Refugees, Asylum Seekers, and Policy in Europe

Timothy J. Hatton and Jeffrey G. Williamson Refugees, Asylum Seekers, and Poliey in Europe Abstract: The number of refugees worldwide is now 12 million, up from 3 million in the early 1970s. And the number seeking asylum in the developed world has in- creased tenfold, from about 50,000 per annum to half a million over the same pe- riod. Governments and international agencies have grappled with the twin prob- lems of providing adequate humanitarian assistance in the Third World and avoiding floods of unwanted asylum seekers arriving on the doorsteps of the First World . This is an issue that is long on rhetoric, as newspaper reports testify, but surprisingly short on economic analysis . This paper draws on the recent lit- erature, and ongoing research, to address aseries of questions that are relevant to the debate. First, we examine the causes of refugee displacements and asylum flows, focusing on the effects of conflict, political upheaval, and economic in- centives to migrate. Second, we examine the evolution of policies towards asy- lum seekers and the effects of those policies, particularly in Europe. Finally, we ask whether greater international coordination could produce better outcomes for refugee-receiving countries and for the refugees themselves. 1 Introduction The worldwide number of refugees has increased by a factor of four since the early 1970s-from about 3 to 12 million. And the number seeking asylum in de- veloped countries has increased by a factor of about ten over the same period- Remark: This paper draws heavily on our own recent work (Hatton and Williamson 2003 ; Hatton 2004) and on our ongoing book World Mass Migration: Two Centuries of Policy and Performance. We are grateful to the organizers of the Kiel Week Conference, where an earlie r version of this paper was presented, and to the participants at the conference for their comments. Hatton acknowledges support through a British Academy Research Readership and Williamson acknowledges financial support from the National Science Foundation SES-0001363. We are both grateful to the Australian National Univers ity for providing us with the opportunity and the env ironment for collaboration on this research project. Williamson also thanks the University of Wisconsin Economics Departrnent, where this paper was cornpleted while on leave from Harvard. F. Foders et al. (eds.), Labor Mobility and the World Economy © Springer-Verlag Berlin Heidelberg 2006 250 TIMOTHY J. HATION AND JEFFREY G. WILLIAMSON from about 50,000 per annum in the early 1970s to half a million in 2001 . This is seen by many as a crisis of growing proportions, and for two reasons. First, there is the humanitarian issue . Most refugees are displaced across borders in the Third World, suffering oppression, poverty and disease. With each new humanitarian crisis comes new criticism of the unwillingness or inability of governments and international agencies to act more decisively to solve the refugee problem, or at least to better ameliorate the condition of the refugees. Second, the arrival of in- creasing numbers of asylum seekers on the doorsteps of the First World has led to fierce political debate about asyl um policies, often fueled by parties of the far right. Despite protest from humanitarian groups, governments have responded to the rising political temperature with a range of measures aimed at deterring asy- lum applications. This paper draws on the recent literature and some of our own analysis to ad- dress aseries of questions that are relevant to these debates. First, what are the causes of refugee displacement? Are they mainly political or economic? Second, how far can wars, political crises and economic conditions explain the apparently inexorable rise in the number of asyl um seekers, particularly in Europe, but in the rest of the OECD as weil? Third, how has asylum policy evolved in Europe and has it been effective in deterring asylum seekers, or has it instead simply de- tlected them elsewhere? Fourth, could greater international cooperation lead to outcomes that are better for refugee-receiving countries as weil as for the refu- gees themselves? And finally, are there better policies? 2 Refugees and Asylum Seekers Refugee and asylum seeker figures co me from estimates compiled by the United Nations High Commission for Refugees (UNHCR). The definition of a refugee is derived from the 1951 Geneva Convention on Refugees, namely someone who, owing to a well-founded fear of persecution, is outside his or her country of normal residence and who is unable or unwilling to return to it. The UNHCR es- timates plotted in Figure 1 show a dramatic increase from the early 1970s to a peak of nearly 18 million in 1992, before falling by a third to 12 million in 2001.' Two further points are worth noting about these refugee totals. First, they exclude an additional 5 million who in 2001 were internally displaced and living Estimates of the total number of refugees differ. The United Stares Committee für Refugees (2003: 3) provides a figure of 14.9 million for 2001, nearly 3 million higher than the UNHCR estimate. Some of the definitional issues in constructing these totals are discussed in UNHCR (2002 : 71-75) and Crisp (1999). Nevertheless, while esti- mates of the total differ, the profile of change over time is essentially the same. R EFUGEES, A SYL UM SEEKERS, AND POLICY IN E UROPE 251 in refugee-like situation s but who were not classified as refugees because they were not outside their country of origin. Second, these refugee stock figures un- dergo considerable turbulence and tumover. When the refugee stock fell by 5.8 million between 1992 and 2001, there were 10.7 million new refugee arrivals and 16.5 million exits. Figure 1: Worldwide Stock of Refugees World Europe NorthA~ 2000 Year 1996199219881984 >-< 1980 Number (0008) 20,000 18,000 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2.000 0 1972 1976 Source: UNHCR (2001: Annex 3). The overwhelming majority of these refugees are located in the Third World, close to their country of origin . As Table 1 shows, there is a very strong correla- tion between the number of refugees that originated in a region and the number who are located within that region. In 2001, 49 percent of refugees originated somewhere in Asia or the Middle East and 46 percent were located there, while 30 percent originated in Africa and 27 percent were located there. Eastem Europe was the source of 11 percent of the refugees and host to 6 percent, some of whom were from outside the region. Western Europe hosted 1.7 million refu- gees from other regions in 2001 (14 percent of the worldwide total) while a fur- ther 646,000 (5 percent) were located in North America. As Figure 1 shows, the number of refugees in Europe rose sharply between the late 1980s and the early 1990s, while the number in North America declined. These trends are consistent with the rapid growth in the number of asylum seekers arriving in Europe . 252 TIMOTHY J. HATTON ANOJEFFREY G. WILLIAMSON Table 1: Refugees by Region of Origin and Location, 1992 and 2001 (Thousands) Region Refugees by origin Refugees by location 1992 2001 1992 2001 GreatLakes Region of Africa 700.2 1,055.3 983.6 1,190 .7 Westand Central Africa 960.1 540.4 950.5 570 .1 Eastand Horn of Africa 1,928 .8 1,364.7 1,784.7 966 .8 Southem Africa 1,757.3 473.6 1,506.0 365.4 North Africa 245.5 206.7 257.2 183.6 The Middle East 1,454.4 901.4 266.3 463.4 SouthWestAsia 4,682.0 3,914.4 5,840.0 4,066.8 Central Asia 60.0 63.4 -3.0 97.8 SouthAsia 503 .6 396.6 579.0 322 .7 East Asiaand thePacific 731.6 667.4 473.0 616.1 Eastern Europe 708 .8 372.9 546.0 294.3 SouthEastern Europe 700.5 897.2 954.4 459.6 Central Europe and the Baltic 60.1 62.8 147.8 18.7 Stares Western Europe 0.0 1.8 1,841.0 1,731.5 North America and the Caribbean 23.8 26.7 769.7 646 .1 Central America 129.1 30.9 853.4 25.8 SouthAmerica 19.4 27.9 22.1 10.5 StatelesslOtherlUnknown 3,132.0 1,025.7 20.7 Total 17,798.5 12,029.9 17,798.5 12,029.9 Source: UNHCR (2002: 84, 88). Figure 2 plots the total number of new asylum applications lodged in 37 in- dustrialized countries (a tlow rather than a stock) by region of asylum.I It docu- ments an enormous surge from about 150,000 per annum in the early 1980s to a peak of more than 850,000 in 1992, falling sharply to 380,000 in 1997 before rising again more recently. It also confirms once again that the bulk claimed asylum in Europe, principally in the 15 countries of the European Union (pre- enlargement). The EU accounted for 68 percent of all applications over the 20- year period, and North America accounted for most of the remainder. The sharp spike in the early 1990s (Figure 3) was accounted for by applications from East- em Europe following the disintegration of the Soviet Union and the contlicts that followed the breakup of the former Yugoslavia. Still, there is evidence of an un- 2 These data are collected by the UNHCR, mainly from national govemments. They are typically first instance claims and they represent the number of applications rather than the number of individuals. On average, eachapplication represents around 1.2 to 1.3 individuals. R EFUGEES, A SYLUM SE EKERS, AND POLI CY IN E UROPE 253 Figure 2: Asylum Applications to Industrialized Countries, 1982-2001 Number (OOOs) 1,000r-----------------------------, 900 800 700 600 500 400 300 200 100 O~:;::=:;:::=:;::::::;:::=_~~~~~~~~~~~~~..____l 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 Year Source: UNHCR (2002: 112-113). Figure 3: Asylum Applications by Source Region, 1982-2001 Number (ooOs) 9OOr-----------------------------, 800 700 600 500 .. . ..................... ........._ ~ ..~ . \. ../ . ···.....Europe "\ Alrica ./ ...•./ .... -----..':.....------;::-:::: ....... Asia 100 200 400 300 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 Year Source: UNHCR (2002: 115-116). derlying upward trend, not only in applications from Europe, but also from Af- rica and Asia. 254 TIMOTH Y J. HATTON AND JEFFREY G. WILLIAM SON The left-hand panel of Table 2 reports the total number of applications from each of the top 20 source countries to the developed world by decade. Not sur- prisingly, Eastem Europe is weil represented with large numbers arriving from Yugoslavia, Romania, Bosnia, and the Russian Federation, as is the Middle East with large numbers from Turkey, Iraq, Iran, and Afghanistan. Asia is also weil represented by China, India, Pakistan, and Sri Lanka, While there is a clear cor- respondence between the number of refugees and the number of asylum seekers, some of the poorest countries (such as Angola, Rwanda, and EthiopialEritrea) did not generate as many asylum seekers as might have been expected given the scale of the conflicts. Indeed, most African refugees do not get much further than a neighboring country, if they manage to leave at all. It is also notable that most of the countries that generated large numbers of asylum seekers in 1992-2001 also generated significant numbers in the previous decade. As we shall see, this historieal persistence is an important feature of asylum flows, especially to Europe. Table 2: Top 20 Sources and Destinations of Asylum Seekers, 1982-2001 1992- 200 1 1982-1991 1992-2001 1982-1 991 Source Country Destination No. per No. per Country No. per No. per 1,000 1,000 1.000 1,000 Yugoslavia FR 817.2 77.5 269.8 17.0 Germany 1,597.3 19.6 996.9 1.7 Iraq 310.8 15.4 52.1 3.4 United Stares 869.0 3.2 437.7 0.2 Turkey 308.8 4.9 499.3 9.6 United Kingdom 576.6 10.0 164.5 0.3 Romania 304.7 13.1 195.4 8.6 Netherlands 358.6 23.2 95.0 0.7 Afghanistan 204.1 10.6 54.0 4.0 Canada 286.3 9.8 239.4 0.9 EI Salvador 196.5 34.7 69.6 14.6 France 281.0 4.8 347.4 0.6 Bosnia and Switzerland 243.5 34.2 170.2 2.6 Herzegovina 186.1 54.4 - Sweden 228.6 25.9 183.2 2.2 Sri Lanka 168.9 9.5 176.8 11.3 Belgium 219.5 21.7 69.7 0.7 Islamic Rep. of Iran 161.1 2.6 195.5 4.0 Austria 128.0 15.9 134.0 1.8 Guatemala 154.8 15.5 38.1 4.9 Denmark 97.4 18.6 45.3 0.9 China 149.7 0.1 21.9 0.0 Australia 89.2 4.9 30.1 0.2 Somalia 147.6 20.1 51.1 7.7 Spain 84.2 2.1 37.5 0.1 India 124.4 0.1 63.2 0.1 Italy 83.4 1.5 55.0 0.1 Pakistan 113.2 0.9 67.9 0.7 Norway 71.1 16.3 32.3 0.8 Russian Federation 105.2 0.7 25.0 0.2 Turkey 54.5 0.9 27.8 0.1 Dem. Rep. of Congo 103.9 2.3 97.5 3.0 Czech Republic 48.1 4.7 3.8 0.0 Algeria 92.6 3.3 - Ireland 39.7 11.0 0.03 0.0 Bulgaria 9 1.2 10.8 47.5 5.3 Hungary 37.5 3.7 4.4 0.0 Nigeria 77. 1 0.8 33.0 0.4 Poland 25.0 0.6 2.4 0.0 Mexico 74.5 0.8 Source: Asylum seeker numbers from UNHCR (2002: 112-113, 115-116); population totals for 1995 and 1985 from United Nations (2003: various tables). REFUGEES, ASYLUM SEEKERS, AND POLlCY IN E UROPE 255 The right-hand panel of Table 2 documents the top 20 countries in the devel- oped world that receive asylum applications. Germany tops the list with a mas- sive 1.6 million applications in 1992-2001 , followed by the United States and the United Kingdom. Nine other EU countries also appear on the list, some of which have very large per capita rates (Belgium , Netherlands, Sweden, Switzer- land). More notable still is the appearance on the list of Turkey , the Czech Re- public , Hungary , and Poland, countries that were major sources of asylum seek- ers in the 1980s. Those countries might have become more attractive havens in the post-Soviet period, but it mayaiso retlect the increasing difficulty of entering Western Europe. While most destination countries experienced an increase in the number of applications between the 1980s and the 1990s, the growth in those numbers varied widely. In Western Europe, large percentage increases occurred in the United Kingdom, the Netherlands, Belgium, and Ireland (from a base close to zero), while modest increases or small declines occurred in France, Switzer- land, Sweden , and Austria. We shall examine the pattern of asylum claims in European destinations later, but first we turn to refugee origins. 3 The Causes of Refugee Flights What causes refugees to tlee their horne country and seek sanctuary abroad? Per- haps the answer seems too obvious to warrant further discussion. For major refu- gee displacements such as those in Rwanda, Somalia, EI Salvador, or Afghani- stan the associated wars and contlicts are well known. Quantitative analyses typically focus on some measure of the incidence and intensity of contlict within the country as the main explanation. Table 3 offers an example using data from sub-Saharan Africa. Here the dependent variable is the total stock of refugees from a given country per thousand of the source country population, across 41 African countries annually from 1987 to 1992. There are 142 cases in this sam- pie where no refugees are observed and another 104 with positive numbers of refugees. We therefore use tobit analysis. The explanatory variables exploit a va- riety of measures of violence and political instability from data assembled by Robert Bates (see Hatton and Williamson 2003) . Given that most of the variables in Table 3 are dummies , the coefficients can be read as the number of refugees per thousand created as a result of the violence event in question. Coups d'etat typically create an eftlux of 35 per thousand , while government crises generate about 18 per thousand, but these effects are only on the borderl ine of significance. The incidence of guerilla warfare yields a more significant coefficient and the effect generates 25 refugees per thousand of the population. However, civil war is the most important variable generating 256 TIMOTHY 1. HATTON AND 1EFFREY G. WILLIAMSON refugees-about 35 refugees per thousand-while each military death in civil war (a proxy for the intensity of the war) generates another 35 refugees per thou- sand. Despite the strength of the civil war effects, the explanatory power of the regression as a whole is still rather low, as reflected in the pseudo R2. This is be- cause of the enormous heterogeneity in the size and intensity of the conflicts, and their refugee-generating effects , none of which is adequately reflected by these crude explanatory variables . Table 3: Explaining Refugee Displacements in Africa RefslPop = -32.4 +35.7 Coups +18.3 Crises (4.8) (1.9) (1.9) +25.2 GuerWar +35.4 CivWar (3.3) (2.9) + 35.4 Deaths, (4.4) pseudo R2 = 0.07, log likelihood = -595.6 Note : r-statistics in parentheses . - Sampie: Balanced panel of 41 countries in sub-Saharan Africa by 6 years (1987-1992). - Variable defin itions: RefslPop : number of refugees per thousand of source country population ; Coups: dummy = 1 for years when there was a political coup d'etat: Crises: dummy=1 for years of government crisis; GuerWar; dummy =1 for years of guerilla warfare; CivWar : dummy =1 for years of civii war; Deaths: num- ber of military deaths in civil war per thousand of the population . - Method: Tobit regres- sion on 246 country/year observations . Source : Hatton and Williarnson (2003: Table 2), where the data are discussed in more detail. The original source for most of the variables is Robert Bates' Africa project, available at http://africa .gov.harvard.edu//. Perhaps these results are unsurprising, but they raise a number of issues. First, if as some believe, refugee flights are determined by economic and demographie forces as well as by politics and violence, then these variables should also playa role. Indeed, some studies do find such effects, but they are generally weak in comparison to polities and violenee . The Afriean data also support that view. When variables sueh as real wage rates and the share of population aged 15-29 were added, they proved to be insignificant. Should we therefore eonelude that eeonomie and demographie faetors play no role? Not neeessarily, sinee war and violenee are highly eorrelated with poor eeonomie eonditions and large young adult populations. And onee the eonflicts cease, the eeonomy rapidly bounees back, so that their effects on refugees may not persist.I 3 Collier (1999) finds that the economies of war-ravaged states grow at 6 percent per year in the five years following the outbreak of peace. Pottenbaum and Kanbur (2001) find that the postwar bounce back in a range of socioeconomic indicators is REFUGEES, ASYLUM SEEKERS , AND POLICY IN EUROPE 257 Second, the regression analysis is applied to the stock of refugees, rather than the tlow into and out of refugee status , If refugees remain displaced for a number of years, even after the end of the contlict or crisis, then lagged values should also matter. However, when the regression in Table 3 is reestimated with the ad- dition of explanatory variables lagged one and two years, the coefficients were not significant, suggesting that persistence is not important. This contrasts with the findings of some other studies where lags are found to matter (Schmeidl 1997; Azam and Hoeftler 2000). There are two possible explanations for our contlicting result. One is that the dynamics cannot be isolated in a short panel where most of the variance is in the cross-section. The other is that most refugee migrations in Africa occur just across borders; they are temporary and quickly reversed. Thus, of thel4.2 million voluntary repatriations in 1992-200 I, 8.3 mil- lion of them were in Africa. It is worth dwelling on the second point for what follows later . Most of the refugee displacements occur when war breaks out or when there is a sudden rise in its intensity. Thereafter the tlood subsides and the stock of refugees rises more gently . When the war or crisis abates, repatriation takes piace, often quickly and on a scale that resembles that of the original displacement. This is particularly true in Africa, where refugees are often displaced to camps just across the border in which conditions are worse even than those in the refugee's war-r