are generally biased toward the highly educated in North America. However, compared with other MENA countries (Egypt, Jordan, Libya, and so on), the expatriation rate of the highly educated is rather high, with estimates varying from 9.4 percent to 18.0 percent (see figure I.7). Because migrant stocks are particularly large in France, even though the share of highly educated Algerian migrants is low, France has drained many highly educated Algerians: on average, one can estimate that about five to six highly educated Algerians out of 100 migrantsreside in France. Morocco Moroccan nationals are predominantly found in France (38.8 percent according to Docquier and Marfouk (DM) figures and 45.8 percent from Dumont and Lemaître’s (DL), followed by Spain (DM: 19.8 percent; DL: 18.2 percent), the Netherlands (DM: 13.5 percent), and Italy (DM: 9.9 percent). Compared with the Tunisians and Algerians, the Moroccans are widely distributed over all European countries. This characteristic, already manifest in the seventies, has been further accentuated in recent years. In France, Moroccans are largely low-educated individuals with only primary education (79 percent DM), whereas this share is 66 percent in Spain where individuals with secondary level education represent 28.6 percent of the total stock of Moroccan migrants (compared with 7.8 percent and 16 percent for France and the Netherlands, respectively). However, the same contrast as for Algeria emerges between Europe and North America: Moroccan migrants are highly educated in Canada (72.2 percent) and the United States (64.2 percent). The expatriation rate of the highly educated is rather high with regard to other MENA countries (on average 9.9 percent, according to DM), with estimates varying from 17 percent to 19.5 percent. Tunisia Emigration from Tunisia is mainly directed to France (DM: 69.9 percent; DL: 78 percent) followed by Italy (DM: 12.2 percent; DL: 9 percent) and Germany (DM: 4.9 percent). In terms of qualification, the stock of Tunisian nationals residing in the traditional European host countries is predominantly made of individuals with only primary education (DM: 80 percent in France; 78.1 percent in Italy). The highly educated population’s expatriation rate ranges from 12.5 percent to 21.4 percent, which is also quite high, compared with other MENA countries. Mashreq (see annex A, tables A15–A22) Egypt - 14 - Emigration from Egypt is mainly directed to the Gulf area (see table I.2). Meanwhile, since the early 1960s, some Egyptians have migrated permanently to the Australia, Canada, the United States, and Western European countries (France, Italy, and the United Kingdom). The preferred destination is the United States, where about two out of five permanent Egyptian migrants are found, followed by Australia and Canada, where about one-fourth of permanent Egyptian migrants live. Table I.2. Major Foreign Communities in the GCC States, 2002 (in thousands) United Country of Saudi Bahrain Kuwait Oman Qatar Arab Total origin Arabia Emirates India 100 295 300 100 1,400 1,000 3,200 Pakistan 50 100 70 70 1,000 450 1,740 Egypt, Arab 275 15 35 1,000 130 1,455 Rep. of 1,000 35 1,035 Yemen, Rep. of 160 110 450 100 820 Bangladesh 160 35 350 160 705 Sri Lanka 60 50 500 120 730 Philippines 50 50 270 110 480 Jordan/Palestine 95 170 265 Syrian Arab 45 80 20 40 145 Rep. 250 250 Iran, Islamic 250 250 Rep. of 120 120 Indonesia 100 100 Sudan 280 1,475 630 420 7,000 2,488 Kuwait Turkey All Source: Kapiszewski 2004, p. 125. Note: GCC = Gulf Cooperation Council. In terms of qualifications, most Egyptians migrants residing in North America are highly educated. The pattern is different in the main European destination countries (France, Italy, and the United Kingdom); nearly a third of Egyptian migrants there have only primary education. Brain drain is less than 5 percent. However, it is likely that the estimated size of the brain drain for Egypt would be much higher if the number of highly skilled Egyptian expatriates living in Persian Gulf countries were included in the calculation. Jordan Among OECD countries, Jordan migrants are predominantly found in the United States (66.4 percent), followed by Germany (9.5 percent), Canada (5.5 percent), and Australia (4.1 percent). In terms of qualifications, most Jordan migrants to North America and Australia have completed tertiary education. This pattern is different in Germany and the United Kingdom, where more than one-third of Jordanian migrants have only primary education. - 15 - On the whole, brain drain is rather limited, with estimates varying between 3.3 percent and 7.2 percent. As with Egypt, however, the size of the brain drain would be much higher, were skilled labor exports to Gulf Cooperation Council (GCC) states accounted for. Lebanon Among OECD countries, Lebanese migrants are predominantly found in North America (Canada and the United States) and Australia, followed by some Western European countries (France, Germany, and Sweden). The preferred destination is the United States, where about one-third of Lebanese migrants live. In terms of qualifications, the share of low-educated Lebanese migrants is rather low in the United States (11.4 percent), but rather high elsewhere (between 30 and 42 percent). Brain drain is very high. It is estimated that about 4 out of 10 highly educated Lebanese reside in an OECD country. Syria Syrian nationals are mostly found in the United States (DM: 40.7 percent), then in Germany (11.4 percent), Canada (10.9 percent), Sweden (9.3 percent), and France (8 percent). In 2000, Syrian nationals in these countries had remarkably similar characteristics. Differences appear in terms of qualification structures across the five dominant countries of Syrian emigration: the share of poorly educated migrants is significantly lower in the United States (16.8 percent) than in Germany (35.7 percent), Canada (29.5 percent), Sweden (40 percent), or France (32.4 percent). Hence, the share of highly educated Syrian migrants is higher in the United States (52 percent) but also in Canada (57 percent). Brain drain indicators show that Syrian highly educated emigration is rather low, ranging from only 4.4 percent to 6.1 percent. GCC States Labor-importing GCC States do not send many migrants to OECD countries. Rather, they have been the main recipients of migrants from Egypt, the Islamic Republic of Iran, Jordan, Syria, and the Republic of Yemen moving to the Gulf primarily to take up employment. An overview of migrant stocks from these countries in OECD countries, computed from Docquier and Marfouk’s database, is provided in annex A, table A23. Other MENA Countries (see annex A, tables A24–A33) Djibouti Few migrants from Djibouti reside in OECD countries (DM: 1 638 individuals in 2000). Islamic Republic of Iran - 16 - Emigration from the Islamic Republic of Iran to OECD countries is mainly directed to North America and some Western European countries (marginally France, the Netherlands, Sweden, and the United Kingdom). The preferred destination is the United States, where about one out of two permanent Iranian migrants live, followed by Germany and Canada, where about one-fifth of permanent Iranian migrants are found. In terms of qualifications, three out of four Iranian migrants residing in North America are highly educated. The pattern is different in Western Europe. Brain drain is rather high, with estimates varying from 8 percent to 18 percent. Iraq Emigration from Iraq to OECD countries is directed to the United States and some Western European countries (Germany, Sweden, the Netherlands, and the United Kingdom). In terms of qualifications, no clear pattern emerges. The share of highly educated Iraqi migrants is slightly higher in Australia, Canada, and the United States than in other destination countries. Brain drain is rather high (6 percent to 11 percent). Libya Libyan migrants are found almost equally in the United States (DM: 33.2 percent) and in the United Kingdom (DM: 32.4 percent). Their shares drop to 7 percent, 6 percent, and 5.9 percent in Germany, Australia, and Canada, respectively. In the United States, the proportion of educated Libyan migrants is overwhelming (80.5 percent) as compared with that of the United Kingdom (42.7 percent) or Germany (27.5 percent). Consequently, the share of poorly educated Libyan migrants is low in the United States (2.1 percent) compared with the other seven main countries of emigration (on average 32 percent for Australia, Canada, France, Germany, Greece, and Switzerland); Brain drain is low at only 2.4 percent. However, this is consistent with the rather low expatriation rate of the Libyan population age 15 and over (1.8 percent, with an average of 2.8 percent for MENA countries). Occupied Palestinian Territory Within OECD, nearly 8 emigrants from the Palestinian territory out of 10 are found in North America, mainly in the United States. In terms of education, Palestinian migrants in North America are more educated on average than those in Europe. The Republic of Yemen - 17 - As with Egypt, emigration from the Republic of Yemen is mainly directed to the Gulf area (see table I.2). Yemenite nationals residing in OECD countries (20,949 in 2000) are predominantly found in the United States (DM: 55.4 percent) and then in the United Kingdom (28.3 percent). The remaining migrants are divided among Germany (4.3 percent), Canada (3.3 percent), France (2.8 percent), and Australia (1.4 percent). The structure of qualification of the Yemenite migrants is almost similar across the United States and the United Kingdom. Additionally, the shares of highly and poorly educated individuals is almost equal in these two main host countries (30 percent, against 31 percent in the United States, and 35 percent, against 38 percent in the United Kingdom). Brain drain is 6 percent (according to DM). II. GRAVITY MODEL ANALYSES OF MIGRATION TRENDS FROM MENA TO OECD COUNTRIES This section empirically investigates the influence of economic, demographic, and political factors on the size and composition of migration flows from MENA to OECD countries in recent years. The focus is on both the supply and demand determinants of immigration patterns. The following questions are addressed: What are the main driving factors of migration from MENA to OECD countries? Are these drivers mainly economic or demographic? Given destination countries’ restrictive immigration policies, do flows respond to economic incentives? Is there a “welfare magnet effect”? The section is structured as follows. The first part briefly surveys the economic literature on international migration and discusses the role of potential key determinants of migration. The second part describes the data sets. The third part analyzes, using gravity models, the determinants of bilateral migration flows from MENA to OECD countries. We first run cross- sectional regressions using as dependent variables expatriation rates by education group in 2000, computed from Docquier and Marfouk’s database. We then run panel data regressions using yearly data on immigrant flows to 14 OECD countries by country of origin between 1990 and 2002. A Review of Migration Theories and Determinants 12 In what follows, we look at the central elements of the theoretical issues in relation to international migration from LDC to industrial countries, as well as some special aspects that we wish to consider in our econometric analysis. Traditional Explanations In the neoclassical approach at a macro level, migration is a consequence of differentiated economic opportunities across regions or countries, especially earnings differentials. 13 Migration 12 This section is mainly based on Rotte and Vogler’s (1998) and Constant and Zimmermann’s (2005) literature review. - 18 - is viewed as the optimal allocation of labor into regions of highest productivity, leading to an equalization of wages if there are no migration costs. In the human capital approach to migration, Sjaastad (1962) gives up the assumption of homogeneous labor. Depending on their skill levels, individuals calculate the present discounted value of expected returns on their human capital in every region or country. Migration occurs if the returns, net of discounted migration costs, are larger in a region or country than those in the country of origin. Every individual evaluates the returns and costs in a different way so that migration may be worthwhile for some individuals in a country and not for others. As a result, in the analysis of migration patterns, one should not only pay attention to aggregate labor market variables (such as wage and unemployment difference), but also consider the importance of the heterogeneity of individuals. Nearly all empirical studies using aggregated data find a statistically significant positive effect of income or wages in the destination country, or of the income and wage differential between the sending and the receiving countries, and a negative effect of income and wages in the sending country. Evidence on the influence of unemployment rates both in the sending and in the receiving countries is more varied. Some studies conform to the predictions of the Harris-Todaro model and find a positive influence of the probability of employment in the destination country on the migration decision. 14 Other studies find that individuals are attracted to regions with shortage of jobs. In our empirical investigation, difference in economic opportunities between origin and host countries, especially income and unemployment rates, are captured by the levels of gross domestic product (GDP) per capita in both the host and origin countries, as well as unemployment rates in the destination country. Labor Market Disequilibrium The labor market disequilibrium issue has to deal with the question of whether and how the native population is affected by immigrants. 15 The evaluation crucially depends on the conditions in the labor markets of the host countries. These markets may be either competitive or in disequilibrium—that is, labor supply equals labor demand or not, respectively. A further decisive aspect is that the labor force in the host country is likely to be heterogeneous and of diverse quality. Then, the immigration impact depends on whether immigrants are uneducated or educated. This is the reason why it is important to understand the determinants of emigration rates of different groups of individuals, namely, the educated and uneducated workers. Disequilibrium situations in labor markets may occur when there are institutional constraints in the market for unskilled labor, such as union wages, minimum wages, or transfers like social assistance, or when the education system is not able to provide the necessary supply of workers for the skilled labor markets. Another issue is whether educated and uneducated workers are complements or substitutes to the local population. A reasonable (and standard) assumption is 13 Hicks 1932. 14 Harris and Todaro 1970. 15 Borjas 1994; Bauer and Zimmermann 1997. - 19 - that high-educated and low-educated workers are complements, which implies that one group becomes more productive (and relatively scarce) when the input of the other group is increased. Self-Selection of Migrants Emigrants are not a random sample of a country’s population. This aspect has already played an important role in the brain-drain debate concerning emigration of the highly educated and its negative consequences for sending country’s economy. The question of self-selection of migrants gained renewed attention with analyses of the performance of migrants in the host labor market. 16 A popular model in the migration literature to analyze the self-selection of migrants is the Roy model. 17 The basic idea can be summarized as follows: When looking at who chooses to emigrate to the United States, one ready-made answer is that workers from low-wage countries will immigrate. This may be true on average, but it is probably too simple. The workers immigrating to the United States are probably not a random subset of the sending country’s workforce. Rather, one should expect that potential migrants make some rough comparison of their wages in the home country and their expected wages in the United States. On average, one would expect those who emigrate to have higher expected earnings in the United States than in the home country and vice versa for those who stay. Hence, the relative wage on observable and unobservable abilities in home and host country determines the kind of selection, while the level of income affects the volume of migration. In our econometric analysis, we shall introduce a proxy for the potential reward for observable and unobservable abilities in the host country, together with GDP per capita levels, to control for both effects. A natural belief is that we may observe differentiated effects of, for instance, the private rate of returns to schooling in the host country on the emigration rates of high-educated and low-educated workers. Push and Pull Factors of Migration In public discussion about present or future international migration from developing countries, the term “migration pressure” is often used. 18 Some authors define migration pressure as the difference in the numbers of people who are willing to migrate under current circumstances and those individuals the country of destination is prepared to accept. From the perspective of the sending country, Bruni and Venturini (1995) define migration pressure as an excess supply of labor regarding the demand of labor. However, the problem remains that excess labor supply does not necessarily result in emigration. Schaeffer (1993), therefore, defines migration pressure in terms of effective demand. The general consensus is that several factors in the sending and the destination countries determine whether an existing “migration potential” results in actual migration. For factors concerning the home country, one refers to the so-called “push” factors; for those concerning the host country, one uses the term “pull” factors. Important push factors for developing countries 16 See Borjas (1994) for an overview. 17 A formal representation of this model for the explanation of international migration flows can be found in Borjas (1987). 18 Rotte and Vogler 1998. - 20 - include population growth and the corresponding unemployment, poverty, and political instability. Pull factors are mainly high wages in the receiving countries, their social security systems, and political stability, as well as a potential past colonial relationship or common culture between host and origin countries. The generosity of the welfare state, therefore, may play an important role in migrants’ decision of choosing country of destination. This is the so called welfare magnet effect. To control for these push and pull effects as determinants of migratory flows, we will apply bilateral “cultural distance” indicators between the host and origin countries (colonial relationship, common language), as well as variables indicative of potential demographic pressure, such as population density, urban population growth, or the age structure of the populations. We also account for political and civil factors in the sending country, as well as use proxies for the welfare magnet effect in the host country. Most attention regarding persistent migration is directed to the so-called network effects. 19 Existing connections between individuals in a host country and friends and relatives left at home increase the likelihood of the latter following the former to the country of destination. In that case, network effects may counteract cultural distance, if the concerned ethnic group is already present in the destination country. Indeed, migrants’ links to their home country reduce the costs and uncertainty for further potential migrants. Moreover, these contacts in the destination country facilitate accommodation and job search for follow-up migrants, which also increases migration incentives. A few studies show that network variables always have a significant and positive impact on migratory flows. 20 From this brief literature review, we now turn to the following questions: How much do the “pure” economic factors, like differences in income and unemployment levels, returns to education, or labor productivity growth, explain migration behavior? How much is explained by other factors such as immigration policies, social networks, cultural and linguistic distance, threat to personal freedom (such as the level of political rights), and civil liberties in home country? Data Collection Cross-Sectional Analysis Our cross-sectional analysis builds on a new comprehensive and consistent database on international migration. 21 The data set (hereafter DM05) describes the stocks of immigrants in all OECD countries by education attainment in 1990 and 2000. 22 As Docquier and Marfouk argue, “regarding statistics, it has long been recognized that migration flow data are less reliable than stock data, due to the impossibility of evaluating emigration and return migration movements.” 23 Migrants are defined as all working-age (25 and over) foreign-born individuals living in an OECD country. Using the category of working-age population maximizes the comparability of the immigration population with data on education attainment in the source countries. The 19 Pedersen, Pytlikova, and Smith 2004. 20 Rotte and Vogler 1999; Pedersen, Pytlikova, and Smith 2004. 21 Docquier and Marfouk 2005. 22 A detailed presentation of the methodology for data collection and the assumptions can be found in Docquier and Marfouk (2005). 23 Docquier and Marfouk 2005, p. 7. - 21 - foreign-born concept (defined as an individual born abroad with foreign citizenship at birth) better captures the decision to emigrate, compared with citizenship used alone. Indeed, in some receiving countries, immigrants’ children often keep their foreign citizenship. 24 However, in a limited number of cases, immigrants are classified only by citizenship (Germany, Greece, Italy, Japan, and the Republic of Korea). 25 Three levels of schooling are distinguished. Low-educated workers are those with primary education (up to eight years of schooling); medium educated workers are those with secondary education (9 to 12 years of schooling); and high-educated workers are those with tertiary education (13 or more years of schooling). This definition of education categories is consistent with Barro and Lee’s data set on education across countries, which is based on the International Standard Classification of Education (ISCED). 26 The preceding literature review and discussions on the determinants of migration have shown that what really matters in the migration process are stocks instead of flows. Indeed, there are many reasons to believe that, with international migration, we are dealing with a disequilibrium situation, as there might be significant rationing. In this context, observed flows may be part of an adjustment to a desired stock of migration. Therefore, rather than estimating emigration flows over a 10-year period (by retrenching the stocks of immigrants in 1990 from the stocks of immigrants in 2000), we prefer to specify our model to explain the stocks of migrants at the end of the period and to include lagged values of our considered determinants of migration as right- hand-side variables. For our purpose, we thus compute expatriation rates by education level s for the year 2000 for each couple of receiving country i and sending country j in the following way: Stock s ij 2000 m s ij , 2000 = (II.1) Population sj 2000 After eliminating observations with missing values or outliers, the sample of bilateral expatriation rates for 2000 amounts to 4,650 couples of LDC-OECD countries, of which 570 are MENA-OECD couples. Other sending subregions include Sub-Saharan Africa (1,410 couples), South Asia (240), Latin America and the Caribbean (900), Europe and Central Asia (780), and East Asia and the Pacific (630). The list of sending and receiving countries is reported in annex B, table B1. Panel Data Approach Our panel data approach makes use of annual migration flows coming mainly from the MPI Web site, an “independent think-tank dedicated to the study of the movement of people worldwide.” 27 24 Docquier and Marfouk 2005. 25 Note that the data exclude a large number of students who temporarily emigrate to complete their education. 26 Barro and Lee 2000. The ISCED classification of education levels into three groups can be made as follows: the group of “primary school educated” includes the categories ISCED 0, 1, and 2; the group of secondary-school educated includes ISCED 3 and 4; and the group of tertiary educated includes ISCED 5 and 6. 27 Available at www.migrationpolicy.org. - 22 - To complete MPI’s database, we also use some of the data contained in the most recent OECD’s annual reports on Trends in International Migration or those made available on the Web sites of destination countries’ National Statistical Institutes. Our two alternative data sets on bilateral immigration stocks and flows are then combined with economic and noneconomic data on both the origin and destination country of each country couple to estimate gravity models. All the bilateral geographic data (distance and dummies for land border, colony, island, and common language) are taken from Glick and Rose (2002). This data set has been updated for many missing data on LDCs and OECD countries. Most data on macroeconomic variables such as GDP per capita (in purchasing power parity, PPP), country population, age-dependency ratios, population density, urban and total population growth, unemployment rates, and literacy rates come from the World Development Indicators and the Penn World Table, release 7.0. Information on origin and destination countries’ share of young population comes from the Population Division of the United Nations. Information on productivity growth comes from OECD’s annual Employment Outlook reports. Demographic factors include population density (PopDens) in destination and source countries, age-dependency ratio in destination and source countries AgeDepend), the share of the population ages 15 to 24 (1524Share) in destination and source countries, and the urban population growth (UrbPopGrowth) in destination and source countries. The higher the relative population growth in the source country and the larger the proportion of the population in the younger adult age-group, the larger the migration pressure is expected. We also include two variables (Literacy and MeanEduc) reflecting the education level of the source and destination countries. To complement the database for some countries, we collected unemployment rates from the OECD’s annual Employment Outlook reports for the years 1990–2000. In particular, for the year 1994, we use unemployment rates by education level following Barro and Lee’s classification (that is, regrouping the different ISCED’s schooling levels into the three main categories described above). Data on schooling participation by country are taken from Barro and Lee’s data set (2000). We also collected the estimated Mincerian private return to education specific to each OECD country from Psacharopoulos and Patrinos (2004). The few missing returns in their study for some European countries were replaced by estimates made available by the PURE Project of the European Commission on the returns to education in European countries. 28 Since political pressure in the source country may influence migration, we include two variables (PolRights and CivLib), which measure the level of political rights and civil liberties in source countries. Each country is assigned a rating for political rights and a rating for civil liberties based on a scale of 1 to 7, with 1 representing the highest degree of rights or freedom and 7 the 28 Asplund 2001. - 23 - lowest one. These two variables come from Freedom House’s Freedom in the World country ratings. Finally, we include a variable (PubExpComp) assumed to capture potential pull factors relating to the “welfare magnet” theories as presented by Borjas (1987, 1999). It is a measure of the level of public social expenditure devoted to unemployment compensation expressed as a percentage of GDP in destination countries. According to the welfare magnet theory, we expect higher migration flows into countries with relatively higher levels of public social expenditure. This variable stems from OECD’s annual Employment Outlook reports. Descriptive statistics for the data sets are given in annex B, table B2. The Determinants of Migration: A Gravity Approach The basic gravity model has been used several times to analyze gross trade flows between countries. 29 It assumes that those flows depend on the size of each of the two trading countries (measured as their GDP) and any determinant of bilateral trade cost, such as distance, contiguity, trade agreements, and so on. Only recently, the gravity approach has been used to explain bilateral migratory flows. 30 In its most basic form, the gravity equation explains the total emigration from one country of origin to a country of destination by using the economic characteristics (population and GDP per capita) of each of the two countries and the bilateral geographic characteristics (distance, common border, access to sea, common language). Occasionally, if the data allow it, additional characteristics of the countries of origin and destination are included to account for immigration policies and other relevant characteristics. In this respect, our empirical strategy follows the general guidelines described in the previous studies. In addition, we stress the impact of new determinants of migratory flows, such as demand factors in the host country, and their potential pull effects. Cross-Sectional Data Analysis Empirical models We run two types of regression using alternative specifications for the dependent variable (total expatriation rate, expatriation rate of the low educated, expatriation rate of the medium educated, and expatriation rate of the high educated). To analyze carefully the effect of bilateral characteristics (such as distance, having a common border or a common language, and so on) on expatriation rates by level of education, we first run a basic gravity equation in which all characteristics of the countries of origin and destination are controlled for through the inclusion of origin and destination dummies. The basic regression is as follows: m s ij , 2000 = c1 log(distance ij ) + c2 Borderij + c3 Language ij + c4Colony ij + c5 Island ij + α i + β j + ν ij (II.2) 29 See, for example, Feenstra, Markusen, and Rose 2001. 30 Rotte and Vogler 1999; Pedersen, Pytlikova, and Smith 2004; Mayda 2005; Peri 2005. - 24 - where msij is the expatriation rate from j to i of individuals belonging to the education group s; αi is a set of 28 country of destination fixed effects and βj is a set of country-of-origin fixed effects. The other variables are relative to each couple (i,j). The same regression is run separately for each skill group using two alternative samples: in the first sample (hereafter the OLDC sample, for Other Least Developed Countries), all countries of origin are included, except countries from the MENA Region (in this case, βj is a set of 127 country-of-origin fixed effects), while in the second sample (hereafter MENA), only countries from the MENA Region are included (in this case, βj is a set of 19 country-of-origin fixed effects). Country-specific effects enlighten us as to whether differentiated country impacts exist on the migration pattern. Naturally, these fixed effects are difficult to interpret, because they capture a wide range of country-specific effects, such as differences in income levels and economic incentives, employment opportunities, demographic pressure, and so on. To identify the impact of all these variables on the magnitude of expatriation rates, we run a second set of regressions in which we explicitly introduce GDP per capita in 1995 (in PPP) for the host (i) and the origin (j) countries, and new determinants of migration reflecting both the labor market and demographic structures of each country. The vector of regressors include the private return to education in i, unemployment rates by educational level in i, the labor productivity growth in i, the average years of schooling among the population of country i, the literacy rate in j, the population density in i and j, the share of population age 15–24 in i and j, the age dependency ratio in i and j, and the urban population growth in i and j. Finally, we account for institutional incentives to migrate by including the share of public expenditure related to unemployment compensation in i and two variables measuring the level of political rights and civil liberties in origin countries. In this type of specification, country fixed effects cannot be identified. However, regional dummies regrouping host and origin countries are added to the vector of regressors. These regional dummies are introduced to test whether migration within particular regions is more intense, once bilateral and country-specific characteristics are controlled for. The regression is as follows: m s ij , 2000 = c1 log(distanceij ) + c2 Borderij + c3 Languageij + c4 Colonyij + c5 Island ij + φi + χ j + δ 1 log(GDPpci ) + δ 2 log(GDPpc j ) + δ 3 PopDensi + δ 4 PopDens j + δ 51524Sharei + δ 6 1524Share j + δ 7 AgeDependi + δ 8 AgeDepend j (II.3) + δ 9UrbPopGrowthi + δ 10UrbPopGrowth j + δ 11 log(MeanEduc) i + δ 12 Literacy j + δ 13 ReturnsEduci + δ 14UnempPrimi + δ 15UnempSeconi + δ 16UnempTerti + δ 17 ProdGrowthi + δ 18 PubExpCompi + δ 19 PolRights j + δ 20 CivLib j + υ ij - 25 - where φi and χ j are respectively subregions of destination fixed effects and subregions of origin fixed effects. As for specification (II.1), the same regression is run separately for each skill group using the two alternative samples defined above. The models are estimated using ordinary least squares (OLS) with reported standard errors being robust to clustering, because pairs of countries are likely to be highly dependent over time. Empirical results Annex B, tables B3 and B4, report the estimation results of the two alternative gravity models presented above. In table B3, columns (1) and (2) report the estimated coefficients of specification (II.1), which is separately estimated for MENA-OECD and OLDC-OECD couples and for each education group. The other six specifications report the coefficients for the same regression run separately for each education group (primary, secondary, and tertiary levels), that is, allowing the explanatory variables to have different effects on the expatriation rate of each group. In this setting, 16 percent to 42 percent of the variation in expatriation rates across couples is explained by geographic and cultural proximity variables and country-specific effects. 31 In Table B3, columns (1) and (2) show that the bilateral variables are not always significant. While distance between two countries and past colonial relationship have the expected sign and enter significantly into the two regressions, sharing a border and speaking a common language are statistically significant in only one of the two specifications. Turning to the variable indicating whether there is an island (or two) in the couple, results show that it is never significant. Considering separate regressions for each education group, few interesting indications emerge. First, the two groups of intermediate and highly educated workers in the OLDC sample are sensitive (negatively) to the presence of a common border, although the less educated are not. By contrast, expatriation rates from the MENA region are not sensitive to the presence of a common border between host and origin country, which is not a surprising result given the limited number of such cases. Second, the linguistic commonality plays a role for only the most educated workers, whose jobs may involve professional skills for which mastering a language is important. Last, the effects of geographic proximity and historical links are significant in all education groups. For countries of the MENA region, however, the group of highly educated workers appears to be less sensitive to distance than the other two groups, suggesting that highly educated people are better equipped to seize good job opportunities in distant countries. Table B4 reports the results obtained for specification (II.2) which replaces the country fixed effects by country-specific effects. Regional dummies regrouping host and origin countries are also included. 31 The hypothesis of joint nullity of the country fixed effects is rejected at the usual confidence intervals, suggesting that the specific conditions in the origin and destination countries play a significant role in explaining expatriation rates. - 26 - A similar grouping as that proposed by Esping-Andersen (1990) was adopted to compute the regional dummies: the group of Northern European countries includes Denmark, Finland, Iceland, Norway, and Sweden; that of Anglo-Saxon countries includes Australia, Canada, Ireland, New Zealand, the United Kingdom, and the United States; that of Continental European countries includes Austria, Belgium, France, Germany, Luxembourg, the Netherlands, and Switzerland; and that of Southern European countries includes Greece, Italy, Portugal, and Spain. The interesting aspect of the regressions reported in Table B4 is that we added a wide range of lagged country-specific effects (referring either to 1990, 1995, or the period from 1990 to 2000), such as income and social indicators (GDP per capita, education or literacy), economic incentives (returns to education, labor productivity growth), employment opportunities (unemployment rates for each education level), various indicators of demographic pressure, and institutional incentives to migrate.32 Let us first focus on the explanatory power of the different models. Respectively, 31 percent and 54 percent of the variability in expatriation rates is explained in the OLDC and MENA samples (compare table B4, columns 1 and 2). At first sight, a noticeable result is that pull factors are often more significant than push factors despite immigration restrictions in most destination countries. This is the case with the impact of the GDP per capita in the host country, which is always highly significant and positive in all regressions, while the effect of the GDP per capita in the origin country is not significantly different from zero in almost all models. The only exception appears in column (4), where the dependent variable is the expatriation rate of low- educated workers originating from the MENA Region. In this regression, per capita GDP in the sending country operates as a push effect. This result might be consistent with the idea that lower levels of per capita GDP in the source country both increase incentives to leave and make it difficult to overcome poverty constraints. Expatriation rate is lower for the least educated individuals because they are too poor to afford the fixed costs of migration. Indeed, further regressions (not reported to save space) revealed that, in the case of the MENA Region, there is a significant negative sensitivity to distance in GDP per capita between host and origin countries for the less educated group. In other words, the more distant the countries in terms of income per capita, the lower the emigration rate of the least educated individuals. This is consistent with the idea that fixed migration costs constitute huge barriers to migration, especially for low-educated people originating from poor countries. These fixed out-of-pocket money costs, such as transportation costs and housing, are generally considered important in human capital migration models. 33 32 Due to missing data for many countries, we were forced to make trade-offs between accuracy of the proxy, data availability, and preservation of the size of our initial samples, as much as possible. Some variables were then chosen to the detriment of others to save on degrees of freedom (this is the case for the literacy rate instead of the mean education for country j). In table B4, many couples of countries present in previous regressions are nonetheless dropped from the estimation because of missing variables. Among OECD countries, note that the regressions are performed without the Czech Republic, Hungary, Japan, the Republic of Korea, Mexico, Poland, the Slovak Republic, and Turkey. For this reason, results in table B4 should not be compared with those presented in table B3. 33 Chiswick 2000. - 27 - All the variables introduced to account for demographic pressures in host and sending countries are found significant in almost all regressions. Here, again, a striking result is that demographic characteristics in the destination country generally have a much more significant impact on the size of expatriation rates than those in the origin country. In particular, the old-age dependency ratio (dependents to working-age population, calculated as the number of people age 65 or more, divided by the number of people ages 15–64) and the urban population growth in the origin country are found to play no role at all. Population density and the share of the population age 15–24 in the origin country are found to be significant, however, but not in all specifications. More specifically, in both OLDC and MENA samples, the expatriation rates of the most educated migrants are also the most reactive ones to the population density in the origin country (compare columns (7) and (8) versus columns (3) and (4). One interpretation of this finding is that the brain drain is responsive to demographic pressure in the origin country. Turning to the demographic characteristics of the destination country, population density, urban population growth, and the age dependency ratio are all found to significantly increase the size of the expatriation rate, while the share of the population age 15–24 is found to decrease it. Estimated coefficients for the population density variable are always larger using the OLDC sample but, in both cases, the effect of population density in the host country is stronger on expatriation of the highly educated. The same holds true for the effects of urban population growth and of the age dependency ratio. As expected, the share of population age 15–24 in the host country exerts a negative effect on expatriation rates. This result might be explained by the demand side of international migration: destination countries with a rising share of elderly (and a decreasing share of young individuals) in their population might be tempted to “import” population from other countries to compensate for population aging. 34 Conversely, the significant positive sign found on this variable in origin countries in the regression of expatriation rates of migrants originating from the OLDCs would seem to reflect the supply side of international migration, namely the push effect—or demographic pressure—exerted by the share of young people in the origin country. Another notable finding concerns the private returns to education in the destination country. This variable is indeed found to be highly significant in all regressions. The size of the estimated coefficients also indicates that the reward for education is an important determinant of expatriation rates across countries. Moreover, not surprisingly, the magnitude of the effect increases with the education category. Hence, this original finding is all the more accurate for the most educated workers for whom access to qualified and well-paid jobs is probably the most important motivation behind migration. This a key issue in the context of international transferability of human capital. 35 Indeed, human capital acquired at home may not be fully transferable to the host country. Then, it is expected that the lower the international transferability of human capital, the higher the earnings disadvantage of the migrants at the time of migration. However, our results suggest that careful comparison by the individuals of the returns to education across potential host countries is an important factor in the decision to migrate to one country instead of another. This is somehow a self-selection or self-correcting 34 Part III provides further discussion regarding the potential needs in migration and skill shortages that demographic trends may entail in OECD countries. 35 Constant and Zimmermann 2005. - 28 - behavior, that is, educated individuals make decisions related to the most appropriate transferability of their acquired human capital. In the case of the MENA Region, the impact of the return to education in the host country is sizeable for the least educated category (above that estimated using the OLDC sample), while it remains larger for the most educated group of migrants (but far below that estimated using the OLDC sample). Let us now turn to the impact of the unemployment rates by education category in the host country. Introducing these variables in the models is revealing because, according to Harris and Todaro’s (1970) theory, low (high) unemployment rate in the destination (source) country will cause higher immigration flows. Besides, we are able to disentangle this effect by education level, because we benefit from differentiated rates in host countries in 1994 across three education groups, which correspond to our three categories of migrants. 36 Interestingly, the traditional prediction mentioned above is not always supported by the data. For instance, the unemployment rate among low-educated workers in the host country exerts a positive impact on the expatriation rate of low-educated individuals originating from the MENA region. This result might be explained by the existence in the sample of host countries with both high unemployment rates and generous welfare allowances. Because they have restricted access to unemployment benefits and social assistance, immigrants have a lower reservation wage than natives and accept to take up low-skilled, low-status, and low-paid work. Many professions (construction, transport, and agriculture) have indeed become associated with immigrant or ethnic minority workers. At the same time, the unemployment rate among the intermediate educated individuals in the host country is found to decrease the expatriation rate of individuals originating from the MENA Region, whether or not they are educated. This finding is well explained by standard theory, namely, the fact that a high unemployment rate in the destination country will cause lower immigration flows. Finally, the unemployment rate among the highly educated individuals in the host country is found to increase the expatriation rates of individuals from both MENA and OLDC regions. This last finding is somewhat puzzling, but it is robust and consistent across regressions and sending regions. 37 A first explanation would be that the migration decision of highly educated individuals in LDCs is based on intermediary job opportunities in the host country rather than on high-level job opportunities, given the lack of transferability of human capital internationally. However, this does not totally explain the positive correlation between the unemployment rate among the highly educated in the host country and the expatriation rates. Another argument relates to the persistence of welfare magnet effects mentioned above, that is, host countries with high unemployment rates are also those with generous welfare states. The welfare magnet effect would then more than counteract the disincentives to migration induced by poor job opportunities in skilled jobs in the host country. This puzzling result deserves further investigation and should be the object of future research. 36 Regrettably, we did not find such disaggregated information for the sending countries. 37 The only exception is for migrants coming from Latin America and the Caribbean for which the unemployment rate of tertiary level residents is insignificant in the model. - 29 - The last four variables in table B4 are also worth considering. Labor productivity growth in host countries produces a strong—and expected—pull effect. Furthermore, the highly educated group is found to respond more strongly to an increase in labor productivity. In other words, when labor productivity is high, demand for labor is rather human-capital intensive. Public expenditure related to unemployment compensation in host countries (as a percentage of GDP) also affects the expatriation rates significantly and positively with, again, a stronger effect for the most educated group. This confirms the welfare magnet theories as presented by Borjas (1987, 1999). The last two variables measuring the level of political rights and civil liberties in origin countries are usually found insignificant in this setting, except for the level of political rights—at the 10 percent level, however—which appears negatively correlated to the expatriation rate of the highly educated individuals originating from the MENA Region. Finally, regional dummies are interesting, because they may capture common features in the immigration policies of some countries. Our results using the MENA sample show that, after controlling for bilateral and country-specific characteristics, Continental and Southern European countries are associated with higher expatriation rates from low-educated individuals than Anglo-Saxon and Northern European countries. However, this ranking is somewhat modified when one looks at the expatriation rates among medium and highly educated individuals: Southern European countries remain alone on top of the destination countries, followed by Continental European, Anglo-Saxon, and Northern countries. This last result is at odds with the generally admitted idea that Anglo-Saxon countries have had immigration policies biased toward the highly educated in the recent period as compared with other destination countries. However, this finding is not totally counterfactual, because Italy had implemented in 1998 policies potentially favoring the admission of foreign skilled workers via the allocation of an annual quota of residence permits to foreigners seeking employment. 38 This may have affected the expatriation rates of highly educated workers in the late 1990s, and hence driven our results. 39 To conclude this section, our findings emphasize the relevance of pull factors in explaining the magnitude of expatriation rates from developing to OECD countries despite restrictive immigration policies in most destination countries. At this stage of the analysis, these results, though not counterintuitive, need more empirical investigation to be confirmed. In particular, it is likely that unobserved time-invariant country-specific effects are at work in the different processes. The “panel data analysis” that follows is thus useful to check for robustness. Panel Data Analysis The major drawback of the approach we have adopted so far lies in the nature of the data used. Inference has been drawn on a cross-section of country data in one time-period. However, heterogeneity across countries in expatriation rates or migration flows is extremely likely and 38 Al-Azar 2005. 39 It is generally tempting in this connection to interpret the size and sign of fixed effects in econometric analysis while, in fact, the latter may capture a wide range of effects, and not just the one we would like to comment on. - 30 - therefore should be accounted for in the model. It is also likely that the business cycle (or “time” effect) affects bilateral migration flows. Ignoring these effects could result in biased estimates. 40 To identify these effects and hence enrich our previous analyses, we use a panel data set. Because time-series on migrants’ stocks do not exist, we use yearly data on immigrant flows into OECD countries by country of origin, between 1990 and 2002, and estimate a panel gravity model. As a consequence, this panel data analysis of migration flows should also be seen as a complement of our previous analysis based on stocks of migrants. Empirical model We exploit the panel structure of the data set and estimate two sets of regressions. In the first one, we introduce dummy variables for both destination and origin countries. This allows us to control for unobserved time-invariant country-specific effects. The basic empirical specification is a follows: Inflows ijt Pop jt = α i + β j + γ t + c1 log(distance ij ) + c2 Borderij + c3 Languageij + c4Colonyij + c5 Islandij + δ1 log(GDPpcit −1 ) + δ 2 log(GDPpc jt −1 ) + δ 3 PopDensit −1 + δ 4 PopDens jt −1 + δ 51524Shareit −1 + δ 61524Share jt −1 (II.4) + δ 7 AgeDepit −1 + δ 8 AgeDep jt −1 + δ 9UrbPopGrowthit −1 + δ10UrbPopGrowth jt −1 + δ11 log(MeanEduc)it −1 + δ12 Literacy jt −1 + δ13UnempRateit −1 + δ14 ProdGrowthit −1 + δ15 PubExpCompit −1 + δ16 PolRights jt −1 + δ17CivLib jt −1 +ν ijt where i is the destination country; j is the origin country, and t is time. Inflowsijt/Popjt is the emigration rate from j to i at time t. As before, Distanceij, Borderij, Languageij, Colonyij, and Islandij are variables measuring both geographic and cultural proximity between countries i and j. All the other variables are included to capture push and pull factors such as economic development, demographic, and political factors. All time-varying explanatory variables are lagged by one year to account for information on which the potential immigrants based their decision to move. In the second set of regressions, we rely on the robust fixed effects “within” estimator, which essentially adds a set of country-pair specific intercepts to the equation, and thus exploits only the time-series dimension of the data set around country-pair averages. The vector of explanatory variables is exactly the same as before, except that we drop regressors that are constant within country pairs (Distance, Border, Language, Colony, and Island). The return to education in the destination country is also dropped, because we could not get a longitudinal series for this variable. Empirical results 40 For example, in specification (2) (table B4), we found a positive coefficient on the destination country’s GDP per capita. Based on this sole result, it is not clear whether immigrants go to countries with higher GDP per capita or whether countries with higher GDP per capita have other characteristics that attract immigrants. - 31 - The results from estimating a gravity model of the gross migration inflows from 101 sending countries to 14 receiving OECD countries on annual unbalanced panel data for the period 1990– 2002 are presented in annex B, tables B5 and B6. 41 Table B5 reports the estimates of the basic specification with year, sending, and receiving countries’ dummy variables. Estimates on the whole sample are reported in column (1). Findings with regard to pull economic factors in destination countries suggest that (log) GDP per capita as a measure of gross income and unemployment rate as a proxy for employment opportunities have no significant impact on migration flows once destination and origin countries’ fixed effects are controlled for. By contrast, welfare state attractors, measured by the level of public social expenditure devoted to unemployment compensation expressed as a percentage of GDP, are found to have a strong and positive impact on migration inflows. Turning to the demographic variables, the share of population age 15–24 in the host country is found to exert a negative effect on migration flows. This result, which is similar to the one obtained using cross-sectional data, is consistent with the idea according to which migration is partly demand driven and used by some countries to compensate for population aging. Last, among the variables affecting the costs of migration, distance between destination and origin countries appears to be of importance. Estimates on the MENA sample are reported in column (2). First, sharing a border and speaking a common language positively and significantly affect emigration rates from countries of the MENA Region. The other findings using this restricted sample do not significantly differ from those reported in column (1): pull economic factors in destination countries are not significant except the share of public social expenditure devoted to unemployment compensation; demographics, in particular the share of the destination country’s population age 15–24, shapes bilateral flows and distance has a dampening effect on migration. Findings also suggest that lower degrees of political rights create emigration incentives. Political instability or a climate of insecurity not only can drive individuals to emigrate but also encourage governments to adopt a benign attitude toward the departure of individuals dissatisfied with their lot, as such persons could cause social unrest. Finally, columns (3), (4), (5), and (6) of table B5 report the results of the same gravity model estimated after splitting the destination countries into subgroups. Contrasted results emerge. The unemployment rate in the host country is found to have a strong significant negative effect on migration flows when the sample is restricted to European Continental countries while its effect is not significant for Anglo-Saxon and Southern countries. The effect of the share of public social expenditure devoted to unemployment compensation also varies between groups; it is significantly positive for European Continental countries and not significant elsewhere. Estimates of a gravity model with country-pair fixed effects are reported in table B6. The country-pair fixed effects allow us to control for time-invariant features of the destination country’s immigration policy that are specific for each sending country. Overall, the explicative power of all the independent variables is much higher once we control for country-pair specific 41 For many OECD countries, yearly data on migration flows disaggregated by country of origin are unavailable. - 32 - effects. Using the MENA sample, both income and employment opportunities in sending countries are now found to have a significant effect on migration (although negative in the case of income), and demographic determinants appear to be strong predictors of migration flows. In particular, the share of young adults in the population of the sending country is found to have a positive effect on migration, as predicted by theory (since the present discounted value of net migration benefits is higher the longer the remaining work-life time). This variable is also an indicator of demographic pressure. In many MENA countries, the economically active population is currently increasing at a more rapid rate than that of the total population, because of a time lag of about 20 years between the decline in fertility and its effect on the number of young persons entering the labor market. Such growth has an impact on the labor market (unemployment is highest among young people and women) and constitutes an important element that influences migration pressure. By contrast, the share of young adults in the population of the receiving country is found to have a negative effect on migration. This latter result suggests here again that demographic imbalances in receiving countries are associated with increased migrant flows. Urban population growth in the sending country is also found to increase emigration rate. Quoting Giubilaro, “New arrivals in the towns are for the most part young persons who have a higher level of education than the average in their areas of origin. Faced with the impossibility of finding employment in the structured sector, this workforce contributes to the increase in the number of precarious and under-paid jobs in the informal sector. The expansion of marginal employment and unemployment in the urban areas fuels socio-economic tensions in the countries concerned and increases emigration pressure.” 42 Last, we find a positive impact on migration flows stemming from literacy rates in the origin countries. Generally, any increase in the literacy rate reduces barriers to migration and is consequently associated with higher migration flows to rich OECD countries. To synthesize our key findings in a comparative way, table II.1 reports the signs and robustness levels of the effects of the determinants of migration using the alternative gravity models described above. Columns (1) and (2) display the results obtained using the cross-sectional analysis of expatriation rates (ratio of total stocks of migrants), while columns (3) and (4) display the findings stemming from the panel data analysis of yearly emigration rates (inflows). 42 Giubilaro 1997, p. 32. - 33 - Table II.1. The Determinants of Migration Using Gravity Models Cross-sectional analysis of total Panel data analysis of bilateral expatriation yearly emigration rates Method and dependent variables rates in 2000 for 1990-2002 s (Ratio of stocks, mij , 2000 )) (Inflows ijt/Popjt) All Other MENA (MENA MENA LDCs + OLDC) Determinants of migration (1) (2) (3) (4) Bilateral characteristics Ln(distance) --- - --- --- Common Border - n.i. +++ Common Language ++ +++ Ever in colonial relationship ++ ++ Island in couple (0, 1, or 2) Country-specific characteristics Log(GDP per capita in host) +++ +++ --- Log(GDP per capita in origin) --- Population density in host +++ +++ Population density in origin ++ --- -- Share of population aged 15-24 in host --- --- --- --- Share of population aged 15-24 in origin + Age dependency ratio in host +++ +++ --- -- Age dependency ratio in origin +++ +++ Urban population growth in host +++ ++ +++ Urban population growth in origin +++ Log(Mean education in host) ++ - +++ +++ Literacy rate in origin ++ ++ +++ Private return to education in host +++ +++ n.i. n.i. Total unemployment rate in host n.i. n.i. --- Unemployment rate (1994) for primary level in host -- + n.i. n.i. Unemployment rate (1994) for secondary level in host --- n.i. n.i. Unemployment rate (1994) for tertiary level in host +++ +++ n.i. n.i. Labor productivity growth in host +++ +++ Share of public expenditure related to unemployment in host +++ +++ +++ ++ Political rights in origin (scale 1 to 7 for high to low rights) +++ Civil liberties in origin (scale 1 to 7 for high to low liberties) --- Source: authors’ calculations. Note: The effects of the bilateral characteristics are based on results reported in annex B, tables B3 and B5, while those of the country-specific characteristics stem from regressions shown in tables B4 and B6. +++, ++, + mean respectively positive and significant effect at the 1 percent, 5 percent, and 10 percent confidence interval, respectively (---, --, and - mean negative effects); n.i. means variable not included in the model; an empty box means insignificant effect at the usual confidence interval (10 percent). For the exact definition of the explanatory variables, see the corresponding tables B3, B4, B5, and B6 in annex B. - 34 - III. CAN MIGRANTS FROM MENA COUNTRIES PROVIDE A SOLUTION FOR LABOR SHORTAGES IN OECD COUNTRIES? One of the key concerns associated with demographic changes in OECD countries is the decline in labor supply caused by population aging. Demographic projections are uncertain, but according to the “average” scenario, the ratio of people over 65 years of age to those between 20 and 64 could double between now and the middle of the century. In some countries, such as Italy, Japan, and Spain, this aging will be much stronger. 43 There have also been many discussions on skill shortages in the highly qualified segment of most OECD economies. 44 Both population aging and skill shortages call for an increase in the number of medium- and high- skilled immigrants. This section provides an overview of the magnitude of skill shortages in OECD countries and examines whether migrants originating from countries of the MENA Region can help filling labor market gaps. Labor Shortages in OECD Countries Identifying and projecting labor shortages for skilled and unskilled personnel is rather difficult, as rapid technological and labor market changes require a continual reevaluation of which skills are needed. Eurostat, the statistical office for the European Union, has recently undertaken an effort to develop an EU-wide job vacancy survey. At the same time, several countries have developed mechanisms to assess needs (see, for example, Italy’s Excelsior system, the United Kingdom’s Employers Skills Survey, and others). In what follows, we first present a general overview of job vacancies in some European countries. We then focus on one OECD country, France, and present a more detailed picture of the location of recruitment difficulties and skill shortages for this country. The discussion heavily draws on a report entitled “Selective Immigration Policies and Needed Skills,” which recently was made available by the French Ministry of the Economy, Finance and Industry. Overview We first present statistics on job vacancies in some European countries made available by Eurostat. These figures concern but a few European countries, and some important countries, such as France, Germany, and Italy, are missing from the list of selected countries. The available statistics do allow us to evaluate potential migration needs with a comparative approach. Figures III.1 through III.4 provide a picture of job vacancies and job vacancy rates at the aggregated level for nine European countries (Austria, Finland, Greece, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom). According to Eurostat, job vacancy is defined as a post newly created, unoccupied, or about to become vacant (i) for which the employer is taking active steps to find a suitable candidate from outside the enterprise concerned and is prepared to take more steps; and (ii) that the employer intends to fill either immediately or in the near future. An occupied post is a post within an organization to which an employee has been assigned. The job vacancy rate (JVR) measures the proportion of total posts 43 Cotis 2003. 44 See, for example, Doudeijns and Dumont 2003; Boswell, Stiller, and Straubhaar 2004. - 35 - that are vacant, according to the definition of job vacancy above, expressed as a percentage as follows: JVR = 100 x number of job vacancies / (number of occupied posts + number of job vacancies) Figures III.1 and III.2 illustrate that, compared with Austria, Finland, Greece, Luxembourg, the Netherlands, Portugal, Spain, and Sweden, the number of job vacancies in the United Kingdom has been much higher over the period 2001–04. This number has to be seen against that of occupied jobs in each country, which shows that the United Kingdom is far on top of the job providers among this short list of European countries. The United Kingdom is then followed by Spain, Sweden, Austria, Portugal, Greece, Finland, and Luxembourg. It is more informative, however, to look at the JVR (see figure III.3). In 2004, with a JVR of 4.2 percent, Greece stood far ahead, followed by the United Kingdom, the Netherlands, Finland, Sweden, Spain, Portugal, and Luxembourg. Figure III.1 Number of Job Vacancies in Some European Countries, 2001–04 Luxembourg Portugal Finland Sweden Austria Spain Greece Netherlands United Kingdom 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 Number of Job Vacancies 2001 2002 2003 2004 Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm Note: Data are missing for 2001–03 for Austria, and for 2001 for Finland and Greece. Figures for Luxembourg, though available, are just very small. - 36 - Figure III.2 Number of Occupied Jobs in Some European Countries, 2001–04 United Kingdom Spain Sweden Austria Portugal Greece Finland Luxembourg 0 5,000,000 10,000,000 15,000,000 20,000,000 25,000,000 30,000,000 Number of Occupied Jobs 2001 2002 2003 2004 Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm Figure III.3 Job Vacancy Rate in Some European Countries, 2001–04 4.5 4 3.5 Job Vacancy rate (%) 3 2.5 2 1.5 1 0.5 0 Greece United Netherlands Finland Sweden Spain Portugal Luxembourg Kingdom 2001 2002 2003 2004 Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm One limitation of the vacancy rate measure is that no single level of vacancies is considered to reflect shortages. There is, however, general agreement that an increase over time of the vacancy rate indicates a tight labor market. Figure III.4 depicts annual changes in job vacancy rates by country. The figure shows irregular drops and increases, although the magnitude of the variations is small and ranges from −0.7 to +0.2 percent only over the period under consideration (2001– 04). Job vacancy rates have been on the rise in five of the seven selected countries between 2003 and 2004. This evolution may reflect a shortage for some economic activities. - 37 - Figure III.4 Annual Changes in Job Vacancy Rate, 2002–04 0.4 0.2 Annual change in job vacancy rate 0 Netherlands United Luxembourg Spain Finland Portugal Sweden Kingdom -0.2 -0.4 -0.6 -0.8 2002 2003 2004 Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm Let us now analyze job vacancies at a more disaggregated level, using the NACE (General Industrial Classification of Economic Activities) classification of economic activities, for 2005 (see figure III.5). 45 The sectors hit hardest by labor market tightness vary from country to country. In the Austria, Portugal, and the United Kingdom, the number of job vacancies is the highest in the category combining sectors F (Wholesale and retail trade), G (Hotel and restaurants), and H (Transport and communication). More than 232,000 jobs in these sectors are vacant in the United Kingdom, 17,000 in Austria, and 6,000 in Portugal. 46 The category that combines sectors K (Public administration), L (Education), and M (Health) also records a high number of job vacancies: it is the second-hardest-hit category in the United Kingdom and the first one in Sweden, and probably in Finland (for this country, data relating to the Finance sector are missing, though). The category including sectors I (Finance) and J (Real estate and firm services), however, is on top of the job vacancy suppliers in the Netherlands, while this category ranks third in the United Kingdom, and second in Austria and Sweden. Among the selected countries, the number of job vacancies in the construction sector in 2005 is low compared with that recorded in the other sectors. This sector has a long tradition of hiring foreign workers and, as a consequence, often appears as the traditional immigration sector. 45 NACE is the European Community’s Classification of Economic Activities (Sector A: Agriculture, hunting and forestry; Sector B: Fishing; Sector C: Manufacturing ; Sector D: Electricity, gas and water supply; Sector E: Construction; Sector F: Wholesale and retail trade; repair of motor vehicles, motorcycles and personal and household goods; Sector G: Hotels and restaurants; Sector H: Transport, storage and communication; Sector I: Financial intermediation; Sector J: Real estate, renting and business activities, consulting; Sector K: Public administration and defense; compulsory social security; Sector L: Education; Sector M: Health and social work; Sector N: Other community, social and personal service activities; Sector O: Activities of households; Sector P: Extra-territorial organizations and bodies). 46 Unfortunately, we do not have data on the level of skills or type of occupations at which shortages are experienced for each of these categories. - 38 - Figure III.5 Number of Job Vacancies by Sector in 2005 250,000 Number of Job Vacancies by Sector 200,000 150,000 100,000 50,000 0 United Netherlands Austria Finland Sweden Portugal Luxembourg Kingdom Agriculture, hunting, forestry and fishing Industry except construction Construction Retail, hotel, transports and communication Finance, real-estate and firm services Public administration (education, health) Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm Note: Data for Agriculture, hunting, forestry, and fishing are missing for Austria, Finland, Portugal, and the United Kingdom. Data on finance are missing for Finland. Another way to scrutinize job vacancies across countries and sectors is reported in figure III.6. It gives a more insightful idea of which sectors have the highest number of job vacancies across a selection of 13 European countries now including some Eastern countries such as the Czech Republic, Estonia, Latvia, Lithuania, the Slovakia Republic, and Slovenia. From this figure, it appears that the most serious recruitment difficulties or labor demand and supply mismatch are in the sector composed of wholesale and retail trade, hotels and restaurants, and transport and communication, with about 344,000 vacant jobs in 2005. This sector is followed by the Public Administration and the Finance sectors which, respectively, record 279,000 and more than 224,000 vacant jobs. Vacancies for the Industry sector (excluding construction) stand far below, with 124,000 vacant jobs. If we add the number of vacant jobs recorded in the Construction sector to this figure, we get a total of 180,000 vacant jobs in 2005, which represents only about half that of the hardest-hit sector. Box III.1. Nurse Shortages in OECD Countries A recent OECD report provides evidence on current nurse shortages and surpluses in OECD countries. According to the report, the majority of OECD countries suffer from nurse shortages. Some countries have published estimates of how many headcounts of full-time equivalent nurses per year over the next decade would be needed to match demand for and supply of nurses: - Australia reports a shortage of around 6,000 registered nurses (around 3 percent of practicing registered nurses); - Canada’s shortage of registered nurses is estimated at 16,000 or 6.9 percent of the current workforce; - The Netherlands has reported a shortage of 7,000 nurses (1 percent of the workforce); - 39 - - In Norway, the shortage has been estimated at 3,300 full-time equivalents, or about 5.4 percent of practicing nurses; - In Switzerland, there are 3,000 (4.6 percent) fewer generalist nurses than required; and - The U.S. government reported a shortage of 110,700 registered nurses (5 percent of the workforce) in 2000. These national shortages are often unevenly distributed geographically and by specialty area. In the United Kingdom, for example, vacancy rates varied from 1.8 percent in district nursing to 4.7 percent in the category of “other psychiatry.” In Australia, employers in rural and remote settings have trouble finding sufficient numbers of qualified staff to work in areas such as long-term care. As explained by the authors of the report, these nurse shortages arise as a result of (i) increasing demand for nurses because of aging populations, new technologies that increase the range of conditions that can be treated, and a greater range of professional activism; and (ii) a falling or slow-growing supply due to fewer younger people entering the workforce, a greater range of professional opportunities, the low social value given to nursing, negative perceptions of nurse conditions of service, and an aging nurse workforce. Source: Simoens, Villeneuve, and Hurst 2005. Data on job vacancies suffer from a strong limitation, however, because they do not differentiate between the frictional and structural causes of vacancies. Job vacancies, in practice, are not only unfilled positions but also recruitment processes a certain amount of time being required to recruit, hire, and train new workers. Moreover, relying on information given by employers may produce a partial picture of the scale and causes of shortages. Despite these methodological problems, all the available data confirm that labor markets are tight in several OECD member countries. Labor shortages are experienced in all sectors, as well as in all kinds of jobs. According to Doudeijns and Dumont, “During the period from the second quarter of 1999 until the third quarter of 2000, employers increasingly complained that shortfalls in the supply of qualified labor caused limits to their production. Increasing difficulties related to labor shortages were reported by employers in Belgium, France, Italy, the Netherlands Austria, Portugal and the UK. Employers also report on tightness not only in labor markets for professionals with knowledge of and experience with the latest technologies, but also for personnel with low and medium level of qualifications. 47 ” Given the aforementioned limitations, the next section provides a more detailed picture of the location of recruitment difficulties and skill shortages for France. Focusing on this country is particularly relevant in the present study, as nearly one out of every two migrants originating from the MENA Region resides in this country. Recruitment Difficulties and Skill Shortages in France A recent report by the French Ministry of the Economy, Finance and Industry provides a detailed picture of recruitment difficulties and skill shortages in France. It is based on three data sources: (i) UNEDIC (Union Nationale pour l'Emploi dans l'industrie et le Commerce) surveys on labor needs; (ii) statistics on demand and supply of labor by sector of activity provided by the ANPE (Agence Nationale pour l’Emploi); and (iii) the report on jobs and qualifications published by 47 Doudeijns and Dumont 2003, p.5. - 40 - DARES (Direction de l’Animation de la Recherche, des Etudes et des Statistiques) and the Commissariat General au Plan. Figure III.6 Number of Job Vacancies by Sector in 2005 400,000 350,000 Number of Job Vacancies 300,000 250,000 200,000 150,000 100,000 50,000 0 Agriculture, Industry except Construction Retail, hotel, Finance, real- Public hunting, forestry construction transports and estate and firm administration and fishing communication services (education, health) Austria Czech Republic Estonia Finland Latvia Lithuania Luxembourg Netherlands Portugal Slovakia Slovenia Sweden United Kingdom Source: http://europa.eu.int/estatref/info/sdds/en/jvs/jvs_a_sm.htm Short-Term Labor Needs UNEDIC surveys on labor needs (enquêtes: “Besoins en main d’oeuvre”) are annual qualitative surveys based on indirect interviews involving a large number of employers of different scale across different sectors and localities in France. According to the 2005 survey, recruitment difficulties are high in the following occupations: • Occupations in the construction industry: skilled workers in heavy construction; skilled workers and technicians in engineering works; plumbers, painters, carpenters, tillers, and so on • Occupations in the restaurants and hotel trade: chefs and cooks, employees in the hotel trade • Personal service occupations: nursery nurses or child minders, security guards, nurses and midwifes, and care assistants As explained by the authors of the report, however, UNEDIC surveys suffer from the fact that they do not allow identifying the causes of these recruitment difficulties. They could be caused by an aggregate shortage of labor (in which case increased immigration could be part of the solution) or by problems of mismatch between labor demand and supply (in which case promoting better matching of people to jobs would be a more adequate response). - 41 - ANPE statistics on labor demand and supply by occupation provide a useful complement to UNEDIC qualitative surveys because they enable us to calculate labor market tightness index by sector of activity. This index is given by the number of jobs offered in a given occupation, divided by the number of job seekers in that occupation. The higher the index, the stronger the tension in the labor market due to labor shortages. Table III.1 shows the 20 occupations with the highest index over the period from July 2004 to June 2005. Table III.1. The 20 Occupations with the Strongest Labor Shortages in France (July 2004–July 2005) Number of jobs offered Occupation / Number of job seekers Skilled workers in the contracting industrya 1.85 Insurance agents and technicians 1.4 Employees in various service occupationsb 1.4 Wood industry unskilled workers 1.3 Chefs and cooks 1.15 Skilled workers in heavy construction 1.1 Technicians in the contracting industryc 1.1 Employees in the hotel trade 1.1 Nurses, midwifes 1.1 Electronic technicians 1.05 Skilled workers in the metal industry 1.05 Mechanical engineering technicians 1.05 Sales representatives 1.05 Skilled workers in the metal industry 0.95 Skilled workers in the process industry 0.95 Butchers, pork butchers and bakers 0.95 Electrical and electronics skilled workers 0.9 HGV and fork lift drivers 0.9 Sales managers and assistants 0.9 Mechanical skilled workers 0.85 Source: DGTPE 2006 Note: The French occupational classification created by DARES (FAP or nomenclature des familles professionnelles) distinguishes 224 occupations in its most detailed version and 84 at a more aggregated level. Occupations considered in the table belong to FAP-84. For a proper definition of each occupation, the reader should refer to FAP (http://www.travail.gouv.fr/IMG/xls/2_FAP_PCS_ROME_23fev06.xls) a. This category includes first-line supervisors/managers of construction trades and extraction workers and skilled construction trades workers. b. This heterogeneous category includes pump attendants, dry cleaners and launderers, self-employed workers such as morticians, directors of advertising or matrimonial agencies, and salaried workers (employees in gaming houses, janitors of changing rooms, and so on). c. This category includes architectural and engineering activities and related technical consultancy (geometers, quantity surveyors, and so on). Most occupations with high recruitment difficulties are associated with a high index of labor market tightness (higher than or equal to 0.85), suggesting that labor shortage more than skill mismatch is the cause of hard-to-fill job vacancies. Exceptions are for personal service occupations for which recruitment difficulties are high, but the index is low: 0.45 for care assistants, 0.2 for nursery nurses or child minders, and 0.65 for security guards. For these occupations, recruitment difficulties seem to be due to problems of mismatch between labor demand and supply. Because of information deficits and other inefficiencies on the labor market, unemployed workers do not acquire information on relevant existing vacancies, and firms or - 42 - employers do not have the information necessary to find people with adequate qualifications. The response to this problem is to promote better matching of people to jobs rather than to increase migration flows. Last, we investigate whether labor market tightness in occupations with high recruitment difficulties is a structural or cyclical phenomenon. To do so, we examine whether occupations with a high labor market tightness index in 2005 already had a high labor market tightness index in 2000. We find that most occupations facing labor shortages in 2005 were already under pressure in 2000. For four occupations in the industrial sector (wood industry unskilled workers, electrical and electronics skilled workers, and skilled workers in the metal industry), however, the number of created jobs has been rather low over the last four years (less than 4,000 created jobs in four years). It thus seems that labor market needs in the short run are particularly high in the following occupations: skilled workers and technicians in the contracting industry; chefs and cooks; employees in the hotel trade; nurses and midwifes; skilled workers in heavy construction; mechanical engineering technicians and skilled workers; butchers, pork butchers, and bakers; sales representatives; and sales managers and assistants. For these particular occupations, labor shortages could be alleviated through domestic labor market reforms (such as training or retraining programs for the currently unemployed, promotion of occupation and geographic mobility, job-search assistance, and so on) or through increased migration. Long-Term Labor Needs To assess labor and skill needs in the longer run, we use DARES projections on job creations and losses in labor supply resulting from the retirement of baby boomers by occupation. Assuming a growth rate of 2 percent per year over the period 2005–15, there would be 750,000 job openings per year. The greatest retirement pressures will be in the following sectors: contracting industry (413,000 vacant jobs due to retirements over the period); tourism and transport (444,000 vacant jobs); accounting and administrative services (641,000 vacant jobs); trade (453,000 vacant jobs); personal services (790,000 vacant jobs); and public administration (500,000 vacant jobs). In net terms, most job creations will be in the tertiary sector both at the skilled and unskilled levels. Most unskilled jobs will be created in the following sectors: personal services (more than 416,000 jobs over the period), transport and tourism (more than 225,000 jobs), trade (more than 196,000 jobs), and contracting industry (more than 116,000 jobs). By contrast, most skilled jobs will be created in the health sector (more than 304,000 jobs) and accounting services (more than 197,000 jobs). Labor and skill needs will be particularly high in the following occupations: • Managers and professional occupations: teachers, managers, and administrative officers; software and information technology professionals; sales managers and technicians; managers and technicians in the contracting industry; and research engineers and researchers - 43 - • Technical and skilled occupations: executive assistants, nurses, sales representatives, intermediate sales and service occupations, administrative clerks, supervisors, and processing occupations • Skilled workers: skilled drivers, skilled workers in processing, skilled workers in storehouse management, and skilled workers in heavy construction • Unskilled occupations: child minders, care assistants, in-home employees, light-duty cleaners Domestic Labor Market Reforms or Increased Migration? The previous discussion has shown that many OECD countries are currently experiencing substantial unsatisfied labor requirements in a number of sectors despite persistently high rates of unemployment. This suggests that, in most cases, labor shortages are not caused by an aggregate shortage of labor, but instead can be attributed to problems of mismatch between labor demand and supply. Jobs remain unfilled because workers lack the relevant qualifications or skills (qualitative mismatch), are reluctant to take up work in particular occupations or geographic areas (preference or regional mismatch), or have insufficient information about job opportunities (mismatch due to information deficits). In addition, employers may be unwilling or unable to offer sufficiently attractive salaries or conditions to encourage occupational or geographical mobility. 48 Increased immigration is not the only policy option to fill labor gaps. Several measures could promote better matching of people to jobs. First, the problem of hard-to-fill vacancies may be addressed through offering more attractive salaries or working conditions. Some particular jobs in the health sector and in the hotel industry are indeed hard to fill because of heavy workloads, the increased use of overtime, and low remuneration. These jobs could be made more attractive through higher wages and a reorganization of working time. Second, labor shortages could be alleviated through relevant training and education measures to increase the supply of qualified domestic workers. Inadequate qualifications are indeed one of the main factors that explain labor tightness in some qualified jobs. Third, policies introducing incentives to encourage mobility may solve the problem of hard-to-fill vacancies in particular regions. Last, any increase in participation rates (especially those of married women) or in the number of years spent in work could help fill labor gaps. However, most of these measures require several years to take effect. Moreover, many low-skilled, low-status, and low-paid professions have become associated with immigrant or ethnic minority workers. Under these conditions, turning to labor migration programs may be a more rapid and effective way to address shortages. 48 Boswell, Stiller, and Straubhaar 2004. - 44 - Can Migrants from the MENA Region Help Fill Labor Market Gaps in OECD Countries? Demographic Prospects A recent paper compares demographic trends in Western and Central Europe 49 with those in Eastern Europe, the Balkans, Turkey, and Central Asia (hereafter EECA-20 50 ) and those in the MENA Region (hereafter MENA-20 51 ). 52 As shown in this report, using Eurostat data and projections by the United Nations, demographic trends in Western and Central Europe and in most countries of the EECA-20 are quite similar: because of low fertility and increasing life expectancy, total population size in these two groups will remain stable during the next 20 years and will start to decrease only during the following decades. But the number of younger people will shrink and the number and share of older people will increase during the same period. As a consequence, the old-age dependency ratio (the ratio of the number of people age 65 or older divided by the number of people age 15 to 64) will dramatically rise. In Western and Central Europe, the ratio is expected to increase from 25 percent in 2005 to 35 percent in 2025 and 55 percent in 2050. A diminishing and aging working- age population necessarily results in a diminishing and aging labor force. At constant participation levels and in the absence of immigration, the labor force is projected to decrease by 66 million people, from 226 million in 2005 to 160 million in 2050 (−29 percent). 53 As shown by this author, neither an increase in overall participation rates or in the participation rate of women, nor an increase in the retirement age would be sufficient to hold the labor force at its 2005 level. In contrast, the situation in the MENA-20 is characterized by higher—but declining—fertility, increasing life expectancy, and sustained demographic growth. Total population in the MENA- 20 will grow steadily from 316 million in 2000 to 492 million by 2025 (+55.7 percent) and to 638 million by 2050 (+102.0 percent). During this period, the number of people between ages of 15 and 64 will more than double, from 187 million in 2000 to 323 million by 2025 (+72.7 percent) and will continue to grow at almost the same rate to 417 million by 2050 (+123.0 percent). 54 Koettl gets comparable trends using a sample restricted to 12 MENA countries. 55 Using projections from the International Labour Organization (ILO) Economically Active Population database, Holzmann and Münz further estimate that during the period 2000–2025, the 49 The 28 EU+EEA countries and Switzerland. 50 The EECA-20 countries consist of Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Croatia, Georgia, Kazakhstan, the Kyrgyz Republic, Macedonia, Moldova, Romania, the Russian Federation, Serbia and Montenegro (including Kosovo), Tajikistan, Turkey, Turkmenistan, Ukraine, and Uzbekistan. 51 The MENA-20 consist of Algeria, Bahrain, Djibouti, the Arab Republic of Egypt, the Islamic Republic of Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Occupied Palestinian Territories, Qatar, Saudi Arabia, the Syrian Arab Republic, Tunisia, United Arab Emirates, and the Republic of Yemen. 52 Münz 2004. 53 Koettl 2006. 54 Münz 2004. 55 Koettl’s (2006) sample includes the Maghreb (Algeria, Morocco, Tunisia) and the Mashreq (Egypt, Jordan, Lebanon, the Syrian Arab Republic) countries, Djibouti, the Islamic Republic Iran, Iraq, the Occupied Palestinian Territories, and the Republic of Yemen. - 45 - job-seeking population will increase by 93 million in the MENA-20. 56 As suggested by these two authors, the main challenge in this region will be to absorb those currently unemployed (compare table III.2 for statistics on unemployment rate by country) and those entering the labor market during the next two decades. To fully cope with this challenge, the MENA-20 countries would have to create 45 million new jobs until 2010 and more than 100 million until 2025. According to Münz, “The current labor market conditions in many MENA-20 countries raise doubts as to whether these economies will be able to absorb the significant expansion of the labor force. As a consequence, migration pressures on the contracting labor markets in Europe will increase.” 57 Table III.2. Unemployment Rates by MENA Country in Recent Years Country Unemployment Rate (%) Year Labor Force in 2002 Maghreb Algeria 28.7 1997 11,472,694 Morocco 17.8 1996 12,162,105 Tunisia 15.9 1997 4,097,709 Mashreq Egypt, Arab Rep. Of 9.4 1998 27,444,152 Jordan 13.7 2001 2,035,570 Lebanon 18.0 1997 1,215,942 Syria, Arab Republic 25.0 1999 5,560,916 GCC States Bahrain 5.0 2000 291,440 Kuwait 1.3 1997 854,535 Oman 10.0 1995 783,990 Saudi Arabia 10.0 1998 7,610,381 UAE 6.7 2000 1,241,670 Other MENA Iran, Islamic Rep. Of 16.2 2000 25,778,539 Iraq 50.0 1999 6,873,680 Libya 30.0 1994 1,954,684 Yemen, Rep. of 30.0 1996 6,165,097 All 18.7 115,543,102 Source: Keller and Nabli 2002. From a strict quantitative standpoint, the diverging demographic trends and structural differences between most European countries and MENA economies confirm that potential synergies need to be developed between the two regions. In particular, increased labor mobility from the MENA Region could compensate for demographic trends in European labor markets in the next decades, while responding to the lack of employment in the home countries. Besides, evidence from the estimated gravity models (section III of this study) does suggest that migration flows from one country to another are strongly driven by demographic features. As shown by Koettl, however, the potential of MENA emigration countries to provide labor migrants to European countries will decrease rapidly after 2020, since the 25–39 age-group, which is the most likely age-group to migrate, will grow at a much lower rate between 2020 and 2050. 58 56 Holzmann and Münz 2004. 57 Münz 2004, p.18. 58 Koettl 2006. - 46 - The central question of interest is not so much whether the MENA region can provide high numbers of working-age individuals or not but, rather, what kind of migrants the region can provide. In other words, what is needed is an evaluation of the quality of the matching between the labor and skill requirements of European labor markets and the skills of people migrating from the MENA region to Europe. Europe’s Skills Requirements and MENA’s Skill Supply: Is There a Match? Providing a definite answer to the above question is rather difficult, particularly given the lack of detailed information on the level and structure of qualifications of MENA countries’ population, on the employment conditions prevailing in the labor markets of these countries (which condition the desire of people to migrate), on human capital transferability, and so on. To get an idea of what the answer could be, however, we can look at the current labor market situation of migrants originating from the MENA Region in OECD countries. In a more prospective approach, it is useful to examine available data on unemployment rates by NACE sector in MENA countries. Current labor market situation of immigrants originating from the MENA Region The current labor market situation of migrants from the MENA Region strongly varies from one OECD country to another, as a result from the diversity of the socioeconomic composition of the migrant populations in destination countries (compare section I). This diversity itself results from the period of migration and from migration and labor-market policies in destination countries. 59 Generally, the earlier cohorts of migrants from the MENA Region to OECD countries were composed mainly of individuals with low qualifications, whereas the more recent ones are composed of people with medium to high qualifications. Available statistics on European countries show that immigrants from Maghreb countries have lower labor force participation rates and higher unemployment rates than natives. In France, in particular, the unemployment rate for North Africans in 2002 (which was 7 percent for natives) was 26 percent for Algerians, 26 percent for Moroccans, and 22 percent for Tunisians.60 Unemployment rates are even higher if one considers the 15–24 age-group alone. The same holds true for women, partly because they are often unskilled workers. Among French women, the unemployment rate was 8 percent in 2002 against 30 percent for Algerian women and 31 percent for Moroccan women. In the Netherlands, the unemployment rate in 1994 was 6.4 percent for nationals but 31 percent for Moroccans. At the same time in this country, the Moroccan population showed the lowest rate of participation in the labor force. Statistics for Spain are given in table III.3 which depicts the distribution of immigrants originating from Maghreb and Mashreq countries by type of activity, in 2001. The employment rate among immigrants from Maghreb and Mashreq countries is relatively low (49.5 percent) compared with the EU-15 average (64 percent). This rate varies from one origin country to another. While it is 44.5 percent for Syrian immigrants, it amounts to nearly 60 59 Fargues 2005. 60 INSEE 2002. - 47 - percent for Tunisians. Overall, a large proportion of immigrants are still studying (19 percent and 18 percent for individuals from Maghreb and Mashreq countries, respectively), while about 11 percent are unemployed and 18 percent to 20 percent do not participate to the labor market for other reasons. Table III.3. Immigrants from Maghreb and Mashreq Countries by Type of Activity in Spain, 2001 (percent) Retirement Other Students Employed Unemployed allowance Total situationa beneficiaries Maghreb 19.4 49.6 11.6 0.7 18.8 100 Algeria 14.1 56.6 14.7 1.1 13.6 100 Morocco 19.8 48.9 11.3 0.7 19.3 100 Tunisia 13.4 58.4 12.5 1.4 14.4 100 Mashreq 17.9 48.2 11.3 2.2 20.3 100 Egypt, Arab Rep. of 15.3 54.5 11.3 2.1 16.8 100 Jordan 19.9 45.6 12.0 2.0 20.6 100 Lebanon 19.0 47.9 10.8 3.5 18.8 100 Syrian Arab Rep. 18.4 44.5 11.4 1.5 24.2 100 Total 19.4 49.5 11.6 0.7 18.8 100.0 Sources: CARIM 2001; authors’ calculations. Note: a. This category includes incapacity benefit beneficiaries, widowed or orphaned benefit beneficiaries, carrying out or sharing domestic chores, and other situation such as uneducated minors, independently wealthy, and so on. Many factors explain the gaps in employment and unemployment rates between natives and immigrants from the MENA Region. The first factor is the reduced employability of this category of labor. With the exception of Greece, Ireland, Italy, and Spain (that is, countries in which labor immigration dominates), immigration flows to most other European OECD countries mainly include categories of individuals admitted on humanitarian or social grounds (annual refugee quotas, asylum seekers, family reunion) who face difficulties in integrating into the labor market. 61 A second factor is the mismatch between the skills of people migrating to Europe and the requirements of Europe’s formal labor markets. Hidden barriers for access to employment (such as a poor understanding of cultural and workplace norms) as well as employers’ discrimination also explain the greater vulnerability of foreign workers to unemployment and their lower degree of employability. To conclude, the current labor market situation of immigrants originating from the MENA Region suggests that a significant proportion of these immigrants is not necessarily endowed with the required skills. To investigate whether this situation can be improved in the near future, the next section aims to identify the sectors of MENA countries in which unemployment is a particularly severe problem (and in which, as a consequence, people may have a stronger incentive to migrate). 61 A significant proportion of population movements within and from MENA results from political instability and armed conflicts. Over the recent period, people from Iraq, the Islamic Republic of Iran, and Algeria have been among the largest groups of asylum seekers in Europe. - 48 - In which sectors of MENA economies is there an excess supply of labor? In what follows, our approach compares the potential for emigration in origin countries by sector with the labor shortages of potential host countries that were detailed in the previous section. One way of doing this is to look at unemployment figures for each segment of the labor market in MENA countries—that is, for each NACE sector listed in footnote 45. High unemployment numbers in some sectors are likely to be due to skill mismatch or to lack of available jobs in these sectors. If the latter is true, migration might be an appropriate means for job seekers to find a suitable job in the same sector abroad. Hence, it is interesting to investigate whether sector- specific labor shortages in receiving countries correspond to complementary sector-specific labor market imbalances in potential sending countries. To do that, we use ILO data on unemployment figures disaggregated by NACE sector for four important sending countries (Algeria, Egypt, Morocco, and, for a comparative purpose, Turkey). To observe whether there are complementary imbalances between sending and receiving countries, compare the histograms in figures III.7 to III.10with the number of job vacancies by sector in 2005 for 13 European countries (see figure III.6). Figure III.7 Unemployment Numbers in Algeria by Sector, 2000–2004 250,000 200,000 Number of unemployed 150,000 100,000 50,000 0 Agriculture, Industry except Construction Retail, hotel, Finance, real- Public Other hunting, forestry construction transports and estate, firm Administration & fishing communication services (education, health) 2000 2001 2002 2003 2004 Source: ILO 2006 - 49 - Figure III.8 Unemployment Numbers in Egypt by Sector, 2000–2004 60,000 50,000 Number of unemployed 40,000 30,000 20,000 10,000 0 Agriculture, Industry except Construction Retail, hotel, Finance, real- Public Other hunting, forestry construction transports and estate, firm Administration & fishing communication services (education, health) 2000 2001 2002 2003 2004 Source: ILO 2006 Figure III.9 Unemployment Numbers in Morocco by Sector, 2000–2004 200,000 180,000 160,000 Number of unemployed 140,000 120,000 100,000 80,000 60,000 40,000 20,000 0 Agriculture, Industry except Construction Retail, hotel, Finance, real- Public Other hunting, forestry construction transports and estate, firm Administration & fishing communication services (education, health) 2000 2001 2002 2003 2004 Source: ILO 2006 - 50 - Figure III.10 Unemployment Numbers in Turkey by Sector, 2000–2004 700,000 600,000 Number of unemployed 500,000 400,000 300,000 200,000 100,000 0 Agriculture, Industry except Construction Retail, hotel, Finance, real- Public Other hunting, forestry construction transports and estate, firm Administration & fishing communication services (education, health) 2000 2001 2002 2003 2004 Source: ILO 2006 Several interesting patterns emerge: • In the case of Egypt (figure III.7), unemployment figures are the highest for the Wholesale and Retail Trade, Hotels, Transports and Communications sector. Interestingly enough, this sector is actually the one facing the most serious recruitment difficulties in European countries, as suggested by figure III.6 (with about 344 thousands vacant jobs in 2005). Unemployment figures in this sector are also relatively high in Turkey and, to a lesser extent, in Algeria. By contrast, the Moroccan case exhibits low unemployment figures in this sector. There have indeed been long-standing labor shortages in the branch of tourism and hotels in this country. Emigration from unemployed Egyptians, Algerians, and Turks in this sector could thus represent a solution to some specific labor shortages in European countries. • Unemployment figures in the Industry, Excluding Construction and Public Administration sectors strongly vary from one country to another. In the Industrial sector, figures are rather low in Turkey, while they are quite high in Morocco and at an intermediate position in Algeria and Egypt. In the public sector, unemployment figures are low in Egypt and Turkey, while they are high in Algeria and Morocco. In terms of job vacancies in European countries, the Public Administration sector was found to be the second-hardest-hit sector. Whether increased migration from unemployed Algerians and Moroccans could alleviate labor shortages in this sector is not clear, however, because many occupations in public administration are not open to nonnationals. • Two cases of noncomplementary imbalances are visible. First, labor shortages in the Finance, Real-Estate[,] and Firm Services sector in European countries cannot be alleviated through increased migration from Algeria, Egypt, Morocco, and Turkey, as - 51 - unemployment figures are rather low in these countries. Second, despite high unemployment figures in the Construction sector for the four origin countries under consideration, this sector in host countries no longer appears to face strong recruitment difficulties. To sum up, available data on labor market imbalances in both sending and receiving countries suggest that increased labor mobility from the MENA region could compensate for labor shortages in European labor markets in the next decades, while responding to the lack of employment in some home countries. This is particularly true in the Wholesale and Retail Trade, Hotels, Transports, and Communications sector. This conclusion needs to be treated with caution, however, because more disaggregated data are lacking on both labor shortages in OECD countries and labor surpluses in MENA countries. Is there a potential for outsourcing? Outsourcing (or contracting out) means delegating operations or jobs from internal production within a business to an external entity (such as a subcontractor) that specializes in that operation. The decision to outsource is often made to lower costs. A related term, offshoring, means transferring work to another country, typically overseas. Offshoring is similar to outsourcing when companies hire overseas subcontractors, but differs when companies transfer work to the same company in another country. Business segments typically outsourced include Information Technology, Human Resources, Facilities and Real Estate Management, and Accounting. Many companies also outsource customer support and call center functions, manufacturing, and engineering. 62 According Bardhan and Kroll, “Firms involved in outsourcing are rapidly gaining ground in China, the Philippines, Malaysia, Russia, Israel and Ireland. While it is difficult to estimate the exact number of jobs created in these countries, tentative evidence suggests that outsourcing has generated at the very least over a million jobs in the 1990s and hundreds of thousands more since the turn of the century”. 63 Bardhan and Kroll outline the following attributes of outsourced jobs: • No face-to-face customer servicing requirement • High information content • Work process is telecommutable and Internet enabled • High wage differential with similar occupation in destination country • Low setup barriers • Low social networking requirement 64 Whether offshoring could gain ground in the MENA Region is not only conditioned by the business environment prevailing in MENA economies but also by labor costs, skill availability, and so on (see figure III.11). 62 Wikipedia 2006. 63 Bardhan and Kroll 2003, p.2. 64 Bardhan and Kroll 2003, p.4. - 52 - Figure III.11 Approximate Value of Offshore Services in Countries that Supply Them Source: McKinsey and Company 2005. As suggested by table III.4, the business climate is a challenge in the region: in Maghreb countries, for example, Algeria and Morocco, respectively rank 128 and 102 out of 155 in terms of “Ease of doing business.” Egypt and Syria are also very poorly ranked, as are Iraq and the Islamic Republic of Iran. By contrast, countries currently involved in offshoring are relatively better ranked: 21 in the case of Malaysia, 28 in the case of South Africa, 79 in the case of the Russian Federation, and 91 in the case of China. There are exceptions, however, as India ranks 116 and the Philippines 113, suggesting that firms consider other elements in their decisions to hire overseas subcontractors. The cost of labor is one of these other elements. Unfortunately, getting information on labor costs by NACE sector is rather difficult. Figure III.12 reports prevailing wages by economic activity for a small list of countries for which data are available. This figure compares wages in some MENA countries with prevailing wages in countries in which offshoring has been gaining ground over the last 10 years (including India, Malaysia, the Philippines, and Russia). In this comparative approach, we have chosen the economic sectors that have been the most prevalent in offshoring. - 53 - Table III.4. Business Environment in the MENA Region Ease of Trading Starting a Getting Protecting Enforcing Closing a Doing Across Business Credit Investors Contracts Business Business Borders Maghreb Algeria 128 109 138 97 84 131 46 Morocco 102 50 146 117 98 29 51 Tunisia 58 40 102 133 53 6 31 Mashreq Egypt, Arab Rep. Of 141 115 142 114 70 118 106 Jordan 74 119 65 124 61 58 70 Lebanon 95 99 66 102 94 142 98 Syrian Arab Rep. 121 135 124 105 146 149 65 GCC States Kuwait 47 87 52 71 69 98 45 Oman 51 61 147 50 79 87 76 Saudi Arabia 38 147 56 74 87 95 67 UAE 69 134 104 96 19 133 130 Other MENA Iran, Islamic Rep. Of 108 54 64 138 132 55 94 Iraq 114 117 133 85 155 74 148 Occupied Palestinian Territories 125 152 129 .. 75 88 155 Yemen, Rep. of 90 151 137 108 68 57 68 Source: World Bank 2005. Note: Ranking covers 155 countries in the world. Figure III.12 Wages for Sectors in Selected Countries Panel A: Wages in the Manufacturing Sector (in US$ per month) Manufacturing India 2001 Russia 1998 Thailand 2001 Egypt 2001 Philippines 1999 Jordan 2001 Malaysia 2001 Qatar 2001 Bahrain 2001 (*) South Africa 2001 (*) 0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 - 54 - Panel B: Wages in the Real Estate, Renting,and Business Activities Sector (US$ per month) Real Estate, Renting and Business Activities Russia 1998 Philippines 1999 Algeria 1996 Thailand 2001 Egypt 2001 Jordan 2001 Qatar 2001 Israel 2001 0.0 200.0 400.0 600.0 800.0 1,000.0 1,200.0 1,400.0 1,600.0 1,800.0 Panel C: Wages in the Financial Intermediation Sector (US$ per month) Financial Intermediation Algeria 1996 Egypt 2001 Russia 1998 Thailand 2001 Philippines 1999 Bahrain 2001 (*) Jordan 2001 South Africa 2001 (*) Qatar 2001 Israel 2001 0.0 500.0 1,000.0 1,500.0 2,000.0 2,500.0 3,000.0 3,500.0 Sources: ILO 2006; authors’ computations. Overall, data suggest that some MENA countries offer significant cost advantages, compared with other countries. In Algeria and Egypt, in particular, wages appear to be lower than in some of the countries currently providing offshored services. Offshoring business process or services is not solely cost driven. In addition to cost advantages, the location choice of offshored services is conditioned by other factors among which are the presence of an English-speaking population (as U.S. businesses dominate the global share of offshoring, accounting for some 70 percent of the total market); a common accounting and legal system based on the common law structure of the United Kingdom and the United States; a time-differential determined by geographic location leading to a 24/7 capability and overnight turnaround time; and a steady and large supply of technically savvy graduates. 65 On some of these particular aspects, compared with other countries, MENA countries are undeniably disadvantaged. Furthermore, barriers in some MENA 65 As explained by Bardhan and Kroll 2003. - 55 - economies often cause high transaction costs for foreign investors. These barriers include arbitrary government behavior affecting the investment climate, corruption, discrimination against foreign investment, insecure or unregulated financial systems, restrictions to the free flow of capital, and the nonadoption of international standards of accounting and arbitration. The question of whether offshoring could gain ground in the MENA region is a complex one. Thus far, the region has played a minor role in the ongoing outsourcing of business service jobs, the latter being mainly generated by U.S. or British firms. Given the prevailing conditions in some of the French-speaking MENA countries (namely, Morocco and Tunisia), these countries could appear in the future offshoring plans of French and other Continental countries. As an illustration, Morocco has a significant French legacy as well as some Spanish. French is the language of business, government, and diplomacy; its civil law draws from the French and Spanish civil laws; and, geographically, it is separated from Spain by the narrow Strait of Gibraltar. - 56 - STATISTICAL ANNEX ANNEX A. DESCRIPTIVE STATISTICS Table A1. Distribution of Migrants by Region of Origin in OECD Countries, 2000 Number of migrants by region of birth Destination Total stock country MENA % OECD countries % Other % of migrants % EEA/EU Austria 29,319 3.2 208,941 22.6 686,227 74.2 924,487 100.0 Belgium 148,557 14.6 577,916 56.7 293,354 28.8 1,019,828 100.0 Denmark 46,859 13.7 116,492 34 179,039 52.3 342,390 100.0 Finland 6,987 6 41,287 35.3 68,815 58.8 117,089 100.0 France 2,329,229 41.6 1,924,101 34.4 1,346,868 24.1 5,600,199 100.0 Germanya 98,750 0.8 1,244,000 9.7 11,538,750 89.6 12,881,500 100.0 Greece 51,033 5.1 185,804 18.6 764,187 76.3 1,001,024 100.0 Hungary 2,664 1 22,347 8.1 250,483 90.9 275,494 100.0 Ireland 4,227 1.3 262,838 78.9 65,939 19.8 333,005 100.0 Italy 292,918 14.5 733,220 36.3 994,796 49.2 2,020,935 100.0 Luxemburg 1,841 1.4 108,484 82.6 21,063 16 131,389 100.0 Netherlandsa 197,982 14.8 292,343 21.9 846,707 63.3 1,337,032 100.0 Polanda 2,735 0.2 139,325 11.1 1,112,124 88.7 1,254,184 100.0 Portugal 2,040 0.3 149,999 25.6 433,893 74.1 585,932 100.0 Slovak Republica 366 0.1 4,064 0.8 514,225 99.1 518,655 100.0 Spain 311,351 16.8 571,999 30.8 973,697 52.4 1,857,047 100.0 Sweden 154,620 16.6 357,498 38.3 420,350 45.1 932,469 100.0 Czech Republica 2,245 0.4 28,035 4.6 580,138 95 610,418 100.0 United Kingdom 173,277 3.8 1,597,491 35.5 2,732,553 60.7 4,503,321 100.0 Total EEA/EU 3,857,000 10.6 8,566,184 23.6 23,823,208 65.7 36,246,398 100.0 EEA/non-EU Norway 29,513 10.1 121,080 41.5 140,828 48.3 291,422 100.0 Other OECD Australiaa 160,836 3.5 2,128,830 46.2 2,315,742 50.3 4,605,408 100.0 Canada 282,930 5.3 2,051,320 38.3 3,021,325 56.4 5,355,575 100.0 Japan 4,101 0.4 532,182 46 621,071 53.7 1,157,354 100.0 Korea, Rep. ofa 0 26,757 19 114,059 81 140,816 100.0 Mexicoa 2,169 0.5 157,650 37.9 255,909 61.6 415,728 100.0 New Zealanda 8,505 1.1 337,656 45.4 397,656 53.5 743,817 100.0 Switzerlanda 39,843 2.3 816,358 47.9 848,747 49.8 1,704,949 100.0 Turkey 26,328 2.7 355,530 36.5 591,403 60.8 973,261 100.0 United States 809,759 2.6 5,824,199 18.4 25,089,139 79.1 31,723,097 100.0 Total Other OECD 1,334,471 2.9 12,230,482 26.1 33,255,051 71.0 46,820,005 100.0 Total 5,220,984 6.3 20,917,746 25.1 57,219,087 68.6 83,357,825 100.0 Sources: Dumont and Lemaître 2005; authors’ calculations. Note: a. Figures to be treated with caution due to the large share of immigrants with unknown country of birth. - 57 - Table A2. Numbers of Algerian Migrants According to which Country Counts Country which counts migrants Difference (2)-(1) Country of Destination Origin country (2) destination country (1) Absolute Relative (%) Belgium 7,221 18,166 10,945 152 France 685,558 932,275 246,717 36 Germany 17,308 9,018 -8,290 -48 Italy 15,750 5,753 -9,997 -63 Spain 23,785 7,079 -16,706 -70 Sweden 531 2,907 2,376 447 United Kingdom 10,672 15,000 4,328 41 Canada 19,095 10,080 -9,015 -47 Morocco 14,392 25,000 10,608 74 Switzerland 3,127 2,924 -203 -6 Tunisia 9,612 30,000 20,388 212 Subtotal 807,051 1,058,202 251,151 31 Other countries n.a. 14,044 Total 1,072,246 Source: Table extracted from CARIM’s Mediterranean Migration 2005 Report. Note: n.a. = not applicable. Table A3. Numbers of Egyptian Migrants According to which Country Counts Country which counts migrants Difference (2)-(1) Country of Destination Origin country (2) destination country (1) Absolute Relative (%) Austria 4,721 14,000 9,279 197 France 15,974 36,000 20,026 125 Germany 14,477 25,000 10,523 73 Greece 7,448 60,000 52,552 706 Italy 40,879 90,000 49,121 120 Netherlands 10,982 40,000 29,018 264 Spain 1,567 12,000 10,433 666 United Kingdom 24,705 35,000 10,295 42 Australia 33,370 70,000 36,630 110 Canada 35,975 110,000 74,025 206 Jordan 124,566 226,850 102,284 82 Switzerland 1,369 14,000 12,631 923 United States 113,395 318,000 204,605 180 Subtotal 429,428 1,050,850 621,422 145 Other countries n.a. 1,685,879 Total 2,736,729 Source: Table extracted from CARIM’s Mediterranean Migration 2005 Report. Note: n.a. = not applicable. - 58 - Table A4. Seven Main Nationalities of Migrants from MENA Countries in Selected English-Speaking OECD Countries Share in total Share in total Seven main stocks of Share of which Seven main stocks of Share of which MENA sending migrants from are poorly MENA sending migrants from are poorly countries MENA countries educated (%) countries MENA countries educated ($) (%) (%) Australia Canada Lebanon 44.4 40.9 Lebanon 24.4 31.6 Egypt, Arab Rep. Iran, Islamic Rep. 21.5 18.4 23.4 10.1 of of Egypt, Arab Rep. Iraq 12.2 39.0 13.6 9.4 of Iran, Islamic Rep. 10.0 17.4 Morocco 9.5 16.5 of Syrian Arab Rep. 4.1 40.1 Iraq 8.2 30.4 Occupied Palestinian 1.7 25.8 Algeria 6.2 9.4 Territories Jordan 1.7 14.0 Syria 5.6 29.5 Subtotal 95.6 Subtotal 91.0 Percent in total Percent in total 3.5 5.3 migrant stocks migrant stocks United Kingdom United States Iran, Islamic Rep. Iran, Islamic Rep. 24.5 19.1 35.4 6.7 of of Egypt, Arab Rep. Iraq 17.2 38.6 13.6 3.6 of Egypt, Arab Rep. 14.3 28.8 Lebanon 13.1 11.4 of Morocco 8.0 28.8 Iraq 10.2 19.9 Algeria 6.5 28.8 Syria 6.7 16.8 Lebanon 6.0 38.6 Jordan 5.5 8.5 Libya 5.0 28.8 Morocco 4.2 5.5 Subtotal 81.5 Subtotal 88.7 Percent in total Percent in total 3.8 2.6 migrant stocks migrant stocks Sources: Docquier and Marfouk 2005; authors’ calculations. - 59 - Table A5. Seven Main Nationalities of Migrants from MENA Countries in Selected OECD Countries of Continental Europe Share in total Share in total Seven main stocks of Share of which Seven main stocks of Share of which MENA sending migrants from are poorly MENA sending migrants from are poorly countries MENA countries educated (%) countries MENA countries educated (%) (%) (%) Belgium France Morocco 74.1 75.5 Algeria 42.9 84.1 Algeria 11.6 75.5 Morocco 35.6 79.3 Tunisia 6.2 75.5 Tunisia 15.5 80.3 Lebanon 3.0 16.7 Lebanon 2.1 29.7 Iran, Islamic Rep. Egypt, Arab Rep. 2.7 14.7 1.3 31.3 of of Egypt, Arab Rep. Iran, Islamic Rep. 1.5 75.5 1.3 11.0 of of Syria 0.9 86.8 Syria 0.8 32.4 Subtotal 100.0 Subtotal 99.4 Percent in total Percent in total 14.6 41.6 migrant stocks migrant stocks Germany Netherlands Iran, Islamic Rep. 26.8 35.7 Morocco 70.7 78.9 of Morocco 20.7 44.3 Iraq 11.3 46.8 Iran, Islamic Rep. Iraq 15.1 35.7 7.0 46.8 of Egypt, Arab Rep. Lebanon 12.8 35.7 4.2 55.0 of Syria 6.4 35.7 Algeria 1.7 55.0 Tunisia 6.2 44.3 Tunisia 1.7 55.0 Algeria 4.3 44.3 Syria 1.3 46.8 Subtotal 92.3 Subtotal 97.8 Percent in total Percent in total n.d. 14.8 migrant stocks migrant stocks Sources: Docquier and Marfouk 2005; authors’ calculations. Note: n.d. = no data - 60 - Table A6. Seven Main Nationalities of Migrants from MENA Countries in Selected OECD Countries of Southern Europe Italy Spain Share in total Share in total Share of which Share of which Seven main stocks of Seven main stocks of are poorly are poorly MENA sending migrants from MENA sending migrants from educated educated countries MENA countries countries MENA countries (%) (%) (%) (%) Morocco 61.6 75.5 Morocco 88.6 66.7 Tunisia 18.3 78.1 Algeria 8.1 43.3 Egypt, Arab Rep. Iran, Islamic Rep. of 10.3 31.7 of 1.0 15.6 Egypt, Arab Rep. Algeria 4.6 60.8 of 0.6 23.8 Iran, Islamic Rep. Syria of 2.5 14.7 0.4 21.9 Lebanon 1.1 26.5 Lebanon 0.3 18.8 Syria 0.8 36.8 Tunisia 0.2 27.8 Subtotal 99.2 Subtotal 99.4 Percent in total Percent in total migrant stocks 14.5 migrant stocks 16.8 Sources: Docquier and Marfouk 2005; authors’ calculations. - 61 - Table A7. Share of Migrants from MENA Countries in the Stocks of Highly Educated Migrants, by OECD Country Share of highly Share of highly educated migrants Total stock of highly Total stock of highly educated migrants from MENA educated migrants educated migrants from MENA countries in the from MENA (all countries of countries in the total stock of highly countries origin) total stock of educated migrants (1) (2) migrants from (%) MENA countries (%) (1)/(2) EEA/EU Austria 7,594 104,805 25.9 7.2 Belgium 17,872 176,960 12.0 10.1 Denmark 8,156 65,486 17.4 12.5 Finland 1,009 22,318 14.4 4.5 France 402,352 1,011,424 17.3 39.8 Germanya 19,500 1,974,500 19.7 1.0 Greece 12,133 153,111 23.8 7.9 Hungary 1,136 54,465 42.6 2.1 Ireland 2,144 128,762 50.7 1.7 Italy 28,600 246,925 9.8 11.6 Luxemburg 592 23,951 32.2 2.5 Netherlandsa 20,006 208,863 10.1 9.6 Polanda 1,466 92,078 53.6 1.6 Portugal 542 113,348 26.6 0.5 Slovak Republica 205 40,617 56.0 0.5 Spain 35,519 404,387 11.4 8.8 Sweden 36,536 207,625 23.6 17.6 Czech Republica 1,188 64,048 52.9 1.9 United Kingdom 72,656 1,374,370 41.9 5.3 EEA/non-EU Norway 4,512 65,535 15.3 6.9 Other OECD Australiaa 46,798 1,542,415 29.1 3.0 Canada 135,410 2,033,490 47.9 6.7 Japan 1,014 279,610 24.7 0.4 Korea, Rep. ofa 45,355 0.0 Mexicoa 891 89,689 41.1 1.0 New Zealanda 2,997 172,407 35.2 1.7 Switzerlanda 12,850 286,685 32.3 4.5 Turkey 8,159 161,869 31.0 5.0 United States 375,729 8,204,473 46.4 4.6 Sources: Dumont and Lemaître 2005; authors’ calculations. Note: MENA countries according to World Bank classification. a. Figures to be treated with caution due to the large share of immigrants with unknown country of birth. - 62 - Table A8. Share of Migrants from MENA Countries in the Migratory Flows to Selected OECD Countries 1995 1996 1997 1998 1999 2000 2001 2002 EEA Austria Flows from MENA countries - 2,703 2,717 2,904 4,501 4,530 1,900 - Share in total flows (%) - 4.7 4.8 4.9 6.2 6.9 2.5 - Denmark Flows from MENA countries 1,936 2,509 2,543 3,729 2,370 3,390 3,421 2,693 Share in total flows (%) 4.9 8.0 9.3 12.9 8.5 11.0 10.2 8.8 Finland Flows from MENA countries - - 996 598 751 531 698 - Share in total flows (%) - - 12.2 7.2 9.5 5.8 6.3 - Germany Flows from MENA countries 21,066 40,004 37,809 28,352 35,343 42,823 47,830 35,845 Share in total flows (%) 2.7 5.7 6.1 4.7 5.2 6.6 7.0 5.4 Netherlands Flows from MENA countries 8,549 14,936 10,067 16,966 7,704 13,915 13,883 9,882 Share in total flows (%) 11.4 17.3 11.6 17.6 8.2 12.8 12.6 9.9 Norway Flows from MENA countries - - - - 3,206 5,548 2,534 - Share in total flows (%) - - - - 9.9 20.0 10.0 - Sweden Flows from MENA countries 4,850 5,348 7,249 8,799 8,521 8,591 10,146 11,272 Share in total flows (%) 12.8 17.1 20.5 23.1 23.1 19.0 21.6 22.2 United Kingdom Flows from MENA countries 1,822 1,750 3,121 2,565 3,187 7,895 1,675 - Share in total flows (%) 3.3 2.8 5.3 3.7 3.3 6.3 1.6 - Other OECD Australia Flows from MENA countries 6,353 5,120 4,725 2,751 4,916 5,364 3,196 7,025 Share in total flows (%) 6.4 6.0 6.1 3.3 5.3 5.0 3.6 7.5 United States Flows from MENA countries 39,405 46,213 37,631 23,999 29,793 40,473 47,222 43,079 Share in total flows (%) 5.5 n.a. 4.7 3.7 4.6 4.8 4.4 4.0 Source: Migration Information Source 2006 - 63 - Table A9. Number of Algerian Migrants Living in OECD Countries, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 116 39 51 26 488 180 194 114 Belgium 9,597 1,200 1,442 718 10,864 8,201 1,750 913 Denmark 368 80 135 62 649 138 256 146 Finland 187 120 49 18 410 181 141 88 France 460,186 430,520 10,636 13,588 511,971 430,440 29,565 51,966 Germany 3,758 1,589 387 697 8,932 3,956 1,019 2,458 Greece 204 54 94 57 171 82 73 17 Hungary 236 140 35 36 295 84 86 125 Iceland - - - - 25 6 10 7 Ireland - - - - - - - - Italy 3,654 2,167 538 558 8,111 4,931 2,273 907 Luxembourg - - - - - - - - Netherlands 1,875 1,112 276 286 3,610 1,984 913 713 Poland 362 215 53 55 406 91 132 175 Portugal - - - - - - - - Slovak Republic 19 4 3 12 Spain 7,121 4,210 1,885 1,025 19,856 8,600 9,299 1,958 Sweden 1,190 281 437 332 1,491 328 552 507 Czech Republic 119 70 17 18 288 27 138 115 United Kingdom 2,820 1,694 282 844 8,949 2,574 2,555 3,821 EEA/non-EU Norway 357 10 138 81 582 12 336 150 Total EEA 492,151 443,500 16,456 18,403 577,117 461,819 49,294 64,190 Other OECD Australia 618 177 153 288 891 135 253 503 Canada 3,435 620 360 2,455 14,035 1,315 915 11,805 Japan - - - - 75 26 22 26 Korea, Rep. of - - - - - - - - Mexico 101 25 25 50 101 - 15 86 New Zealand 33 3 12 18 99 3 33 51 Switzerland 4,730 199 3,045 1,234 5,824 291 3,098 1,824 Turkey - - - - - - - - United States 4,433 243 1,158 3,032 9,682 439 2,190 7,053 Total Other OECD 13,350 1,267 4,753 7,077 30,707 2,209 6,526 21,348 Total 505,501 444,767 21,209 25,480 607,824 464,028 55,820 85,537 Source: Docquier and Marfouk 2005. Table A10. Main Destination Countries of Algerians within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. France 84.2 10.2 5.8 84.1 2. Spain 3.3 9.9 46.8 43.3 3. Canada 2.3 84.1 6.5 9.4 4. Belgium 1.8 8.4 16.1 75.5 5. United States 1.6 72.8 22.6 4.5 Total 93.2 Sources: Docquier and Marfouk 2005; authors’ calculations. - 64 - Table A11. Number of Moroccan Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 187 96 68 23 767 333 286 148 Belgium 61,385 6,603 7,319 3,072 69,485 52,453 11,193 5,839 Denmark 1,566 427 381 163 3,003 875 877 407 Finland 281 226 50 5 942 496 330 116 France 328,499 284,702 13,248 19,753 425,096 337,051 33,369 54,676 Germany 35,412 14,968 3,651 6,571 42,678 18,904 4,869 11,744 Greece 181 48 83 51 352 169 150 34 Hungary 32 19 5 5 40 11 12 17 Iceland 29 8 11 6 90 20 37 24 Ireland - - - - - - - - Italy 54,085 32,073 7,959 8,256 108,253 81,755 19,960 6,538 Luxembourg - - - - - - - - Netherlands 92,126 54,632 13,557 14,062 147,699 116,474 23,629 7,596 Poland - - - - - - - - Portugal - - - - - - - - Slovak Republic - - - - 9 2 - 7 Spain 107,961 92,460 11,408 4,093 216,470 144,434 61,969 10,066 Sweden 3,146 1,039 1,064 549 4,072 1,303 1,547 896 Czech Republic 37 22 6 6 88 9 42 34 United Kingdom 6,460 3,879 647 1,934 10,903 3,136 3,112 4,655 EEA/non-EU Norway 1,745 98 578 192 3,082 236 1,603 454 Total EEA 693,134 491,300 60,033 58,741 1,033,030 757,662 162,986 103,250 Other OECD Australia 881 204 209 468 1,075 205 316 554 Canada 14,440 3,445 2,130 8,865 21,720 3,585 2,450 15,685 Japan - - - - 150 51 43 52 Korea, Rep. of - - - - - - - - Mexico 328 126 126 76 323 94 105 121 New Zealand 48 6 12 27 117 15 36 51 Switzerland 5,661 215 3,747 1,362 9,082 478 5,476 2,310 Turkey - - - - - - - - United States 16,879 989 4,998 10,892 29,670 1,625 8,900 19,145 Total other OECD 38,237 4,985 11,222 21,690 62,137 6,054 17,327 37,918 Total 731,370 496,285 71,255 80,431 1,095,166 763,716 180,313 141,168 Source: Docquier and Marfouk 2005. - 65 - Table A12. Main Destination Countries of Moroccans within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. France 38.8 12.9 7.8 79.3 2. Spain 19.8 4.7 28.6 66.7 3. Netherlands 13.5 5.1 16.0 78.9 4. Italy 9.9 6.0 18.4 75.5 5. Belgium 6.3 8.4 16.1 75.5 6. Germany 3.9 27.5 11.4 44.3 7. United States 2.7 64.5 30.0 5.5 Total 94.9 Sources: Docquier and Marfouk 2005; authors’ calculations. - 66 - Table A13. Number of Tunisian Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 570 302 235 33 1,857 922 781 154 Belgium 5,167 530 1,203 725 5,848 4,415 942 491 Denmark 273 43 110 47 527 107 233 102 Finland 114 87 22 5 272 172 78 22 France 180,670 154,817 10,160 12,580 184,603 148,243 13,726 22,634 Germany 13,291 5,618 1,370 2,466 12,833 5,685 1,464 3,531 Greece 126 33 58 35 266 128 113 26 Hungary 22 13 3 3 27 8 8 11 Iceland - - - - 13 3 5 3 Ireland - - - - - - - - Italy 18,536 10,992 2,728 2,829 32,227 25,161 5,843 1,223 Luxembourg - - - - - - - - Netherlands 2,192 1,300 323 335 3,520 1,934 890 695 Poland - - - - - - - - Portugal - - - - - - - - Slovak Republic - - - - 13 4 2 6 Spain 566 258 156 153 605 168 307 130 Sweden 2,144 650 797 327 2,489 747 1,095 498 Czech Republic 55 33 8 8 130 6 57 65 United Kingdom 2,000 1,201 200 599 2,641 760 754 1,128 EEA/non-EU Norway 289 24 106 52 447 36 242 98 Total EEA 226,014 175,900 17,478 20,198 248,319 188,497 26,541 30,817 Other OECD Australia 407 160 82 165 401 123 134 144 Canada 2,305 430 235 1,640 4,260 535 250 3,475 Japan - - - - 98 34 29 34 Korea, Rep. of - - - - - - - - Mexico 50 - - 50 26 - 26 - New Zealand 18 3 - 15 33 - 9 12 Switzerland 4,214 143 2,955 900 5,443 265 3,325 1,328 Turkey - - - - - - - - United States 3,616 300 1,046 2,270 5,555 390 1,625 3,540 Total other OECD 10,610 1,036 4,318 5,040 15,817 1,347 5,398 8,533 Total 236,625 176,936 21,796 25,238 264,135 189,844 31,939 39,350 Source: Docquier and Marfouk 2005. Table A14. Main Destination Countries of Tunisians within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. France 69.9 12.3 7.4 80.3 2. Italy 12.2 3.8 18.1 78.1 3. Germany 4.9 27.5 11.4 44.3 4. Belgium 2.2 8.4 16.1 75.5 5. United States 2.1 63.7 29.3 7.0 6. Switzerland 2.1 24.4 61.1 4.9 Total 93.3 Sources: Docquier and Marfouk 2005; authors’ calculations. - 67 - Table A15. Number of Egyptian Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 3,663 1,643 1,188 832 7,905 3,113 2,278 2,514 Belgium 1,263 43 255 402 1,431 1,080 231 120 Denmark 611 48 187 166 880 88 329 288 Finland 169 111 21 37 373 156 86 131 France 13,776 5,096 2,996 5,280 15,576 4,873 3,701 7,002 Germany 4,968 2,100 512 922 7,457 3,303 851 2,052 Greece 2,969 778 1,362 828 5,233 2,505 2,224 505 Hungary 175 104 26 27 218 62 63 92 Iceland - - - - 16 4 7 4 Ireland - - - - - - - - Italy 11,192 6,637 1,647 1,708 18,076 5,723 7,812 4,541 Luxembourg - - - - - - - - Netherlands 5,063 3,003 745 773 8,819 4,847 2,231 1,741 Poland - - - - - - - - Portugal - - - - - - - - Slovak Republic - - - - 19 5 2 12 Spain 899 311 225 363 1,450 346 631 474 Sweden 1,538 174 413 752 1,852 194 528 991 Czech Republic 63 38 9 10 145 9 37 98 United Kingdom 19,470 11,692 1,949 5,829 19,557 5,625 5,583 8,349 EEA/non-EU Norway 235 2 67 93 306 2 105 166 Total EEA 66,055 31,780 11,603 18,021 89,314 31,936 26,697 29,081 Other OECD Australia 30,444 6,296 7,356 16,792 30,668 5,633 7,931 17,104 Canada 25,500 3,105 3,545 18,850 31,025 2,905 3,205 24,915 Japan 368 144 96 113 580 198 168 200 Korea, Rep. of - - - - - - - - Mexico 403 126 126 151 329 59 108 162 New Zealand 390 66 105 201 969 48 219 588 Switzerland 4,772 97 2,428 1,983 4,316 85 1,710 2,211 Turkey 1,523 983 255 235 - - - - United States 59,216 2,621 12,881 43,714 96,660 3,480 18,010 75,170 Total Other OECD 122,616 13,438 26,792 82,039 164,547 12,409 31,351 120,351 Total 188,671 45,218 38,394 100,060 253,861 44,344 58,048 149,432 Source: Docquier and Marfouk 2005. Table A16. Main Destination Countries of Egyptians within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 38.1 77.8 18.6 3.6 2. Canada 12.2 80.3 10.3 9.4 3. Australia 12.1 55.8 25.9 18.4 4. United Kingdom 7.7 42.7 28.5 28.8 5. Italy 7.1 25.1 43.2 31.7 6. France 6.1 45.0 23.8 31.3 Total 77.2 Sources: Docquier and Marfouk 2005; authors’ calculations. - 68 - Table A17. Number of Jordanian Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 123 32 51 40 316 97 95 124 Belgium 236 16 53 47 0 0 0 0 Denmark 321 80 115 52 582 141 214 120 Finland 41 29 6 6 112 53 24 35 France 512 112 84 244 639 154 118 367 Germany 5,645 2,209 532 1,595 5,595 2,000 510 2,196 Greece 563 151 251 162 967 371 408 188 Hungary 97 57 14 15 121 49 34 39 Iceland 0 0 0 0 13 3 5 3 Ireland 0 0 0 0 0 0 0 0 Italy 1,059 628 156 162 1,329 191 586 552 Luxembourg 0 0 0 0 0 0 0 0 Netherlands 480 285 71 73 463 217 134 113 Poland 0 0 0 0 0 0 0 0 Portugal 0 0 0 0 0 0 0 0 Slovak Republic 12 0 2 10 Spain 0 0 0 0 543 75 245 223 Sweden 622 147 226 208 828 166 323 298 Czech Republic 35 21 5 5 78 0 20 58 United Kingdom 1,620 1,034 172 414 1,983 765 513 705 EEA/non-EU Norway 46 2 12 20 54 3 25 23 Total EEA 11,401 4,803 1,748 3,043 13,635 4,283 3,256 5,054 Other OECD Australia 1,590 283 358 949 2,428 339 568 1,521 Canada 1,590 410 250 930 3,265 615 345 2,305 Japan 0 0 0 0 60 21 17 21 Korea, Rep. of 0 0 0 0 0 0 0 0 Mexico 25 0 0 25 20 7 0 13 New Zealand 27 3 3 21 132 9 30 81 Switzerland 157 10 72 50 286 11 131 118 Turkey 0 0 0 0 0 0 0 0 United States 21,990 2,863 7,449 11,678 39,140 3,320 12,165 23,655 Total other OECD 25,379 3,569 8,132 13,653 45,331 4,321 13,256 27,714 Total 36,780 8,372 9,880 16,697 58,966 8,605 16,513 32,768 Source: Docquier and Marfouk 2005. Table A18. Main Destination Countries of Jordanian Migrants within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 29.3 43.1 37.0 19.9 2. Germany 12.7 39.3 9.1 35.7 3. Sweden 12.6 35.0 28.0 24.0 4. Netherlands 9.6 24.3 28.9 46.8 5. United Kingdom 9.6 35.6 25.9 38.6 6. Canada 7.6 55.4 14.1 30.4 7. Australia 7.1 46.8 14.2 39.0 Total 88.4 Sources: Docquier and Marfouk 2005; authors’ calculations. - 69 - Table A19. Number of Lebanese Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 320 144 123 53 955 393 327 235 Belgium 1,859 59 348 625 2,783 466 1,077 1,241 Denmark 2,895 990 597 297 6,772 2,748 1,670 586 Finland 83 53 18 12 230 116 64 50 France 20,900 6,865 3,296 9,001 25,155 7,483 4,311 13,361 Germany 22,379 8,757 2,108 6,325 26,320 9,409 2,398 10,331 Greece 1,315 352 585 378 1,281 241 703 338 Hungary 96 57 14 15 120 48 34 38 Iceland - - - - 20 5 8 5 Ireland - - - - - - - - Italy 1,395 827 205 213 1,869 496 717 656 Luxembourg - - - - - - - - Netherlands 1,812 1,075 267 277 1,960 917 566 477 Poland 273 162 40 42 306 47 75 165 Portugal - - - - - - - - Slovak Republic - - - - 29 7 2 19 Spain - - - - 751 141 366 244 Sweden 11,842 5,986 2,738 1,530 14,279 5,997 5,140 2,285 Czech Republic 74 44 11 11 163 6 51 104 United Kingdom 5,760 3,675 612 1,473 8,245 3,179 2,134 2,932 EEA/non-EU Norway 402 23 118 75 872 65 486 189 Total EEA 71,406 29,069 11,081 20,325 92,110 31,763 20,129 33,256 Other OECD Australia 52,380 22,399 13,027 16,954 63,446 25,962 19,920 17,564 Canada 37,610 13,525 5,390 18,695 55,615 17,590 7,545 30,480 Japan - - - - 28 10 8 10 Korea, Rep. of - - - - - - - - Mexico 3,174 1,864 479 831 2,741 907 1,001 676 New Zealand 342 81 96 135 384 54 147 129 Switzerland 8,194 924 4,789 1,144 3,779 241 1,922 1,309 Turkey - - - - - - - - United States 67,214 11,033 20,954 35,227 92,685 10,605 27,290 54,790 Total other OECD 168,914 49,826 44,735 72,986 218,679 55,368 57,833 104,958 Total 240,320 78,895 55,816 93,312 310,789 87,131 77,963 138,214 Source: Docquier and Marfouk 2005. Table A20. Main Destination Countries of Lebanese within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 66,4 60,4 31,1 8,5 2. Germany 9,5 39,3 9,1 35,7 3. Canada 5,5 70,6 10,6 18,8 4. Australia 4,1 62,6 23,4 14,0 5. United Kingdom 3,4 35,6 25,9 38,6 6. Italy 2,3 41,5 44,1 14,4 Total 91.1 Sources: Docquier and Marfouk 2005; authors’ calculations. - 70 - Table A21. Number of Syrian Migrants living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 338 153 123 62 1,346 527 349 470 Belgium 910 64 178 234 858 745 113 - Denmark 226 53 50 45 744 191 239 183 Finland 36 18 10 8 137 60 29 48 France 8,900 3,112 1,024 3,504 9,354 3,028 1,296 5,030 Germany 6,888 2,695 649 1,947 13,311 4,758 1,213 5,225 Greece 1,491 399 663 428 1,811 695 765 352 Hungary 431 256 63 66 538 216 151 172 Iceland - - - - 15 3 6 4 Ireland - - - - - - - - Italy 837 496 123 128 1,410 519 495 396 Luxembourg - - - - - - - - Netherlands 2,239 1,328 330 342 2,629 1,230 759 640 Poland 448 266 66 68 502 45 99 348 Portugal - - - - - - - - Slovak Republic - - - - 92 12 9 70 Spain - - - - 1,048 229 502 317 Sweden 5,959 2,772 1,361 1,046 10,837 4,335 3,034 2,059 Czech Republic 182 108 27 28 378 14 96 262 United Kingdom 1,760 1,123 187 450 3,192 1,231 826 1,135 EEA/non-EU Norway 133 12 34 38 373 19 206 102 Total EEA 30,778 12,855 4,888 8,393 48,575 17,856 10,187 16,812 Other OECD Australia 4,270 1,681 843 1,746 5,805 2,327 1,277 2,201 Canada 8,530 3,130 1,250 4,150 12,745 3,755 1,685 7,305 Japan - - - - 66 23 19 23 Korea, Rep. of - - - - - - - - Mexico 1,083 831 101 151 727 327 187 191 New Zealand 24 6 6 12 189 30 51 72 Switzerland 1,239 92 640 349 1,357 87 646 472 Turkey 2,919 1,510 729 660 - - - - United States 30,449 6,598 9,178 14,673 47,660 7,990 14,895 24,775 Total other OECD 48,514 13,848 12,747 21,741 68,549 14,539 18,760 35,039 Total 79,292 26,703 17,635 30,134 117,124 32,395 28,947 51,851 Source: Docquier and Marfouk 2005. - 71 - Table A22. Main Destination Countries of Syrians within OECD Countries Composition by level of education (%) Main destination Share in total stocks countries within OECD (%) Tertiary Primary Secondary education education education 1. United States 40.7 52.0 31.3 16.8 2. Germany 11.4 39.3 9.1 35.7 3. Canada 10.9 57.3 13.2 29.5 4. Sweden 9.3 19.0 28.0 40.0 5. France 8.0 53.8 13.9 32.4 6. Australia 5.0 37.9 22.0 40.1 7. United Kingdom 2.7 35.6 25.9 38.6 8. Netherlands 2.2 24.3 28.9 46.8 Total 90.1 Sources: Docquier and Marfouk 2005; authors’ calculations. - 72 - Table A23. Migrant Stocks from GCC States in OECD countries Saudi United Arab Bahrain Kuwait Oman Qatar Arabia Emirates EEA/EU Austria 5 123 6 14 100 24 Belgium 0 0 0 0 0 0 Denmark 27 424 0 5 42 12 Finland 0 32 0 1 7 5 France 24 307 5 51 433 90 Germany 0 0 0 0 396 0 Greece 6 25 3 15 126 13 Hungary 0 60 2 3 16 29 Iceland 0 0 0 0 0 0 Ireland 0 0 0 0 0 0 Italy 0 0 0 0 0 0 Luxembourg 0 0 0 0 0 0 Netherlands 36 814 110 19 376 145 Poland 0 0 0 0 0 0 Portugal 0 0 0 0 0 0 Slovak Republic 2 11 0 1 5 7 Spain 5 49 7 12 63 14 Sweden 30 412 4 12 164 53 Czech Republic 5 27 0 24 6 12 United Kingdom 2,335 3,066 508 433 2,488 694 EEA/non-EU Norway 13 64 0 0 28 11 Total EEA 2,487 5,413 645 589 4,250 1,110 of which highly educated 35.1% 33.1% 33.3% 31.0% 35.6% 34.2% Other OECD Australia 314 1,289 63 93 320 148 Canada 410 3,565 40 215 1,165 290 Japan 3 11 7 5 76 9 Korea, Rep. of 0 0 0 0 0 0 Mexico 0 0 0 0 356 0 New Zealand 138 315 18 21 108 189 Switzerland 59 168 13 9 324 79 Turkey 0 0 0 0 0 0 United States 1,227 12,505 476 688 10,028 1,322 Total other OECD 2,151 17,853 617 1,031 12,376 2,037 of which highly educated 68.8% 78.4% 93.4% 91.7% 74.5% 85.4% Total 4,638 23,266 1,261 1,621 16,626 3,146 of which highly educated 49.8% 64.8% 60.0% 67.9% 60.8% 63.1% Sources: Docquier and Marfouk 2005; authors’ calculations. - 73 - Table A24. Number of Iranian Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 3,711 1,436 1,686 589 9,282 3,471 3,087 2,724 Belgium 2,612 75 668 643 2,529 371 339 1,819 Denmark 4,822 272 965 550 5,601 601 1,935 1,067 Finland 314 239 41 34 1,423 759 429 235 France 13,369 2,192 2,712 7,036 15,249 1,670 3,419 10,160 Germany 59,000 22,000 6,000 31,000 55,292 19,766 5,038 21,704 Greece 618 265 231 122 905 496 262 148 Hungary 172 102 25 26 215 86 60 69 Iceland - - - - 19 4 8 5 Ireland - - - - - - - - Italy 3,152 1,869 464 481 4,471 656 1,884 1,931 Luxembourg - - - - - - - - Netherlands 8,193 4,858 1,206 1,251 14,586 6,824 4,213 3,549 Poland - - - - - - - - Portugal - - - - - - - - Slovak Republic - - - - 26 5 6 15 Spain - - - - 2,534 395 1,341 797 Sweden 32,134 5,274 14,275 9,848 38,478 4,617 16,930 15,006 Czech Republic 76 45 11 12 172 5 62 104 United Kingdom 23,170 12,494 2,082 8,594 33,619 6,429 15,632 11,558 EEA/non-EU Norway 3,458 69 1,410 898 6,442 133 3,538 2,117 Total EEA 154,802 51,191 31,777 61,083 190,844 46,289 58,183 73,008 Other OECD countries Australia 9,073 1,532 930 6,611 14,326 2,493 2,444 9,389 Canada 21,950 2,785 3,390 15,775 53,150 5,375 6,745 41,030 Japan 1,237 485 321 380 3,977 1,361 1,153 1,375 Korea, Rep. of - - - - - - - - Mexico 101 - - 101 184 46 22 116 New Zealand 720 36 207 423 1,725 126 621 699 Switzerland 5,004 228 2,483 1,530 5,027 187 2,317 2,112 Turkey 6,074 1,825 2,243 1,824 7,650 2,760 2,195 2,625 United States 164,283 10,837 36,925 116,521 250,785 16,910 55,475 178,400 Total other OECD 208,442 17,728 46,499 143,165 336,824 29,258 70,972 235,746 Total 363,243 68,919 78,276 204,248 527,668 75,547 129,155 308,754 Source: Docquier and Marfouk 2005. - 74 - Table A25. Main Destination Countries Iranians within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 47.5 71.1 22.1 6.7 2. Germany 10.5 39.3 9.1 35.7 3. Canada 10.1 77.2 12.7 10.1 4. Sweden 7.3 39.0 44.0 12.0 5. United Kingdom 7.3 34.4 46.5 19.1 6. France 2.9 66.6 22.4 11.0 7. Netherlands 2.8 24.3 28.9 46.8 8. Australia 2.7 65.5 17.1 17.4 Total 91.0 Sources: Docquier and Marfouk 2005; authors’ calculations. - 75 - Table A26. Number of Iraqi Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 287 119 81 87 2,245 944 575 726 Belgium 261 7 48 86 - - - - Denmark 1,427 228 366 319 7,352 1,570 2,119 1,650 Finland 128 90 24 14 1,697 1,071 377 249 France 1,776 488 228 872 2,292 776 346 1,170 Germany 2,768 1,083 261 782 31,206 11,156 2,844 12,249 Greece 1,311 930 277 104 2,373 1,454 785 134 Hungary 245 145 36 37 306 123 86 98 Iceland - - - - 16 4 7 4 Ireland - - - - - - - - Italy 418 248 62 64 - - - - Luxembourg - - - - - - - - Netherlands 6,004 3,560 883 916 23,615 11,049 6,821 5,745 Poland 341 202 50 52 382 36 61 268 Portugal - - - - - - - - Slovak Republic - - - - 24 2 2 19 Spain - - - - 472 93 194 185 Sweden 14,861 3,498 3,876 5,094 31,121 7,469 8,714 10,892 Czech Republic 100 59 15 15 235 8 39 187 United Kingdom 11,500 7,337 1,223 2,940 23,559 9,083 6,098 8,378 EEA/non-EU Norway 510 33 111 164 4,121 265 2,172 1,037 Total EEA 41,937 18,028 7,540 11,548 131,016 45,101 31,238 42,992 Other OECD Australia 4,346 1,246 793 2,307 17,383 6,772 2,476 8,135 Canada 5,610 1,440 800 3,370 18,755 5,710 2,650 10,395 Japan - - - - 32 11 9 11 Korea, Rep. of - - - - - - - - Mexico 25 25 - - 149 50 40 59 New Zealand 342 144 78 102 3,810 609 945 1,704 Switzerland 690 50 317 211 3,038 405 1,402 659 Turkey 2,596 983 574 940 - - - - United States 35,553 7,378 12,398 15,777 72,245 14,370 26,745 31,130 Total other OECD 49,162 11,266 14,960 22,707 115,413 27,928 34,267 52,093 Total 91,099 29,294 22,500 34,255 246,429 73,029 65,505 95,086 Source: Docquier and Marfouk 2005. Table A27. Main Destination Countries of Iraqi Migrants within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 29.3 43.1 37.0 19.9 2. Germany 12.7 39.3 9.1 35.7 3. Sweden 12.6 35.0 28.0 24.0 4. Netherlands 9.6 24.3 28.9 46.8 5. United Kingdom 9.6 35.6 25.9 38.6 6. Canada 7.6 55.4 14.1 30.4 7. Australia 7.1 46.8 14.2 39.0 Total 88.4 Sources: Docquier and Marfouk 2005; authors’ calculations. - 76 - Table A28. Number of Libyan Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 0 0 0 0 182 85 41 56 Belgium 178 26 32 32 0 0 0 0 Denmark 28 5 9 4 70 18 25 12 Finland 15 10 5 0 44 23 7 14 France 732 528 88 104 836 435 113 288 Germany 614 259 63 114 1,484 657 169 408 Greece 267 70 123 74 484 232 206 47 Hungary 91 54 13 14 114 33 33 48 Iceland 0 0 0 0 0 0 0 0 Ireland 0 0 0 0 0 0 0 0 Italy 773 458 114 118 0 0 0 0 Luxembourg 0 0 0 0 0 0 0 0 Netherlands 178 105 26 27 352 193 89 69 Poland 0 0 0 0 0 0 0 0 Portugal 0 0 0 0 0 0 0 0 Slovak Republic 12 5 0 6 Spain 0 0 0 0 181 27 69 85 Sweden 0 0 0 0 207 43 77 70 Czech Republic 29 17 4 4 63 3 19 41 United Kingdom 4,590 2,756 459 1,374 6,856 1,972 1,957 2,927 EEA/non-EU Norway 31 2 10 5 45 2 23 16 Total EEA 7,526 4,292 947 1,871 10,929 3,728 2,828 4,088 Other OECD Australia 1,103 473 290 340 1,274 442 423 409 Canada 385 60 40 285 1,250 115 90 1,045 Japan 0 0 0 0 8 3 2 3 Korea, Rep. of 0 0 0 0 0 0 0 0 Mexico 76 25 50 0 15 0 0 15 New Zealand 42 6 6 30 78 3 24 42 Switzerland 420 17 247 88 565 35 267 184 Turkey 0 0 0 0 0 0 0 0 United States 5,388 116 1,542 3,730 7,024 149 1,220 5,655 Total other OECD 7,414 697 2,175 4,473 10,215 747 2,026 7,353 Total 14,939 4,989 3,122 6,344 21,144 4,475 4,854 11,441 Source: Docquier and Marfouk 2005. Table A29. List of main destination countries of Libyans within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 33.2 80.5 17.4 2.1 2. United Kingdom 32.4 42.7 28.5 28.8 3. Germany 7.0 27.5 11.4 44.3 4. Australia 6.0 32.1 33.2 34.7 5. Canada 5.9 83.6 7.2 9.2 6. France 4.0 34.4 13.5 52.0 7. Switzerland 2.7 32.6 47.3 6.2 Total 91.1 Sources: Docquier and Marfouk 2005; authors’ calculations. - 77 - Table A30. Number of Migrants from Occupied Palestinian Territory Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 0 0 0 0 295 76 90 129 Belgium 0 0 0 0 0 0 0 0 Denmark 24 1 9 7 0 0 0 0 Finland 0 0 0 0 0 0 0 0 France 0 0 0 0 619 210 78 331 Germany 0 0 0 0 0 0 0 0 Greece 0 0 0 0 0 0 0 0 Hungary 0 0 0 0 0 0 0 0 Iceland 0 0 0 0 0 0 0 0 Ireland 0 0 0 0 0 0 0 0 Italy 12 7 2 2 0 0 0 0 Luxembourg 0 0 0 0 0 0 0 0 Netherlands 0 0 0 0 0 0 0 0 Poland 0 0 0 0 0 0 0 0 Portugal 0 0 0 0 0 0 0 0 Slovak Republic 0 0 0 0 0 0 0 0 Spain 0 0 0 0 115 28 43 44 Sweden 785 269 202 198 941 292 282 301 Czech Republic 21 13 3 3 55 1 16 37 United Kingdom 1,800 1,148 191 460 1,989 767 515 707 EEA/non-EU Norway 0 0 0 0 307 9 137 96 Total EEA 2,642 1,438 407 670 4,321 1,383 1,161 1,645 Other OECD Australia 61 11 9 41 2,480 639 651 1,190 Canada 0 0 0 0 5,175 1,040 795 3,340 Japan 0 0 0 0 0 0 0 0 Korea, Rep. of 0 0 0 0 0 0 0 0 Mexico 0 0 0 0 0 0 0 0 New Zealand 0 0 0 0 105 3 18 69 Switzerland 174 6 85 53 304 21 148 111 Turkey 0 0 0 0 0 0 0 0 United States 18,112 2,967 6,088 9,057 20,275 1,917 6,737 11,621 Total Other OECD 18,347 2,984 6,182 9,151 28,339 3,620 8,349 16,331 Total 20,989 4,422 6,589 9,821 32,660 5,002 9,510 17,977 Source: Docquier and Marfouk 2005. - 78 - Table A31. Main Destination Countries of Palestinians within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD (%) Tertiary education Secondary education Primary education 1. United States 62.1 57.3 33.2 9.5 2. Canada 15.8 64.5 15.4 20.1 3. Australia 7.6 48.0 26.3 25.8 4. United Kingdom 6.1 35.6 25.9 38.6 5. Sweden 2.9 32.0 30.0 31.0 Total 94.5 Sources: Docquier and Marfouk 2005; authors’ calculations. - 79 - Table A32. Number of Yemenite Migrants Living in OECD Countries, by Education Level, 1990 and 2000 1990 2000 Total Primary Secondary Tertiary Total Primary Secondary Tertiary EEA/EU Austria 15 4 11 0 37 13 8 16 Belgium 14 0 1 1 0 0 0 0 Denmark 32 4 12 4 74 11 22 18 Finland 3 2 0 1 15 5 5 5 France 184 64 8 56 593 277 96 220 Germany 278 109 26 79 897 321 82 352 Greece 150 40 67 43 25 9 10 5 Hungary 87 51 13 13 108 43 30 34 Iceland 0 0 0 0 0 0 0 0 Ireland 0 0 0 0 0 0 0 0 Italy 44 26 6 7 0 0 0 0 Luxembourg 0 0 0 0 0 0 0 0 Netherlands 94 56 14 14 187 87 54 45 Poland 0 0 0 0 0 0 0 0 Portugal 0 0 0 0 0 0 0 0 Slovak Republic 0 0 0 0 15 2 1 12 Spain 0 0 0 0 12 1 2 9 Sweden 0 0 0 0 114 24 34 39 Czech Republic 44 26 6 7 98 4 19 75 United Kingdom 2,500 1,595 266 639 5,927 2,285 1,534 2,108 EEA/non-EU Norway 0 0 0 0 34 2 21 6 Total EEA 3,443 1,977 430 864 8,135 3,085 1,918 2,944 Other OECD Australia 189 16 53 120 303 31 73 199 Canada 355 60 40 255 700 80 85 535 Japan 0 0 0 0 13 4 4 4 Korea, Rep. of 0 0 0 0 0 0 0 0 Mexico 0 0 0 0 0 0 0 0 New Zealand 27 3 9 9 48 3 15 27 Switzerland 36 4 18 8 141 22 63 31 Turkey 0 0 0 0 0 0 0 0 United States 4,090 1,469 1,628 993 11,609 3,612 4,520 3,477 Total other OECD 4,697 1,552 1,748 1,385 12,814 3,752 4,760 4,273 Total 8,140 3,529 2,178 2,249 20,949 6,837 6,678 7,218 Source: Docquier and Marfouk 2005. Table A33. Main Destination Countries of Yemenite within OECD Countries Share in total stocks Composition by level of education (%) Main destination countries within OECD Tertiary education Secondary education Primary education 1. United States 55.4 30.0 38.9 31.1 2. United Kingdom 28.3 35.6 25.9 38.6 3. Germany 4.3 39.3 9.1 35.7 4. Canada 3.3 76.4 12.1 11.4 5. France 2.8 37.1 16.2 46.7 6. Australia 1.4 65.7 24.1 10.2 95.6 Sources: Docquier and Marfouk 2005; authors’ calculations. - 80 - ANNEX B. ESTIMATION RESULTS Table B1. List of Sending and Receiving Countries Included in the Econometric Analysis Sending Countries Caribbean & Sub-Saharan Africa Asia Eastern Europe MENA South America a a Angola Afghanistan Argentina Albania Algeria Benin Armenia Barbadosa Azerbaijana Bahrain Botswana Bangladesh Belize Belarus Djiboutia Burkina Faso Buthan Bolivia Bosnia & Herze. Egypt, Arab Rep. of Burundi Cambodia Brazil Bulgaria Iran, Islamic Rep. of Cameroon Chinaa Chile Croatia Iraqa Cape Verde Fijia Colombia Estonia Jordan Central African Rep. India Costa Rica Georgiaa Kuwait Chad Indonesia Cubaa Kazakhstana Lebanona Comoros Kazakhstan Dominicaa KyrgyzRepublic?a Libyaa Congo, Rep. of Kiribatia Dominican Republic Latvia Morocco Congo, Dem. Rep. ofa Lao PDR Ecuador Lithuania Occ. Palestinian Territorya Côte d'Ivoire Malaysia El Salvador Macedonia, FYRa Oman Equatorial Guineaa Maldives Grenadaa Moldova Qatara Eritreaa Mongolia Guatemala Romania Saudi Arabia Ethiopia Myanmara Guyanaa Russian Federation Syrian Arab Rep. Gabona Nepal Haiti Ukraine Tunisia Gambia, The Pakistan Honduras United Arab Emirates Ghana Papua New Guineaa Jamaica Yemen, Rep. of Guineaa Philippines Mexico Guinea-Bissaua Samoaa Nicaragua Kenya Solomon Islandsa Panama Lesotho Sri Lanka Paraguay Liberia Tajikistan Peru Madagascara Thailand St. Kitts and Nevisa Malawi Tongaa St. Luciaa Mali Turkmenistan St. Vincent & Grena.a Mauritania Uzbekistan Surinamea Mauritius Vanuatua Trinidad and Tobago Mozambique Vietnam Uruguay Namibia Venezuela, R. B. de Niger Nigeria Rwanda São Tomé and Principea Senegal Seychellesa Sierra Leonea Somaliaa South Africa Sudan Swaziland Tanzania Togo Uganda Zambia Zimbabwe - 81 - Table B1 (continued) Receiving OECD Countries Continental Northern Southern Anglo-Saxon Eastern Europe Other OECD countries Europe Europe Europe a Australia Austria Denmark Spain Czech Republic Japana a a Canada Belgium Finland Greece Hungary Korea, Rep. ofa a a a a a Ireland Luxembourg Iceland Italy Poland Mexicoa New Zealanda France Norway Portugala Slovak Republica Turkeya United Kingdom Germany Sweden United States Netherlands Switzerlanda Note:a. These countries are included only in the cross-sectional analysis but not in the panel data analysis. Table B2. Summary Statistics of Variables Used in the Econometric Analyses, 1990–2002 Whole sample MENA sample Mean Std. deviation Mean Std. deviation Dependent variables Total expatriation rate in 2000 0.0036 0.0310 0.0010 0.0044 Expatriation rate of the low educated in 2000 0.0017 0.0156 0.0006 0.0034 Expatriation rate of the medium educated in 2000 0.0057 0.0694 0.0008 0.0033 Expatriation rate of the high educated in 2000 0.0209 0.1785 0.0036 0.0156 Yearly emigration rate (per 1,000) 0.098 0.568 0.057 0.145 Independent variables Log (GDP per capita in PPP) in host 10.2 0.3 10.2 0.3 Log (GDP per capita in PPP) in origin 7.0 1.2 8.0 1.2 Population density in host 105.2 128.7 105.2 128.7 Population density in origin 104.1 158.8 110.9 197.2 Share of population age 15–24 in host 0.135 0.015 0.135 0.015 Share of population aged 15–24 in origin 0.191 0.021 0.193 0.023 Age dependency ratio in host 0.504 0.035 0.504 0.035 Age dependency ratio in origin 0.750 0.176 0.727 0.160 Urban population growth in host 0.008 0.006 0.008 0.006 Urban population growth in origin 0.031 0.023 0.036 0.016 Log(mean education) in host 2.3 0.2 2.3 0.2 Literacy rate in origin 0.731 0.220 0.690 0.151 Total unemployment rate in host 0.079 0.046 0.079 0.043 Unemployment rate of the low educated (1994) in host 0.115 0.052 0.115 0.052 Unemployment rate of the medium educated (1994) in host 0.081 0.038 0.081 0.038 Unemployment rate of the high educated (1994) in host 0.051 0.029 0.051 0.029 Private rate of return to education in host 0.074 0.021 0.074 0.021 Labor productivity growth in host 0.020 0.014 0.020 0.014 Share of public expenditure related 0.017 0.012 0.017 0.012 to unemployment compensation in host Political rights in origin (scale 1 to 7) 4.1 1.9 5.4 1.5 Civil liberties in origin (scale 1 to 7) 4.2 1.5 5.3 1.2 - 82 - Table B3. Cross-Sectional Gravity Model: Specification (1) Dependent Variable: Expatriation rate in 2000 (ratio of stocks, mijs , 2000 ) Total Total Primary Primary Secondary Secondary Tertiary Tertiary Other LDCs MENA Other LDCs MENA Other LDCs MENA Other LDCs MENA (1) (2) (3) (4) (5) (6) (7) (8) Bilateral characteristics Ln(distance) −0.0173*** −0.0016* −0.0058*** −0.0021** −0.0297*** −0.0012* −0.0927*** −0.0013 (5.08) (1.69) (3.79) (2.28) (4.68) (1.70) (5.24) (0.55) Common Border −0.0145* 0.0030 0.0060 0.0020 −0.0374*** 0.0067 −0.1353*** 0.0019 (1.95) (1.01) (0.92) (0.89) (3.08) (1.15) (4.57) (0.53) Common Language 0.0060 0.0025** 0.0006 0.0010 0.0101 0.0018** 0.0693*** 0.0086** (1.56) (2.47) (0.39) (1.27) (1.52) (2.42) (3.35) (2.47) Ever in colonial relationship 0.0318** 0.0076** 0.0252*** 0.0080** 0.0546 0.0031*** 0.0755*** 0.0209*** (2.25) (2.49) (2.70) (2.38) (1.62) (2.63) (2.76) (3.46) Island in couple (0, 1, or 2) −0.0029 0.0015 −0.0014 0.0005 −0.0055 0.0014 −0.0056 0.0056 (1.25) (0.73) (1.38) (0.27) (1.37) (0.76) (0.45) (1.34) Host country fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Origin country fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Constant 0.1497*** 0.0123* 0.0508*** 0.0160** 0.2568*** 0.0087* 0.7916*** 0.0104 (4.96) (1.72) (3.67) (2.30) (4.71) (1.67) (5.13) (0.60) Observations 3467 534 3467 534 3467 534 3467 534 R2 0.23 0.37 0.16 0.35 0.16 0.38 0.24 0.42 Note: Robust t-statistics to clustering are in parentheses. ***, **, and * mean, respectively, significant at the 1 percent, 5 percent, and 10 percent levels. - 83 - Table B4. Cross-Sectional Gravity Model – Specification (2) Dependent Variable: Expatriation rate in 2000 (ratio of stocks, mijs , 2000 ) Total Primary Secondary Tertiary Total Primary Secondary Tertiary Other Other Other Other MENA MENA MENA MENA LDCs LDCs LDCs LDCs (1) (2) (3) (4) (5) (6) (7) (8) Bilateral characteristics Ln(distance) −0.0204*** −0.0040*** −0.0066*** −0.0041** −0.0534*** −0.0024*** −0.1455*** −0.0087** (3.95) (2.65) (4.50) (2.58) (3.09) (2.75) (2.93) (2.45) Common Border −0.0070 0.0098*** 0.0116 0.0069*** −0.0569* 0.0258*** −0.2198** 0.0001 (0.36) (4.23) (0.62) (2.83) (1.74) (18.45) (2.38) (0.01) Common Language 0.0053 0.0006 0.0001 −0.0009 0.0045 0.0006 0.0794** 0.0019 (1.34) (0.34) (0.13) (0.62) (0.43) (0.60) (2.05) (0.49) Ever in colonial relationship 0.0011 0.0127*** 0.0036* 0.0139*** 0.0018 0.0049*** −0.0003 0.0264*** (0.28) (2.76) (1.96) (2.76) (0.19) (2.84) (0.01) (2.98) Island in couple (0, 1, or 2) 0.0104*** 0.0157*** 0.0047*** 0.0157*** 0.0076 0.0075*** 0.1029*** 0.0328*** (2.98) (3.40) (3.66) (3.17) (1.46) (3.67) (2.66) (3.38) Country-specific characteristics Log(GDP per capita in 1995 (PPP) in host) 0.1395*** 0.0369*** 0.0333*** 0.0215** 0.3817*** 0.0350*** 1.0291*** 0.1311*** (3.87) (3.25) (3.06) (2.23) (3.66) (4.23) (2.62) (3.21) Log(GDP per capita in 1995 (PPP) in origin) 0.0007 0.0010 0.0005 0.0013** 0.0045 0.0000 −0.0119 0.0008 (0.64) (1.52) (1.29) (2.10) (1.36) (0.08) (1.24) (0.47) Population density in 1990 in host (in hundreds) 0.0086*** 0.0024*** 0.0020*** 0.0014* 0.0212*** 0.0023*** 0.0708*** 0.0083*** (3.64) (2.96) (3.03) (1.96) (3.63) (3.63) (2.80) (3.19) Population density in 1990 in origin (in hundreds) 0.0009** −0.0023*** 0.0004* −0.0022*** 0.0032* −0.0010*** 0.0108* −0.0052*** (2.24) (3.21) (1.70) (2.94) (1.77) (3.46) (1.78) (3.15) Share of population age 15–24 in 1990 in host −0.0106*** −0.0053*** −0.0030*** −0.0042*** −0.0232*** −0.0037*** −0.0909*** −0.0157*** (4.13) (3.94) (4.57) (3.17) (3.98) (4.51) (3.12) (4.01) Share of population age 15–24 in 1990 in origin 0.0009* 0.0001 0.0005** 0.0000 0.0027* 0.0001 −0.0028 −0.0001 (1.84) (0.23) (2.18) (0.11) (1.81) (0.70) (0.54) (0.09) Age dependency ratio in host (mean 1990–2000) 0.4077*** 0.1971*** 0.1197*** 0.1547*** 0.9754*** 0.1273*** 3.2148*** 0.6036*** (4.25) (4.12) (4.65) (3.30) (3.79) (4.65) (3.00) (4.36) Age dependency ratio in origin (mean 1990–2000) 0.0089 0.0008 0.0021 0.0019 0.0535 −0.0017 −0.0273 −0.0090 (0.83) (0.19) (0.85) (0.49) (1.38) (0.70) (0.41) (0.77) Urban population growth in 1990 in host 0.0061*** 0.0025** 0.0016* 0.0017* 0.0168*** 0.0016*** 0.0430*** 0.0093*** (3.77) (2.51) (1.94) (1.67) (3.32) (3.07) (2.79) (3.73) Urban population growth in 1990 in origin 0.0002 −0.0002 0.0001 −0.0002 0.0000 0.0000 0.0049 −0.0004 (0.53) (0.56) (0.95) (0.67) (0.03) (0.27) (1.22) (0.58) Log(Mean education in 1990 in host) 0.0286** −0.0093* 0.0030 −0.0141** 0.1178*** −0.0018 0.1281 0.0009 (2.54) (1.66) (0.58) (2.38) (2.85) (0.62) (1.16) (0.07) - 84 - Literacy rate in origin (mean 1990–2000) 0.0001** −0.0000 0.0000** −0.0000* 0.0003** −0.0000 0.0003 −0.0000 (2.31) (0.73) (2.01) (1.80) (2.03) (0.44) (1.07) (0.05) Private return to education in host 0.8298*** 0.3107*** 0.1964*** 0.2335*** 2.0261*** 0.2294*** 7.1318*** 0.9274*** (3.71) (3.53) (3.65) (2.94) (3.69) (3.87) (2.88) (3.29) Unemployment rate (1994) for primary level in host −0.39** 0.15* −0.02 0.21** −1.54*** −0.02 −2.15 0.21 (2.52) (1.72) (0.29) (2.29) (3.11) (0.62) (1.40) (1.11) Unemployment rate (1994) for secondary level in host 0.03 −0.67*** −0.15 −0.71*** 1.33** −0.21** −2.36 −1.47*** (0.17) (3.00) (1.53) (2.96) (2.19) (2.36) (1.24) (3.37) Unemployment rate (1994) for tertiary level in host 0.53*** 0.69*** 0.23*** 0.66*** 0.47* 0.33*** 6.20*** 1.72*** (3.61) (3.61) (4.31) (3.22) (1.77) (3.80) (3.42) (4.22) Labor productivity growth in host (mean 1990–2000) 0.0296*** 0.0093*** 0.0070*** 0.0060** 0.0762*** 0.0074*** 0.2365*** 0.0320*** (3.85) (3.50) (3.38) (2.52) (3.56) (4.08) (2.85) (3.70) Public expenditure related to unemployment 0.0233*** 0.0051*** 0.0050** 0.0025* 0.0684*** 0.0054*** 0.1666** 0.0199*** compensation in host (mean 1990–2000; % GDP) (3.76) (2.78) (2.56) (1.66) (3.58) (3.94) (2.46) (2.87) −0.0000 −0.0009 −0.0000 −0.0006 0.0003 −0.0007 0.0009 −0.0054* Political rights in 1995 in origin (scale 1 to 7) (0.03) (0.97) (0.19) (0.67) (0.27) (1.30) (0.30) (1.74) Civil liberties in 1995 in origin (scale 1 to 7) −0.0005 0.0004 −0.0001 0.0001 −0.0021 0.0006 −0.0010 0.0046 (0.76) (0.41) (0.44) (0.11) (0.95) (0.98) (0.23) (1.33) Subgroups of host countries fixed effects [Ref. Anglo-Saxon countries] Northern countries −0.0502*** 0.0000 −0.0106** 0.0064 −0.1614*** −0.0073*** −0.3236** −0.0238* (3.63) (0.01) (2.09) (1.43) (3.38) (2.64) (2.23) (1.69) Continental European countries −0.0069*** 0.0047** −0.0014 0.0053** −0.0242*** 0.0002 −0.0549** 0.0079 (2.84) (1.97) (0.91) (2.02) (3.22) (0.16) (2.52) (1.54) Southern European countries 0.0691*** 0.0266*** 0.0173*** 0.0186*** 0.1797*** 0.0186*** 0.5229*** 0.0864*** (3.87) (3.72) (3.87) (2.89) (3.64) (4.06) (2.61) (3.79) Subgroups of origin countries fixed effects [Ref. Sub-Saharan Africa] East Asia & Pacific 0.0053** 0.0010 0.0157** 0.0451* (2.23) (1.59) (1.97) (1.93) Europe & Central Asia −0.0173*** −0.0048** −0.0471*** −0.1295*** (3.55) (2.25) (2.98) (2.97) Latin America & Caribbean 0.0098*** 0.0026*** 0.0235*** 0.0995** (3.49) (3.51) (3.48) (2.30) South Asia 0.0009 −0.0006 0.0058 −0.0154 (0.40) (0.74) (0.70) (0.94) Constant −1.5154*** −0.3976*** −0.3507*** −0.2146** −4.2750*** −0.3756*** −10.6830** −1.4774*** (3.80) (3.19) (2.85) (2.11) (3.57) (4.07) (2.54) (3.28) Observations 1183 221 1183 221 1183 221 1183 221 R2 0.31 0.54 0.25 0.55 0.24 0.65 0.20 0.55 Note: Robust t-statistics to clustering are in parentheses. ***, **, and * mean, respectively, significant at the 1 percent, 5 percent, and 10 percent levels. - 85 - Table B5. Panel Data Gravity Model: Specification (1) Dependent Variable: Yearly Emigration Rates for 1990–2002 (Inflowsijt/Popjt) ALL MENA MENA X MENA X MENA X MENA X NORTHER ANGLO- CONTINEN SOUTHER N EUROPE SAXON TAL N EUROPE EUROPE (1) (2) (3) (4) (5) (6) Bilateral characteristics Ln(distance) −0.468*** −0.093*** −0.013 0.035 −0.979*** −0.448 (4.65) (2.79) (0.91) (0.54) (5.44) (0.69) Common Border 0.394 0.295*** 0.000 0.000 0.000 0.000 (0.99) (5.83) (.) (.) (.) (.) Common Language 0.049 0.137*** 0.011* 0.051 −0.029 0.000 (0.71) (2.73) (1.86) (0.95) (0.94) (.) Ever in colonial relationship 0.061 0.062 0.000 −0.187** 0.116* 0.000 (0.90) (0.97) (.) (2.16) (1.81) (.) Island in couple (0, 1, or 2) 0.503 −0.292 0.051 −2.577 −0.981 0.000 (1.14) (0.65) (1.66) (0.66) (1.18) (.) Country characteristics Log (GDP per capita) in host −0.066 −0.416 −0.079* 0.190 −1.506* 0.000 (0.20) (1.39) (1.74) (0.50) (1.97) (.) Log (GDP per capita) in origin −0.061 0.007 −0.006 −0.033 −0.001 −0.119 (1.09) (0.11) (0.59) (0.35) (0.01) (0.44) Population density in host 0.001 −0.001 −0.007* −0.009 0.001 0.058 (1.05) (0.58) (1.72) (0.58) (0.14) (0.87) Population density in origin −0.000 −0.000*** −0.000** −0.000 0.000 −0.000 (0.76) (2.71) (2.24) (1.55) (0.16) (0.26) Sh. of pop. age 15–24 in host −12.657*** −3.053* −0.805** 1.637 0.202 0.000 (3.03) (1.97) (2.64) (0.35) (0.05) (.) Sh. of pop. age 15–24 in origin −1.150 −0.283 −0.192 2.180 −1.446 −3.877 (0.62) (0.28) (1.21) (1.07) (1.47) (0.98) Age dependency ratio in host −4.679*** −1.054 0.752 −0.011 −0.725 0.000 (3.62) (1.02) (1.52) (0.00) (0.24) (.) Age dependency ratio in origin 0.696** 0.335* −0.004 0.751* 0.479** −0.077 (2.27) (1.68) (0.13) (1.97) (2.50) (0.17) Urban pop. growth in host −1.261 3.043** 0.397 3.651 −9.625 0.000 (1.26) (2.18) (1.58) (1.02) (1.53) (.) Urban pop. growth in origin −0.374 0.442* −0.008 0.430 0.363 1.643 (1.40) (1.68) (0.15) (1.23) (1.10) (1.21) Log (Mean education) in host 0.789*** 0.472 −0.039 2.216 3.795* 0.000 (2.58) (1.50) (0.62) (1.03) (1.89) (.) Literacy rate in origin 0.730* 0.622 0.051 0.379 2.305*** 1.434 (1.70) (1.57) (1.01) (0.47) (2.88) (0.89) Unemployment rate in host −0.228 −0.670 −0.215* 2.105 −4.654*** 0.000 (0.31) (1.05) (1.74) (1.18) (2.87) (.) Labor productivity growth in host 0.182 0.086 0.006 0.379 −0.223 0.000 (0.66) (0.71) (0.21) (0.73) (0.41) (.) Share of public expenditure related to 6.832*** 2.288* −0.351 −1.533 8.479** 0.000 unemployment in host (2.59) (1.66) (0.72) (0.23) (2.08) (.) Political rights in origin −0.003 0.018** 0.001 0.019 0.013 0.028 (0.33) (2.44) (1.16) (1.43) (1.05) (0.76) Civil liberties in origin 0.005 −0.016 −0.001 −0.008 0.000 −0.121 (0.28) (1.56) (0.41) (0.68) (0.01) (1.18) Constant 5.703* 4.510 0.896* −3.111 14.215* 0.100 (1.94) (1.46) (1.92) (0.43) (1.71) (0.01) Observations 9,504 1,558 526 486 415 131 R2 0.29 0.47 0.40 0.55 0.82 0.59 Note: Robust t-statistics are in parentheses. ***, **, and * mean, respectively, significant at the 1 percent, 5 percent, and 10 percent levels. Year, destination country, and origin country dummy variables included but not shown. - 86 - Table B6. Panel Data Gravity Model: Specification (2) Dependent Variable: Yearly Emigration Rates for 1990–2002 (Inflowsijt/Popjt) ALL MENA MENA X MENA X MENA X MENA X NORTHE ANGLO- CONTIN SOUTHE RN SAXON ENTAL RN EUROPE EUROPE EUROPE (1) (2) (3) (4) (5) (6) Log (GDP per capita) in host −0.124 −0.394*** −0.081 0.238 −1.785** 0.000 (0.54) (3.35) (1.16) (0.32) (2.46) (.) Log (GDP per capita) in origin −0.097*** 0.018 −0.006 −0.034 0.084 −0.119 (2.71) (0.49) (0.76) (0.47) (1.18) (0.44) Population density in host 0.002 −0.001 −0.009* −0.018 0.002 0.058 (0.56) (0.53) (1.82) (0.78) (0.31) (0.98) Population density in origin −0.000 −0.000** −0.000*** −0.000 −0.000 −0.000 (1.24) (2.47) (3.37) (1.48) (0.31) (0.17) Sh. of pop. age 15–24 in host −13.91*** −3.651*** −0.871** 5.055 −0.118 0.000 (8.49) (4.60) (2.35) (0.66) (0.02) (.) Sh. of pop. age 15–24 in origin −0.974 −0.211 −0.189** 1.750** −2.037*** −3.877 (1.35) (0.56) (2.46) (2.31) (2.80) (1.46) Age dependency ratio in host −4.272*** −1.052** 1.057* −0.734 −0.794 0.000 (4.85) (2.49) (1.91) (0.11) (0.30) (.) Age dependency ratio in origin 0.602*** 0.304*** −0.003 0.684*** 0.354** −0.077 (4.07) (4.54) (0.24) (5.41) (2.29) (0.15) Urban pop. growth in host −0.441 3.098*** 0.408 0.148 −9.109 0.000 (0.18) (2.60) (1.49) (0.02) (1.35) (.) Urban pop. growth in origin −0.395 0.461*** −0.012 0.517 0.091 1.643 (1.52) (2.58) (0.33) (1.52) (0.21) (1.19) Log (Mean education) in host 0.953*** 0.496*** −0.076 3.940 4.312*** 0.000 (4.86) (5.06) (1.15) (1.26) (3.69) (.) Literacy rate in origin 0.615** 0.613*** 0.044 0.598 1.915*** 1.434 (2.57) (2.87) (1.06) (1.45) (3.93) (0.85) Unemployment rate in host −0.317 −0.589*** −0.231** 1.885 −4.784*** 0.000 (0.77) (2.80) (2.01) (0.71) (5.26) (.) Labor prod. growth in host 0.059 0.026 0.003 0.489 0.003 0.000 (0.15) (0.13) (0.06) (0.59) (0.00) (.) Share of public expenditure 7.902*** 2.093** −0.410 −2.954 9.326* 0.000 related to unemployment in host (3.88) (2.10) (0.98) (0.34) (1.73) (.) Political rights in origin −0.007 0.014*** 0.001 0.018** 0.011 0.028 (1.14) (2.81) (1.27) (1.97) (0.96) (0.83) Civil liberties in origin 0.008 −0.016*** −0.001 −0.012 −0.005 −0.121*** (0.98) (3.38) (0.70) (1.32) (0.54) (3.49) Constant 2.870 3.364** 1.155 −11.766 7.246 −3.395 (1.07) (2.50) (1.29) (1.61) (0.88) (0.76) Observations 9,649 1,558 526 486 415 131 R2 0.68 0.81 0.65 0.86 0.89 0.59 Note: Robust t-statistics are in parentheses. ***, **, and * mean, respectively, significant at the 1 percent, 5 percent, and 10 percent levels. Year dummies are included but not shown due to space limitations. The list of receiving and sending countries included in the sample and summary statistics on the variables are provided in Tables B1 and B2. - 87 - REFERENCES Al-Azar, R. 2005. “Italian Immigration Policy: the Reversal of the Tide.”, mimeo. Asplund, R. ed. 2001. Public Funding and Private Returns to Education – PURE – Final Report. Series of the European Commission, DG Research. www.etla.fi/PURE. Bardhan, A. D., and C. Kroll. 2003. “The New Wave of Outsourcing.” Fisher Center Research Report No. 1103. Fisher Center for Real Estate & Urban Economics, University of California, Berkeley. Barro, R. J., and J.-W. Lee. 2000. “International Data on Educational Attainment: Updates and Implications.” CID Working Paper No. 42, April, Center for International Development, Harvard University, Cambridge, MA. Bauer, T., and K. F. Zimmermann. 1997. “Integrating the East: The Labor Market Effects of Immigration.” In S. W. Black, ed., Europe’s Economy Looks East - Implications for Germany and the European Union, 269–306. Cambridge: Cambridge University Press. Borjas, G. J. 1987. “Self-selection and the Earnings of Immigrants.” American Economic Review 77: 531–53. Borjas, G. J. 1994. “The Economics of Immigration.” Journal of Economic Literature 32: 1667– 1717. Borjas, G. J. 1999. “Immigration and Welfare Magnets.” Journal of Labor Economics 17(4): 607–37. Boswell, C., S. Stiller, and T. Straubhaar. 2004. “Forecasting Labour and Skills Shortages: How Can Projections Better Inform Labour Migration Policies?” Hamburg Institute of International Economics, Migration Research Group. http://ec.europa.eu/employment_ social/employment_analysis/docs/forecast_short3.pdf. Bruni, M., and A. Venturini. 1995. “Pressure to Migrate and Propensity to Emigrate: The Case of the Mediterranean Basin.” International Labor Review 134: 377–400. CARIM (Euro-Mediterranean Consortium for Applied Research on International Migration). 2001. Population and Housing Census 2001. Robert Schuman Centre for Advanced Studies, European University Institute, Florence. CARIM (Euro-Mediterranean Consortium for Applied Research on International Migration). 2005. Mediterranean Migration 2005 Report, Robert Schuman Centre for Advanced Studies, European University Institute, Florence. Chiswick, Barry R., and T. Hatton. 2002. “International Migration and the Integration of Labor Markets.” IZA Discussion Paper No. 559, Institute for the Study of Labor, Bonn. - 88 - Constant, A., and K. F. Zimmermann. 2005. “Immigrant Performance and Selective Immigration Policy: A European Perspective.” IZA Discussion Paper No. 1715, August, Institute for the Study of Labor, Bonn. Cotis, J.-P. 2003. “Population Ageing: Facing the Challenge.” OECD Observer No. 239, September, Paris. DGTPE (Direction Générale du Trésor et de la Politique Économique). 2006. “Immigration sélective et besoins de l’économie française”, Rapport, French Ministry of the Economy, Finance and Industry, Paris. Docquier, F., and A. Marfouk. 2005. “International Migration by Educational Attainment (1990- 2000), Release 1.1.” In C. Ozden and M. Schiff, eds., International Migration, Remittances and Development. New York: McMillan and Parlgrave. Doudeijns, M., and J-C. Dumont. 2003. “Immigration and Labour Shortages: Evaluation of Needs and Limits of Selection Policies in the Recruitment of Foreign Labour.” Presented at the conference on “The Economic and Social Aspects of Migration,” 21-22 January 2003, Brussels Dumont, J-C., and G. Lemaître. 2005 “Counting Immigrants and Expatriates in OECD Countries: a New Perspective.” OECD Social, Employment, and Migration Working Paper No. 25, Organisation for Economic Co-operation and Development, Paris. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. London: Polity Press. Fargues, P., ed. 2005. Migrations méditerranéennes, rapport 2005. Cooperation Project on the Social Integration of Immigrants, Migration, and the Movement of Persons, financed by the ECMEDA Programme, Florence: EUI-RSCAS, CARIM Consortium. Feenstra, R., J. A. Markusen, and A. K. Rose. “Using the Gravity Equation to Differentiate Among Alternative Theories of Trade.” Canadian Journal of Economics 34(2): 430–447. Giubilaro, D. 1997. “Migration from the Maghreb and Migration Pressures: Current Situation and Future Prospects.” International Migration Paper No. 15, Geneva, ILO. Glick, R., and A. Rose. 2002. “Does a Currency Union Affect Trade? The Time-series Evidence.” European Economic Review 46(6): 1125–1151. Harris, J. R., and M. P. Todaro. 1970. “Migration, Unemployment and Development: A Two- Sector Analysis.” American Economic Review 60: 126–42. Hatton, T., and J. G. Williamson. 2002. “What Fundamentals Drive World Migration?” NBER Working Paper No. 9159, National Bureau of Economic Research, Cambridge, MA. Hicks, J. 1932. The Theory of Wages. London: MacMillan. - 89 - Holzmann, R., and R. Münz. 2004. “Challenges and Opportunities in International Migration for the EU, Its Member States, Neighboring Countries and Regions: A Policy Note.” Social Protection Discussion Paper Series No. 0411, Social Protection Unit, Human Development Network, World Bank. ILO (International Labour Organization). 2006. http://laborsta.ilo.org/ INSEE (Institut National de la Statistique et des Études Économiques). 2002. Les immigrés en France, coll. Repères, Paris. Kapiszewski, A. 2004. “Arab Labour Migration to the GCC States.” In IOM (International Organization for Migration) eds., Arab Migration in a Globalized World, 115–133. Geneva: ILO. Keller, J., and M. Nabli. 2002. “The Macroeconomics of Labor Market Outcomes in MENA over the 1990s: How Growth Has Failed to Keep Pace with a Burgeoning Labor Market.” World Bank, Washington, DC. Koettl, J. 2006. “Prospects for Management of Migration Between Europe and the Middle East and North Africa.”, OECD, mimeo. Labib, A. 1996. “L'immigration maghrébine en Italie: du transit à l'installation.” Hommes et Migrations No. 1194. Mayda, A. M. 2005. “International Migration: A Panel Data Analysis of Economic and Non- Economic Determinants.” Georgetown University, May. Migration Information Source. 2006. http://www.migrationinformation.org/ Münz, R. 2004. “Migrants, labour markets and integration in Europe: a comparative analysis”, Global Migration Perspectives 16, Global Commission on International Migration (GCIM), Geneva. Nugent, J. B. 2002. “Why Does MENA Trade So Little?” Paper presented to the Middle East Region Group at the World Bank, Washington, DC. OECD. (Organisation for Economic Co-operation and Development). Various years. Employment Outlook. Pedersen, P. J., M. Pytlikova, and N. Smith. 2004. “Selection or Networks Effects? Migration Flows into 27 OECD Countries, 1999-2000.” IZA Discussion Paper No. 1104, April, Institute for the Study of Labor, Bonn. Peri, G. 2005. “International Migration: Some Comparisons and Lessons from the EU.” University of California, September. - 90 - Psacharopoulos, G. and H. A. Patrinos. 2004. “Returns to Investment in Education: A Further Update.” Education Economics 12 (2): 111–34. Rotte, R., and M. Vogler. 1998. “Determinants of International Migration: Empirical Evidence for Migration from Developing Countries to Germany.” IZA Discussion Paper No. 12, Institute for the Study of Labor, Bonn. Rotte, R. and M. Vogler. 1999. “The Effects of Development on Migration: Theoretical Issues and New Empirical Evidence.” IZA Discussion Paper No. 46, Institute for the Study of Labor, Bonn. Schaeffer, P. V. 1993. “A Definition of Migration Pressure Based on Demand Theory.” International Migration 31: 43–72. Simoens, S., M. Villeneuve, and J. Hurst. 2005. “Tackling Nurse Shortages in OECD Countries.” OECD Health Working Paper No. 19, Directorate for Employment, Labour and Social Affairs, Organisation for Economic Co-operation and Development, Paris. Sjaastad, L. A. 1962. “The Costs and Returns of Human Migration.” Journal of Political Economy 70: 80–93. Wikipedia 2006. “Outsourcing.” Wikipedia, The Free Encyclopedia. http://en.wikipedia.org/w/index.php?title=Outsourcing&oldid=53209906 World Bank. 2005. Doing Business database.