Trends and Patterns of Global Refugee Migration

Abstract

This paper studies long‐term trends and patterns in global refugee migration. We explore the intensity, spread, and distance of refugee migration at a global, regional, and country level between 1951 and 2018. The analysis did not detect a long‐term increase in the global intensity of refugee migration. Primarily depending on levels of conflict, refugee numbers have fluctuated at levels of between 0.1 and 0.3 percent of the world population. Apparent increases in numbers of the globally displaced are driven by the inclusion of populations and countries that were previously excluded from the data. While refugee populations continue to be concentrated in countries with low‐to‐medium income levels, the analysis reveals several geographic shifts in refugee migration. Refugees tend to come from a shrinking number of origin countries and move to an increasing variety of destination countries. This trend seems to reflect a concentration of recurrent conflict cycles in a relatively small number of countries and a parallel increase in the number of safe destinations. Although the vast majority of refugees remain near to origin countries, the average distance between origin and destination countries has increased over time, presumably linked to the greater ease of travel and migration‐facilitating networks.

Introduction

Refugee migration is at the center of political debates and has been the subject of an expanding body of academic research. Most research focuses on the movement of particular refugee populations, generally as a consequence of the outbreak of violent conflicts in various parts of the world. Few studies investigate the long‐term trends and patterns of refugee migration (see Marfleet ²⁰⁰⁷). Because the study of refugee migration tends to be highly policy driven (Black 2001; Bakewell ²⁰⁰⁸), most studies focus on pressing issues such as refugee reception and support, the geographies of recent refugee migrations, or immediate responses to specific refugee urgencies. This bias toward the “present” often obstructs the analysis of longer‐term trends of refugee migration as well as the factors explaining structural changes in these trends. Moreover, because of its more‐limited coverage in terms of years or its focus on asylum flows to Western countries, prior research has been insufficiently able to provide a more long‐term, global perspective.

To fill this gap, this paper makes a comprehensive analysis of trends and patterns of global refugee migration between 1951 and 2018. Drawing on the UN High Commissioner for Refugees (UNHCR) Population Statistics Database, its aim is to provide a global analysis of long‐term post‐WWII trends and patterns in terms of the intensity, geographic reach, and regional orientation of refugees migration. As such, this paper also goes beyond the usual focus on the number of refugees by studying how geographic patterns of refugee migration in terms of countries and regions of origin and destination have been changing. Such an analysis can help us to improve our understanding of the factors that could explain changing patterns of refugee migration. The UNHCR database covers refugee stocks—that is the number of refugees that are present in a country or region at a certain moment in time—and is therefore not suitable for studying short‐term refugee migration or circular flows. Yet “stock” data are used to assess long‐term refugee trends, which reflects the focus of this paper.

Although this study will refer to internally displaced persons or “IDPs” ¹ where relevant, the focus is on international refugees. ² The main reason for this is the limited availability of reliable long‐term data on IDPs. Although IDP statistics exist from 1969 onward (Schmeidl and Jenkins 1999, 2003), the data are less “systemized” than refugee statistics, in the sense that national governments are responsible for the reporting (Schmeidl 2000). Many IDPs move unrecorded within the boundaries of their nation‐states, which makes the coverage of IDP statistics generally less reliable than that of refugee statistics. Second, the focus on international refugees is warranted because of this study's focus on macrolevel, global shifts in the distribution of refugee populations.

The analysis in this paper builds on prior research that has given a number of valuable insights into the trends and drivers of refugee migration. First, earlier studies have shown that the scale of refugee migration has fluctuated considerably since the mid‐twentieth century (see Gatrell ²⁰¹³; FitzGerald and Arar ²⁰¹⁸) and that the vast majority of refugees in the world stay close to their countries of origin in what is often referred to as the “Global South” (see Puerto Gomez et al. ²⁰¹⁰). Most people fleeing civil war and human rights abuse are either forcibly displaced within their own country or seek refuge in a country nearby (Hatton 2020). In 2017, for instance, about 80 percent of refugees worldwide resided in countries neighboring their origin country and 85 percent were in developing countries (UNCHR ²⁰¹⁸). This seems to challenge the idea of refugees traveling greater distances; but, thus far, there has been no evidence of the evolution of these patterns over time.

Second, prior research indicates that fluctuations in refugee numbers are primarily determined by levels of conflict, oppression, and political stability in origin countries (Schmeidl 1997, 2001; Neumayer ²⁰⁰⁵; Moore and Shellman ²⁰⁰⁷). Hatton (2020) argued that the most important origin‐country variables determining asylum flow to Western countries are political terror and the lack of civil liberties; civil war seems to matter less, perhaps because war per se does not necessarily confer refugee status (see Hatton ^{2009, 2017, 2020}). Conflict also affects people's resources and livelihood opportunities and can have a destructive effect on the physical and economic infrastructure of countries, factors which may also prompt people to leave (Lindley 2010). In that sense, the socioeconomic context of the origin country, deteriorated by conflict, also plays a role in driving refugee migration.

Prior research has refuted the idea that refugee migration is, to any significant degree, directly driven by socioeconomic factors in destination countries. This is quite different from processes of labor migration, which tend to be primarily driven by destination‐country labor demand and recruitment (Piore 1979; Mayda ²⁰¹⁰; Ortega and Peri ²⁰¹³; Ambrosini ²⁰¹⁶; de Haas et al. ²⁰¹⁹). Such factors were found to play a relatively minor role in asylum migration to Western countries (Hatton 2020). This resonates with a statistical study on global refugee destinations by Moore and Shellman (2007), which showed that economic opportunities as well as political freedom in destination countries played a minor role in destination choices for refugees. They argued that “Instead of responding to greater wage opportunities and institutional democracy, people go where others have gone before them, usually crossing a nearby border” (Moore and Shellman, 2007, 831).

What does play a more significant role in destination choice, as with other forms of migration, is the location of family and community members already living abroad, with new refugees often following the beaten track (Lindley 2010; de Haas, Miller, and Castles ²⁰²⁰; Müller‐Funk ²⁰¹⁹). The most powerful variable determining the destination choice of asylum‐seekers is the population of previous migrants from the same origin country (Hatton 2020). The geographic distance between origin and destination countries has a strongly negative effect on the number of asylum applications, reflecting the greater costs and hazards associated with refugee migration (Hatton 2020). Another important factor in destination choice is access to resources. Generally, the socioeconomic background of forced migrants determines “routes taken, means of migration and destinations reached” (Van Hear ²⁰⁰⁶, 125) and, as such, those who flee from low‐income countries often have little influence on the destination choice. Fransen, Vargas‐Silva, and Siegel (2018) described how Burundians living close to the border crossed it on foot when conflict erupted, whereas those who lived farther from the border became internally displaced. These findings corroborate the common observation that the vast majority of refugees tend to stay in neighboring countries. Apart from reasons such as cultural familiarity and social proximity, a lack of resources enabling refugees to travel farther and pay intermediaries or smugglers to reach more faraway destinations is an important explanatory factor (see also Müller‐Funk ²⁰¹⁹).

For governments of destination countries, the greater degree of social identification with refugee populations from neighboring countries often makes it more difficult to close the borders and to prevent citizens from helping coethnic and coreligious people to cross them. Sometimes, nearby states welcome refugees from neighboring countries (at least initially) for various political, diplomatic, and economic reasons. For instance, in response to the influx of asylum seekers—mainly from Syria—to Europe in 2015 and 2016, Jordan leveraged its refugee‐hosting capacity in order to negotiate financial and other support from wealthy states in the “Global North” (Arar 2017), while Turkey was also able to use the presence of millions of Syrian refugees to its economic and diplomatic advantage (İçduygu and Üstübici 2014).

Asylum policies in destination countries often do matter but are generally of secondary importance in determining destination choice. For instance, in a statistical study on the determinants of refugee migration for 19 Western destination countries, Hatton (2009) found that the tightening of asylum policies between 2001 and 2006 reduced the number of asylum applications but that the policy restrictions only accounted for about a third of the decline in applications (Hatton 2009, 183). Analyses using quantified policy indexes show that border controls and restrictive asylum policies have a significant deterrent effect while welfare policies do not (Hatton 2004, 2009, 2017). Hatton (2020) argued that the 2016 agreement between the European Union and Turkey showed that arrivals of asylum seekers can be curbed drastically only through the application of draconian policies and in close cooperation with “transit” countries. There is also evidence that restrictive immigration policies can push prospective asylum seekers into an undocumented status (Czaika and Hobolth 2016) or re‐route migration to other destinations.

A new age of refugee migration?

Public discourse and scholarly analysis have recently suggested that levels and patterns of refugee migration have undergone quantitative and qualitative changes. Since 2010, refugee numbers seem to have surged and refugees are increasingly applying for asylum in Western countries. Governments and international organizations have regularly claimed that refugee numbers are increasing rapidly. In 2020, UNHCR reported that the global displaced population is “at a record high” with almost 80 million individuals (refugees, IDPs, and asylum seekers) being forcefully displaced by the end of 2019. It also claimed that the number of refugees (26 million) is “the highest ever seen.” ³ Such alarmist statements tend to be replicated by other international organizations, ⁴ are amplified by mass media, and have therefore fueled a certain perception that the world has entered a new age of refugee migration, characterized by intensified refugee movement as a result of conflict, oppression, and climate change and various forms of environmental havoc. Refugee movements as a result of “disasters” have been described as “one of the biggest humanitarian challenges of the 21st century.” ⁵

A related idea is that globalization—the growing interconnectedness of the world—has boosted the refugee movement. With falling costs of communication and travel, more and more refugees would find ways to flee their homelands to reach a wider variety of increasingly distant destinations, allegedly increasing pressure on Western asylum systems. Recent increases in refugee movements from countries such as Syria to Europe and from Central America to the United States seem to confirm this impression. For instance, Brell, Dustmann, and Preston (2020, 30) argued that “In coming years, the outflow of refugees from poorer regions of the world seems likely to continue undiminished, given the continued political fragility of populous and growing countries from which migration to safer locations is increasingly easy.”

Hatton (2020) suggested that increasing numbers of asylum applications in the Western world should be understood against the background of the evolution of international policy—such as the growing international alignment with the 1951 UN Refugee Convention and the EU's Common European Asylum System—which would have “provided clear incentives for spontaneous migration from poor, strife‐prone countries to the developed world” (2020, 91). Increasing asylum applications in Western countries would also be fueled by the growth of transit routes and migrant networks (Hatton 2020; see also Capps et al. ²⁰¹⁹).

However, there are reasons to question the validity of these claims as well as the extent to which the extrapolation of recent trends into the future is warranted. First of all, the recent increase in refugee arrivals in Europe and North America may reflect the simple fact that conflicts such as in Syria and Central America have occurred in geographic proximity to Western destinations. We should therefore not jump to the conclusion that this is part of some long‐term trend without properly investigating it first.

Second, a general analysis of long‐term trends in population mobility by Czaika and de Haas (2015) has questioned the idea that international migration has accelerated as a consequence of globalization. The authors found that, while, since WWII, the overall intensity of global migration had remained stable at levels of around 3 percent of the world population, the main changes in global migration were in terms of the geographic direction of migration flows. This is particularly seen in the demise both of Europe as a global source of settlers and migrants and South America as a global migration destination, as well as the emergence of Europe and the Middle East as new global migration destinations. Migrants from an increasingly diverse array of Asian, Latin American, and African countries have been increasingly concentrating in a small and shrinking pool of prime destination countries. Czaika and de Haas, therefore, concluded that the global migration map has become more skewed—with more countries converting from net immigration into net emigration countries—and that this reflects the asymmetric nature of globalization processes. This finding defies simplistic reasonings according to which globalization, in combination with global inequalities, would automatically result in intensified population movements. In line with this, the same study found that, between 1960 and 2000, the average distance traveled between origin and destination countries only increased modestly, from 3,000 to 3,700 km.

Although this analysis included all forms of migration—refugee‐ and nonrefugee‐related—it justifies a critical examination of the assumptions that “globalization” is leading to increasing refugee migration and that there is a structurally growing trend in the proportion of asylum seekers reaching Western countries. Such claims cannot be taken for granted and may, in fact, reflect a Western‐centric (research) bias—according to which most refugees and other migrants want to move to Western countries—and an unfounded extrapolation of recent events into the future. This gives reason to hypothesize that, in line with global migration trends, the main changes in refugee migration have been directional—geographic rather than in terms of intensity and distance. Civil wars are increasingly concentrated in a relatively small number of lower income countries that experience recurrent cycles of war (Blattman and Miguel 2010). Afghanistan, Burundi, and Sudan are notable examples of countries engulfed in this “conflict trap” (Collier and Sambanis 2002). From this, we can hypothesize that refugees increasingly originate from a small number of relatively poorer countries that are involved in protracted conflicts. However, the recent occurrence of conflicts in Libya, Syria, and Venezuela exemplifies the fact that civil conflicts are not unique to low‐income countries.

The perception of an increasing trend in refugee numbers may also have been influenced by the use of the post‐1951 period as a benchmark for our numerical analyses on refugee migration. This was the year in which the 1951 Convention Relating to the Status of Refugees was signed, which prescribes the rights and obligations of refugees and provides a formal definition of a refugee. ⁶ As a consequence of the convention, data collection on this type of migration was initiated by UNHCR and systemized over time. The absence of refugees as an internationally recognized legal category before 1951 explains the absence of systematic data on them from earlier periods. Yet only a cursory glance at historical sources is enough to show that large‐scale refugee movements are certainly not a new phenomenon (see Marrus ^{1985, 2002}; Skran ¹⁹⁹⁵; Zolberg ¹⁹⁸³; Zolberg, Suhrke, and Aguayo ¹⁹⁸⁹; and Gatrell ²⁰¹³ for historical overviews). Zolberg (1983), for example, describes how early‐modern state formation in Western Europe between the late‐fifteenth and the late‐seventeenth centuries led to various population displacements, totaling at least one million people. During World War I (1914–1918) an estimated 10 million Europeans were displaced, either internally or internationally (Gatrell 2007, 2008, 2013). Combined with the aftermath of the Balkan War (1912–1913), an estimated 12 million people were displaced after World War I (Gatrell 2013).

The Russian Revolution (1917–1920) prompted the displacement of 1 million people (Gatrell 2013). The Greek–Turkish War (1919–1922) displaced approximately 1.5 million people, while an estimated 465,000 Spanish refugees fled to France during the Spanish Civil War (1936–1939). The Second Sino‐Japanese War between China and Japan (1937–1945) led to the estimated displacement of 60–95 million people (MacKinnon ²⁰⁰⁸; Muscolino ²⁰¹⁰). An estimated 3–10 million people were killed as a consequence of Japanese aggression in Asia between 1931 and 1945 (Rummel 1998) and approximately 60 million Europeans were displaced during World War II (Proudfoot 1956). The Holocaust involved the systematic persecution, deportation, and murder of around 6 million European Jews by the Nazi regime (de Haas, Miller, and Castles ²⁰²⁰).

In the aftermath of World War II, the global displaced population had risen to 175 million—approximately 8 percent of the world population at the time and much higher than current levels even by absolute, let alone relative, numbers. The end of World War II witnessed the mass population movements in Europe of Holocaust survivors, displaced persons, and ethnic groups, such as the approximately 12 million ethnic Germans expelled as part of ethnic cleansing policies in Eastern Europe (Ther 1996). Postcolonial processes of state formation also reinforced displacement. In 1947, for example, when India and Pakistan gained independence from Britain, between 10 and 15 million Hindus and Muslims crossed the borders of the two new nation‐states after partition (Gatrell 2013).

Displacement was therefore at a record high in the mid‐twentieth century compared to both the early‐ and the late‐twentieth centuries (Gatrell 2013). These massive displacements were an important impetus for the establishment of organizations like the UNHCR and the Intergovernmental Committee for European Migration (the predecessor of the International Organization for Migration) in an effort to find more effective international responses to situations of human displacement (de Haas, Miller, and Castles ²⁰²⁰). This also provided the framework and incentive for the registration of refugee populations.

The fact that refugee records only started after the mid‐century peak and that early data were scant means that recent numbers may be inflated if they are not put into historical context. More generally, this begs the question of whether recent increases in refugee numbers as well as asylum applications in Western countries reflect a structural change in trends and patterns of global refugee migration, or a “normal” and therefore temporary response to the recent increase in conflict levels in particular countries, with refugee numbers likely to go down again after conflicts subside.

Data and methodology

One of the factors complicating long‐term, global analyses have long been the limited availability and quality of data, a situation which has recently been changing. A first longitudinal study in forced migration (Schmeidl 1997) created a dataset (1969–1990), extended a few years later by Moore and Shellman (2004) (1951–1995); in more recent years UNHCR picked up on these prior efforts by compiling the Population Statistics Database covering refugee data since 1951. UNHCR published its first annual statistical overview in 1994 (UNHCR ²⁰⁰⁰). Although far from perfect or complete, this database does allow us to conduct long‐term analyses of global refugee migration.

We use the UNHCR Population Statistics Database—which provides worldwide macrolevel statistics on the numbers of refugees—to study trends in refugee migration after 1951. ⁷ The database draws on various data sources, including survey data; census data; and registration data collected by local governments, UNHCR field offices, and NGOs. Although the UNHCR database is the most comprehensive one available, it has several limitations. Besides changing definitions and administrative practices across countries, the most important shortcoming which stands in the way of the analysis of refugee migration trends is the limited coverage in terms of older data for some countries. In order to avoid misinterpreting data, the paper will take these limitations into account and perform analyses using all available data as well as more‐limited data selections that only include countries for which data were available for the entire period we were interested.

For our analysis, we constructed indices that capture the intensity, spread, and distance of refugee migration, based on methodologies developed by Czaika and de Haas (2015) to analyze global migration trends. Combining these indices enables us to assess potential changes in the diversification of refugee migration over time. To this end, we studied the intensity, spread, and distance of refugee migration since 1951 (see Table 1). First, the intensity of refugee migration is defined as the refugee rate, which is measured as the number of refugees as a percentage of the world population. We study the intensity of refugee migration from an origin‐ as well as destination‐country perspective, by calculating the number of refugees as a share of origin‐ or residence‐country populations.

TABLE 1.

Dimensions of refugee migrations

Dimensions	Origin country perspective	Destination country perspective
Intensity	Refugee emigration intensity	Refugee immigration intensity
Spread	Refugee emigration spread	Refugee immigration spread
Distance	Refugee emigration distance	Refugee immigration distance

SOURCE: Czaika & De Haas (2015).

Second, the spread of refugee migration is defined as “the global spread of migrants across all possible bilateral (country‐to‐country) migration corridors” (Czaika and de Haas ²⁰¹⁵, 296–297). This measurement is based on the Hirschman–Herfindahl index (Hirschman 1964), a statistical measure that can be used to capture the concentration of migrants in bilateral country corridors. Following Czaika and de Haas (2015), we subtracted the score from 1 to gain a measurement of spread instead of concentration. The index provides a score between 0 and 1, with higher scores indicating a more equal spread of refugee migration across bilateral corridors. We also calculated the global refugee emigration spread, which is an indicator of the spread of refugees across origin countries, and the global refugee immigration spread, which measures the spread of refugees across destination countries, calculated as one minus the sum of squares of the share of (emigrant or immigrant) refugees of the global refugee population for each country i in the world:

$$ES=1-\sum _{i=1}^{n=226}{\left(\frac{E{M}_i}{M}\right)}^2$$

$$IS=1-\sum _{i=1}^{n=226}{\left(\frac{I{M}_i}{M}\right)}^2.$$

Finally, distance is measured as the geographic distance (in kilometers) between refugees’ origin and residence countries. We use the GeoDist database (Mayer and Zignago 2011) for the distance variable, which includes measures of bivariate distances between all countries in the world. We use the distance measurement calculated with “the great circle formula,” which uses bilateral distances between the most populated cities within countries. The relevance of absolute geographic distance may seem limited in a globalizing world in which cultural or legal barriers may be as important as, or more important than, geographic distance. Yet geographic distance still matters in terms of travel costs and the likelihood of crossing significant legal barriers (such as immigration restrictions) or the occurrence of cultural and social differences. As with the spread measure, in this paper, we distinguish between the average distances that refugees traveled from an origin‐ and a residence‐country perspective. This means that we calculated the average distance traveled for all refugees from each origin country as well as all refugees living in each destination country.

The intensity, spread, and distance of refugee migration after 1951

Global patterns: Intensity, spread, and distance

To analyze the evolution of refugee migration over recent years, we calculated global refugee populations (“stocks”) both in absolute terms and relative to the world population between 1951 and 2018 (Figure 1). For comparative purposes, we also added data on the number (or “stock”) of IDPs from 1969 onward. The early IDP data (1969–2000) are derived from the Global Refugee Project (Schmeidl and Jenkins 1999). The statistics are estimates based on annual UNHCR and US Committee for Refugees and Immigrants (USCRI) reports. Later IDP statistics (after 2000) are derived from the UNHCR Population Database and are based on data from the Internal Displacement Monitoring Centre (IDMC). Throughout this section, we use the term “global displacement” when we refer to both IDPs and refugees.

FIGURE 1.

Global displacement: 1951–2018NOTES: Refugee data are derived from the UNHCR Population Statistics Database. Data for IDPs for the 1969–2000 period are derived from Schmeidl and Jenkins (1999). Data for IDPs from 2000 onward are derived from the UNHCR database and provided by IDMC.

Figure 1 indicates that global displacement has increased drastically over the past five decades. This corroborates common perceptions of soaring displacement. Whereas, in 1951, approximately 1.8 million displaced persons were reported, numbers had risen to 44 million by the early 1990s, after which global displacement decreased again in the early 2000s, followed by a peak after 2005. This decrease coincides with the above‐mentioned shift in data sources used for the IDP statistics. It is not clear exactly why the IDMC statistics are much lower than those of USCRI, but this is likely related to differences in measurement or the inclusion versus exclusion of certain populations. This highlights the sensitivity of forced displacement statistics to changes in definitions or measurements and the importance of interpreting such data with the greatest possible care.

By 2018, the figure of the global displaced had risen to 62 million. However, Figure 1 also shows that most of this statistical increase since the early 2000s is driven by a sharp increase in IDP numbers—those included in UNHCR statistics increased sharply from 4.2 million in 2003 to 41.4 million by 2018. A large part of the increase in IDP numbers seems to have been driven by improved statistical coverage for this group as well as a looser definition of who belongs to this category. For instance, in 2007, individuals in “IDP‐like situations” were added to the number of IDPs and removed from the “others of concern” group in which they used to be classified. There were 12 countries included in the data in 2003; this increased to 31 countries by 2018. In that same period and notwithstanding periodic fluctuations, international refugee numbers have remained fairly stable. Refugee numbers decreased between 1993 and the early 2000s, from roughly 16 to 9 million. After the early 2000s, refugee numbers increased to 20.4 million by 2018. Another important observation is that refugee and IDP statistics follow the same trends and fluctuations over time.

To measure the intensity or relative magnitude of refugee migration, Figure 1 also depicts global displacement as a percentage of the world population between 1951 and 2018. This analysis reveals that refugee statistics have fluctuated at levels of between 0.1 and 0.3 percent of the world population over the past seven decades and that the intensity of refugee migration has gone down since the 1990s despite an upturn in recent years. Even though the total number of displaced as a share of the world population has increased from 0.10 to 0.27 percent since the early 2000s, this increase is mainly driven by the inclusion of IDP data. Between the early 2000s and 2018, the number of IDPs as a percentage of the world population increased from 0.07 to 0.5 percent, while that of refugees relative to the world population decreased from 0.33 percent in 1992 to 0.27 percent in 2018. Another reason to nuance the idea of a global increase in refugee populations is the fact that Palestinian refugees—a total of 904,122 in 1951—are excluded from UNHCR statistics, as they formally fall under the protection of the United Nations Relief and Works Agency for Palestine Refugees.

Overall, Figure 1 refines the idea that the intensity of refugee migration has increased. When looking solely at refugees, the current intensity of refugee migration is indeed higher than in the 1950s; however, it is also significantly lower today than in the 1990s, when it peaked at 0.33 percent. Taking a more long‐term perspective, the picture becomes even more nuanced. As mentioned above, 175 million people—approximately 8 percent of the world population—were displaced worldwide in the aftermath of World War II (Gatrell 2013). This is significantly more than the 62 million people—0.8 percent of the world population—who were displaced in 2018.

The common perception that refugee movements are rapidly increasing can also be linked to another assumption—the idea that the level of violent conflicts has been increasing. However, this idea also seems to be based on perception rather than fact because, since WWII, the level of violent conflicts and government oppression seems to have shown a decreasing rather than an increasing trend. For instance, the intensity of conflicts as measured by the number of battle‐related deaths in state‐based conflicts has shown a decreasing trend (see Figure 2). From a peak at 550,000 in the annual number of deaths around 1949–1950 (Chinese and Greek civil wars and the Korean War), the number of battle‐related deaths dropped to 30,000 in 1955 to rise to another peak of 290,000 in 1972 (Vietnam War) and 240,000 in 1984 (Iran–Iraq War, Afghanistan War). Since 1989, the annual number of battle‐related deaths has never exceeded 100,000 in 25 years, not even during the Bosnian War (1992–1995). After reaching a peak at around 80,000 in 1999–2000 (DRC war and the Ethiopia–Eritrea war), the number dropped to an absolute low of 21,000 in 2010. Recent conflicts in Syria, the Middle East, and South Sudan made numbers rise to new peaks in 2014 (105,000) and 2017 (134,000) but they have dropped again recently. If we look at the nature of political regimes, we also see a clear post‐WWII trend toward decreased autocratic governance. While, in 1946, the share of the world population living in democratic countries hovered at levels from 30 and 35 percent between 1947 and 1982, it rose to 50 percent in 1992, increasing again to 57 percent in 2005–2006. Although this percentage has been stagnating at levels of around 55 percent in recent years, the long‐term trend is clearly toward democracy, although it is too early to tell whether recent attacks on democracy will herald a future increase in autocracy. ⁸

FIGURE 2.

Number of battle‐related deaths in state‐based conflicts: 1946–2018NOTE: UCDP/PRIO Armed Conflict Dataset, Uppsala Conflict Data Program, and Peace Research Institute Oslo.

Such data give reason to think that earlier UNHCR data underrepresented true refugee numbers. Research has shown that conflict intensity, measured by the number of battle‐related deaths, is positively correlated to levels of forced migration (Melander, Öberg, and Hall ²⁰⁰⁹; Turkoglu and Chadefaux ²⁰¹⁹). Yet, in the early 1950s, the number of battle‐related deaths in state‐based conflicts was at an all‐time high, while reported refugee stocks (Figure 2) were relatively low. The number of global battle‐related deaths also reached a peak in the 1970s, a figure which is not mirrored in the refugee data, probably because many of the displaced were unaccounted for.

The poor geographic coverage of older refugee statistics is another reason to question the idea that the intensity of refugee migration has increased. The collection of refugee statistics has improved significantly over time, particularly after 1990 (Crisp 1999), and an increasing number of countries have been included in UNHCR data. A long‐term analysis of refugee trends (as in Figure 1) can therefore be potentially misleading because statistics are not comparable between years. The long‐term increase in refugee migration up to 1990, as suggested by Figure 1, may therefore largely reflect improvements in measurement rather than a real increase. The collection of reliable IDP statistics presents an even greater challenge to governments and humanitarian organizations because IDPs often move unregistered and tend to have shorter cycles of displacement. There is also a lack of clear definition by agencies and governments of what actually constitutes an “IDP” (see, e.g., Bennett ¹⁹⁹⁸). Even more so than in the case of refugees, rising IDP numbers may therefore reflect data improvements rather than actual changes.

To illustrate improved data coverage over time, Table 2 gives an overview of refugee data provided by the UNCHR Population Database between 1951 and 2018. The database contained information on 21 countries in 1951, increasing to 76 in 1970, 114 in 1980, 147 in 1990, 194 in 2000, 211 in 2010, and 216 in 2018. For the early decades, refugee data were unavailable for many countries, particularly those outside the Western world. In 1951, the 10 countries with the highest numbers of refugees included in the UNHCR database are the United States, France, Austria, Germany, the UK, Australia, Canada, Belgium, Sweden, and Hong Kong. ⁹ As described before, data on approximately 900,000 Palestinian refugees are absent from the 1951 data; including this group would increase the number of reported refugees in that year from 1.8 to 2.7 million.

TABLE 2.

Data overview: 1951–2018

	1951	1960	1970	1980	1990	2000	2010	2018
Net refugee‐hosting countries	21	19	50	76	88	99	95	106
Net refugee‐origin countries	−	1	25	38	59	92	104	86
Countries with equal numbers of refugees who left and who are hosted	−	−	1	−	−	3	12	24
Total number of countries	21	20	76	114	147	194	211	216

NOTE: Based on data from the UNHCR Population Statistics Database.

Moreover, in the 1950s, information on origin countries is missing. The tenfold increase in the number of countries covered between 1970 and 2018 shows that the UNHCR database should be used with caution. The number of net refugee‐origin countries increased between 1990 and 2018, which seems to reflect an increase in civil wars in the 1990s after the end of the Cold War (Blattman and Miguel 2010). The 1990s were a particularly volatile decade, with insurgencies arising in various countries such as Algeria, Afghanistan, Burundi, Liberia, Rwanda, Sierra Leone, and the former Yugoslavia. This confirms that fluctuations in the global refugee population primarily reflect those in the incidence of conflicts for more‐recent decades. Because of the better coverage and reliability of recent data, subsequent analyses will focus on the 1980–2018 period.

Global emigration and immigration intensity

In Figure 3, we make a distinction between the intensity of refugee migration from both origin‐ and destination‐country perspectives. The emigration intensity is calculated as the average percentage of refugees relative to the origin‐country populations, whereas the immigration intensity is calculated as the average percentage of refugees relative to the destination‐country populations, for each year and for a fixed set of countries (69 destination countries and 51 origin countries) for which we have data for each year since 1980. The figure shows a relatively stable trend in refugee immigration intensity over time. Between 1980 and 2018, refugees constitute on average between 0.3 percent and 1 percent of the destination‐country populations.

FIGURE 3.

Refugees relative to origin‐ and destination‐country population (%): 1980–2018NOTES: Refugee data are derived from the UNHCR Population Statistics Database. The fixed set of countries includes 69 destination countries and 51 origin countries for which we have data for each year since 1980. The emigration intensity is calculated as the average percentage of refugees relative to the origin‐country populations, whereas the immigration intensity is calculated as the average percentage of refugees relative to the destination‐country populations.

The highest peak in refugee migration intensity over the 1980–2018 period occurred in the 1990s—which saw increasing civil conflict around the world, such as in the Balkans, the Horn of Africa, and the African Great Lakes region—and after 2010, when conflict broke out or intensified in countries like Iraq, Libya, Syria, and Yemen, alongside continuing conflict in, inter alia, Somalia and Sudan. In 2018, the top five countries hosting the greatest percentages of refugees relative to their total populations included Lebanon, Jordan, Nauru, Turkey, and Chad. Taking an origin‐country perspective, the intensity of refugee emigration varied between 2.5 and 3.5 percent of origin‐country populations between 1980 and the mid‐1990s. Between 2000 and 2010, these levels dropped to around 1.5 percent. After 2012 the emigration intensity went up again. In 2018, countries with the highest refugee emigration intensity were Syria, South Sudan, Afghanistan, Somalia, and Burundi. A global overview (maps) of the origin and destination countries n 2018 can be found in the Appendix.

However, the most important finding is that the post‐1980 increase in refugee migration suggested by Figure 1 is not reproduced in this analysis. If we use a time‐invariant set of countries, the pattern is one of (conflict‐related) fluctuation rather than an overall increase. What appeared to be an increase in refugee migration reflects the growing number of countries included in UNHCR data rather than a real increase in refugee migration.

The global spread of refugee migration

Table 3 shows the top 15 refugee‐destination and refugee‐origin countries in 1980, 1990, 2000, 2010, and 2018, both in absolute numbers (in millions) and as shares of global refugee numbers. Refugees concentrate in a relatively small number of destinations, although the level of concentration has undergone some changes. The percentage of global refugees residing in the top 15 refugee‐hosting countries decreased from 84 to 72 percent between 1980 and 2010, after which it increased slightly again between 2010 and 2018. However, 75 percent is still 10 percentage points lower than in the 1980s. This is a first indication that the global spread or diversification of refugee migration in terms of destination countries has been increasing.

TABLE 3.

Top 15 refugee‐hosting and refugee‐sending countries: 1980–2018

	1980			1990			2000			2010			2018
Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population
Countries of residence
Somalia	2.00	24.13	Iran	4.17	24.00	Ethiopia	0.20	1.63	Pakistan	1.90	18.02	Turkey	2.87	18.46
Pakistan	1.43	17.23	Pakistan	3.26	18.72	Kenya	0.21	1.70	Iran	1.07	10.17	Pakistan	1.35	7.04
DRC	0.61	7.37	Sudan	1.03	5.93	Uganda	0.24	1.95	Syria	1.01	9.53	Uganda	1.01	5.85
Sudan	0.49	5.95	Malawi	0.93	5.33	Zambia	0.25	2.07	Germany	0.59	5.63	Sudan	0.98	5.41
United States	0.40	4.87	Germany	0.82	4.69	Armenia	0.28	2.31	Jordan	0.45	4.27	Germany	0.94	5.33
Iran	0.33	3.98	Ethiopia	0.77	4.45	China	0.29	2.42	Kenya	0.40	3.82	Iran	0.79	4.91
Australia	0.30	3.67	United States	0.46	2.67	DRC	0.33	2.74	Chad	0.35	3.30	Lebanon	0.69	4.76
China	0.26	3.17	Somalia	0.46	2.65	Sudan	0.41	3.42	China	0.30	2.85	Bangladesh	0.67	4.55
Thailand	0.26	3.15	DRC	0.42	2.39	Guinea	0.43	3.52	United States	0.26	2.51	Ethiopia	0.45	4.53
Burundi	0.23	2.83	Mexico	0.36	2.05	Serb. & Kos.	0.48	3.99	UK	0.24	2.26	Jordan	0.45	3.59
Tanzania	0.16	1.93	Guinea	0.33	1.87	United States	0.51	4.19	Bangladesh	0.23	2.17	DRC	0.42	2.65
UK	0.15	1.79	China	0.29	1.65	Tanzania	0.68	5.61	Venezuela	0.20	1.91	Chad	0.39	2.26
Cameroon	0.13	1.54	Costa Rica	0.28	1.59	Germany	0.91	7.47	France	0.20	1.90	Kenya	0.38	2.11
France	0.12	1.47	Cote d'Ivoire	0.27	1.57	Iran	1.87	15.40	Yemen	0.19	1.80	Cameroon	0.32	1.91
Uganda	0.11	1.36	Burundi	0.27	1.54	Pakistan	2.00	16.50	India	0.18	1.75	France	0.30	1.85
Total/Cum. %	7.00	84.43		14.10	81.11		9.09	74.94		7.59	71.90		15.00	75.21
Countries of origin
Ethiopia	2.57	30.98	Afghanistan	6.34	36.45	Afghanistan	3.59	29.57	Afghanistan	3.05	28.95	Syria	6.65	33.37
Afghanistan	1.73	20.93	Ethiopia	1.35	7.74	Various/unknown	1.12	9.23	Iraq	1.68	15.96	Afghanistan	2.68	13.45

	1980			1990			2000			2010			2018
Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population	Country	Net refugee stock (in million)	Percentage of total refugee population
Various/unknown	1.42	17.18	Mozambique	1.25	7.18	Burundi	0.57	4.68	Somalia	0.77	7.30	South Sudan	2.29	11.46
Angola	0.45	5.42	Iraq	1.13	6.52	Iraq	0.53	4.34	DRC	0.48	4.52	Myanmar	1.15	5.74
Vietnam	0.34	4.15	Liberia	0.74	4.23	Bosnia and Herzegovina	0.50	4.16	Myanmar	0.42	3.94	Somalia	0.95	4.76
Rwanda	0.29	3.46	Sudan	0.52	3.01	Sudan	0.49	4.08	Colombia	0.40	3.75	Sudan	0.72	3.63
Uganda	0.22	2.71	Vietnam	0.50	2.85	Somalia	0.48	3.92	Sudan	0.39	3.67	DRC	0.72	3.61
Chad	0.22	2.67	Somalia	0.47	2.70	Angola	0.43	3.58	Vietnam	0.34	3.21	CAR	0.59	2.96
Cambodia	0.19	2.33	Angola	0.41	2.34	Sierra Leone	0.40	3.32	Eritrea	0.22	2.11	Eritrea	0.51	2.54
Burundi	0.17	2.05	Rwanda	0.36	2.08	Eritrea	0.38	3.11	China	0.18	1.75	Burundi	0.39	1.94
DRC	0.12	1.41	Sri Lanka	0.21	1.20	DRC	0.37	3.06	Serbia and Kosovo	0.18	1.74	Iraq	0.37	1.87
Lao P. D. Rep.	0.11	1.27	Burundi	0.19	1.10	Vietnam	0.37	3.06	Various/unknown	0.17	1.58	Vietnam	0.33	1.68
El Salvador	0.09	1.11	Chad	0.18	1.06	Croatia	0.34	2.76	CAR	0.16	1.56	Nigeria	0.28	1.39
Philippines	0.09	1.09	Western Sahara	0.17	0.95	Azerbaijan	0.28	2.34	Turkey	0.15	1.39	Rwanda	0.25	1.24
Namibia	0.06	0.67	Lao P. D. Rep.	0.14	0.78	Liberia	0.27	2.20	Sri Lanka	0.14	1.34	Various/unknown	0.22	1.09
Total/Cum. %	8.07	97.40		13.95	80.20		10.12	83.42		8.73	82.77		18.09	90.74

NOTE: Refugee data are derived from the UNHCR Population Statistics Database.

From an origin‐country perspective, trends show more variation over time. In 1980, the top 15 refugee‐origin countries represented 97 percent of all refugees included in the UNHCR database. Between 1980 and 1990, this percentage decreased to around 80 percent, after which it increased again to 83 percent in 2000 and 2010, to peak in 2018 at 91 percent. Overall, the data analysis shows a higher level of diversification in terms of destination compared to origin countries. While the bulk of the global refugee population has come from a relatively small number of countries (Syria, Afghanistan, South Sudan, Myanmar, Venezuela, and Somalia in 2018, for instance), the number of destination countries seems to have been increasing and diversifying. Another trend is that more low‐ and middle‐income countries enter the top 15 refugee‐hosting countries. This could, however, be due to more low‐ and middle‐income countries starting to collect statistics on refugee populations in recent years. So, this trend may partly be a statistical artifact.

To provide a statistical measure for the geographic distribution of refugee migration in terms of both origin and destination countries, we use the (inversed) Hirschman‐Herfindahl Index. Figure 3 shows the global spread of refugee migration between 1980 and 2018, as well as the spread of refugees in terms of origin and destination countries. As described, higher scores indicate a more equal spread of refugee migration across origin and destination countries, respectively.

Figure 4 shows that the global refugee population has spread more equally across bilateral corridors. This increase particularly occurred in the 1986–1991 period, after which it stabilized at values of around 0.95 from the early 1990s. The global spread or distribution of refugees is lower than that of nonrefugee migrants—a level which, as reported by Czaika and de Haas (2015, 296), hovered at around 0.99 between 1980 and 2000. This shows that refugees tend to be more concentrated in particular countries of origin and destination. The increasing global spread of refugees seems to be particularly driven by a post‐1980 diversification of destination countries, particularly in more recent years. Trends for the diversity of refugee migration in terms of origin countries are quite different and less linear. After a period of relative stability in the 1980s, the global refugee‐origin spread increased rapidly between 1980 and the mid‐1990s. This reflected the outbreak of new conflicts in the post–Cold War era—particularly in the Balkans, the Horn of Africa, and the African Great Lakes region. However, after 1995 the origin‐country diversity decreased, reflecting a drop in the number of prime refugee‐producing countries. The refugee destination‐country spread is consistently higher than the emigrant spread and the gap seems to have widened between the early 1990s and 2000. To check whether this reflected the inclusion of more countries in recent data, we ran the analysis with a fixed set of origin and destination countries for which we have data for all years since the 1980s. Using these data, we find the same trends as the analysis using the nonfixed dataset.

FIGURE 4.

Global spread of refugees: 1980–2018NOTES: Refugee data are derived from the UNHCR Population Statistics Database. The fixed set of countries includes 69 destination countries and 51 origin countries for which we have data for each year since 1980.

Global distances traveled by refugees

Figure 5 shows the average geographic distance between refugees’ country of destination and country of origin (in kilometers) for each year since 1980. These calculations have been weighted by the number of refugees in each corridor containing refugees. To crosscheck the validity of these findings, we also included this measurement for a fixed set of origin countries from 1980 to 2018. The results show that the average geographic distance between origin and destination country has increased over time but that the trends are nonlinear. This trend is visible using measurements both of all countries and the fixed set of origin countries. Between 1980 and 2018, the average distance that refugees traveled increased by 40 percent, from approximately 1,000 km to around 1,500. These increases are most apparent in the 1990s and 2000s. The sharp drop after 2006 seems to be largely driven by the inclusion of people in “refugee‐like” situations in UNHCR refugee numbers since 2007, ¹⁰ among whom there are many Colombians and Haitians who are residing in the Americas. This observation shows how changes in definitions can bias aggregated figures.

FIGURE 5.

Average geographic distance (in km): 1980–2018NOTES: Refugee data are derived from the UNHCR Population Statistics Database. The fixed set of countries includes 69 destination countries and 51 origin countries for which we have data for each year since 1980. Calculations have been weighted by the number of refugees in each corridor containing refugees.

On average, refugees travel shorter distances than “regular” migrants. Czaika and de Haas (2015) found that migrants traveled on average 3,657 km in 2000. This was a 16 percent increase compared to 1980 when the average distance was 3,128 km. This may reflect the fact that refugees prefer to stay close to home and more often lack the resources to travel long distances and acquire essential paperwork such as passports and visas. Although refugees travel shorter distances than regular migrants, the rate with which the average distance increased over time is higher than for “regular” migrants.

Regional patterns: Intensity, spread, and distance

Figure 6 displays the distribution of refugees by region of origin and destination between 1980 and 2018. Most refugees originated from Asia and Africa, particularly in the 1990s and late 2010s. As shown in Table 3, nine of the top 15 origin countries in the 1990s were African countries (Ethiopia, Mozambique, Liberia, Sudan, Somalia, Angola, Rwanda, Burundi, and Chad), while the peak in Asia mainly reflects enduring conflicts in Afghanistan, Iraq, and Vietnam. Together, these countries accounted for 46 percent of the global refugee population in 1990. In 2018, approximately 10 and 7 million refugees originated from Asia and Africa, respectively. In that year, 32 percent of the world refugee population originated from Syria, followed by 15 percent from Afghanistan. The third most important origin country in Asia was Myanmar, which mirrors the increasing outflows of Rohingya refugees. Recent increases in refugee numbers from Africa largely reflect increasing refugee outflows from South Sudan, Eritrea, and the Central African Republic. Europe and the Americas have mostly been refugee‐receiving regions, although a small peak can be observed in Europe in the early 1990s, which was a result of refugees from the former Yugoslavia moving to Western Europe as well as Albanians arriving in Italy.

FIGURE 6.

Refugee emigration (top) and immigration (bottom) by region (in millions)NOTES: Refugee data are derived from the UNHCR Population Statistics Database. Refugee statistics are presented by region in absolute numbers.

Figure 6 also shows that regions with high refugee outmigration (particularly West Asia and East and Central Africa) also host relatively large refugee populations, indicating that most refugees reside within their origin region. For example, 6 million refugees originated from Africa in 2018, while the continent also hosted 5.5 million refugees in that same year. The main refugee‐hosting countries in Africa in 2018 were Uganda and Ethiopia, neighboring countries to refugee‐producing ones such as South Sudan, Somalia, and the Democratic Republic of the Congo (). The same pattern can be observed for Asia: 10 million refugees originated from the region, which hosted 8.6 million refugees. Analyses of country‐level statistics confirmed that most Asian refugees are hosted in the region: Syrian refugees mainly reside in Lebanon, Jordan, and Turkey, many Afghan refugees are in “protracted” situations in Iran and Pakistan, and many Vietnamese refugees are in a protracted situation in China. Countries in Europe and the Americas tend to host more refugees than they “produce.”

Focusing on the regional refugee emigration intensity, Figure 7 reveals that Asia and Africa also have the highest refugee emigration intensity. In the 1990s, this peaked in Africa and Asia, reaching between 0.94 and 0.26 percent, respectively. However, refugee emigration intensity declined significantly in later years, almost reaching that of Europe, Oceania, and the Americas in 2010, with intensity in Africa reaching levels of 0.28 percent. Between 2010 and 2018, refugee migration from Asia and Africa increased again in the wake of violent conflicts in Syria, South Sudan, and Somalia. The refugee emigration intensity in Europe showed a peak in the 1990s following a conflict in the former Yugoslavia and the wake of the collapse of the Soviet Union. Compared to other world regions, the refugee emigration intensity in the Americas and Oceania has been comparatively low and stable over time.

FIGURE 7.

Refugee emigration (top) and immigration intensity (bottom) by region (%)NOTES: Refugee data are derived from the UNHCR Population Statistics Database. The intensity of refugee migration is defined as the refugee rate, which is measured as the number of refugees (immigrants or emigrants) as a percentage of the region's population.

Figure 7 also shows the refugee immigration intensity for each region. From the mid‐1980s onward, Africa has consistently had the highest intensity. Europe is the second most important region in terms of the number of refugees relative to its population, reflecting its geographic proximity to conflict zones within both Europe and the Middle East. Particularly in the mid‐1990s and after 2015, relative refugee immigration intensity increased in Europe. After the Syrian conflict erupted in 2011, the number of refugees as a percentage of Europe's population increased from 0.22 in 2010 to 0.37 in 2018, reaching levels similar to those in the early 2000s. Although Asia hosts the most refugees in absolute terms, in relative terms refugee immigration intensity is less significant, despite the outbreak of conflict in Syria, after which many refugees settled in Turkey, Jordan, and Lebanon. The Americas and Oceania consistently have the lowest refugee immigration intensity. ¹¹

Regional destination patterns

The previous figures revealed that regions with high refugee outmigration also tend to host the most refugee immigrant populations. Table 4 shows that 83.46 percent of refugees resided in their region of origin in 2018. The global percentage shows a steady decline from the 1980s, when 98.27 percent of global refugees resided in their origin region, after which the figure steadily declined to 96.08 percent in 1990, 89.15 percent in 2000, and 84.13 percent in 2010. Although the majority of refugees continue to reside in their region of origin, these statistics suggest diversification of destinations and further travels of refugees over time.

TABLE 4.

Regional origins of refugees by residence region: 1980–2018

Residence region	Origin region	1980	1990	2000	2010	2018
Africa	Africa	99.74	99.79	99.47	95.86	94.96
	Americas	0.00	0.00	0.00	0.00	0.00
	Asia	0.00	0.07	0.41	3.63	5.02
	Europe	0.00	0.00	0.01	0.00	0.00
	Oceania	0.00	0.00	0.00	0.00	0.00
	Various/Unknown	0.26	0.13	0.11	0.51	0.02
	Stateless	0.00	0.00	0.01	0.00	0.00
Americas	Africa	0.00	0.71	17.00	12.76	15.23
	Americas	14.59	8.39	10.77	57.43	43.20
	Asia	0.00	14.41	26.87	23.74	28.67
	Europe	0.00	8.80	29.86	4.36	4.00
	Oceania	0.00	0.00	0.06	0.22	0.04
	Various/unknown	85.41	67.69	15.31	1.12	8.74
	Stateless	0.00	0.00	0.14	0.38	0.12
Asia	Africa	4.14	0.04	1.20	3.84	2.87
	Americas	0.00	0.00	0.00	0.00	0.00
	Asia	95.75	99.56	98.40	96.06	97.11
	Europe	0.00	0.00	0.40	0.10	0.02
	Oceania	0.00	0.00	0.00	0.00	0.00
	Various/unknown	0.11	0.39	0.01	0.00	0.00
	Stateless	0.00	0.02	0.00	0.00	0.00
Europe	Africa	0.12	1.28	6.55	19.81	21.04
	Americas	0.38	1.30	0.44	0.75	0.48
	Asia	0.56	6.64	15.25	40.54	62.52
	Europe	2.92	3.71	35.03	28.91	8.43
	Oceania	0.00	0.00	0.00	0.01	0.00
	Various/unknown	96.03	84.47	42.52	9.15	5.70
	Stateless	0.00	2.60	0.21	0.84	1.82
Oceania	Africa	0.00	0.00	12.69	7.14	8.66
	Americas	0.00	0.00	1.09	0.73	0.30
	Asia	0.32	6.46	48.25	87.88	81.69
	Europe	0.00	0.00	34.57	1.39	0.30
	Oceania	0.00	0.00	0.03	0.82	1.34
	Various/unknown	99.68	93.54	3.16	1.00	1.27
	Stateless	0.00	0.00	0.20	1.04	6.45
Global ‐ refugees residing in their origin region		98.28	96.08	89.15	84.13	83.46

NOTES: Refugee data are derived from the UNHCR Population Statistics Database.

However, there are large differences across regions. Table 4 shows that the vast majority of refugees residing in Africa and Asia also originate from these regions, with regional refugee retention rates ranging from 94 to 100 percent, respectively. In 2018, Sudan, Uganda, and Ethiopia were the main refugee‐destination countries in Africa, with refugees mostly originating from South Sudan. In Asia, the top three refugee‐hosting countries were Turkey, Pakistan, and Iran, hosting mainly refugees from Syria and Afghanistan. The situation is quite different in other world regions. In Europe, there has been a clear shift in the regional origins of refugees over time. In 2000, Europe mainly hosted refugees from other European countries, such as Bosnia and Herzegovina, Serbia and Kosovo, and Croatia; however, over time, the shares of refugees originating from Asia (Afghanistan and later Syria) and Africa increased. In 2018, approximately 83 percent of refugees residing in Europe came from other continents. In that same year, only 2 percent of refugees residing in Oceania originated from there. Oceania, particularly Australia, mainly hosts refugees from Asia.

Background analysis revealed that the stark increase in the number of refugees in the Americas originating from the region—from 14 percent to 59 percent between 2006 and 2007—is a statistical artifact reflecting the aforementioned inclusion of intra‐regional migrants in “refugee‐like” situations who were previously excluded. In 2000, the Americas still hosted a sizeable number of refugees from Europe, predominantly people who fled the wars in the former Yugoslavia. The largest group of refugees in the Americas, however, continues to be from the Americas and the Caribbean—particularly from Colombia, Venezuela, Haiti, El Salvador, and Guatemala. Asian refugees residing in the Americas were mostly from China.

Region‐specific distance covered by refugees

One of the general assumptions in the literature is that refugees travel increasingly long distances because of the falling costs of travel and communication. While this was confirmed on a global scale (Figure 4), region‐specific trends can help us to further understand the factors explaining these global tendencies. Table 5, therefore, examines the average distances traveled by refugees originating from different world regions. These patterns reveal that, at a regional level, average distances between origin and destination countries increased between the 1980s and the 2000s, slightly decreasing after 2010.

TABLE 5.

Average distance traveled from and to different regions (in km): 1980–2018

	1980	1990	2000	2010	2018
From origin region
Africa	1,059	1,112	1,692	2,266	1,650
Americas	1,008	2,363	3,228	1,745	2,460
Asia	890	1,429	1,676	1,698	1,558
Europe	1,867	5,810	2,239	1,831	1,898
Oceania	−	12,786	12,488	11,511	5,370
To residence region
Africa	1,029	1,070	1,104	1,322	1,144
Americas	789	8,270	9,013	5,235	6,389
Asia	925	1,027	1,030	1,007	959
Europe	3,707	4,381	3,304	3,894	3,915
Oceania	4,464	4,464	12,549	9,471	10,581

NOTES: Refugee data are derived from the UNHCR Population Statistics Database. Numbers are rounded.

On average, refugees originating from Oceania travel the longest distances, although these fluctuate substantially over time. The long‐distance traveled from Oceania is mostly because of high shares of refugees originating from Fiji who are resettled in Canada and the United States. Refugees from Europe traveled the second‐longest distances in the 1990s, which mainly reflects refugees from the former Yugoslavia who moved to the United States. After the 1990s, the average distance that refugees traveled from European countries decreased and remained stable at around 2,000 km, which is comparable to the average distances that refugees traveled from countries in Asia, Africa, and the Americas in later years.

Refugees traveling to Oceania and the Americas crossed the longest distances by far. The United States, Australia, and New Zealand host relatively large populations of resettled refugees. Even though resettlement numbers are small, this might have influenced the results upward because the distances crossed are large. For example, the largest group of resettled refugees in the United States traditionally consisted of refugees originating from the DRC, whereas in more recent years the United States hosted relatively large groups of Ethiopians, Cameroonians, Somalians, and Guineans, many of them coming through resettlement programs.

The average distance that refugees travel from their origin countries to Europe is stable over time and hovers at around 4,000 km. This finding corresponds with the earlier finding that the vast majority of refugees residing in Europe originate from other continents. Refugees residing in Africa and Asia traveled the shortest distances, between approximately 1,000 and 1,200 km, with the pattern remaining stable over time. This finding aligns with the general observation that most refugees originating from these regions tend to reside in neighboring countries in the same region.

Refugees from and to low‐, middle‐, and high‐income countries

We now turn to country‐level analyses to test the idea that refugees increasingly come from low‐income countries because recurrent cycles of conflict tend to concentrate on a limited number of them. We grouped the origin countries in our data into three equally sized groups based on their GDP (see Figures 8 and 9). The findings confirm that most refugees originate from low‐income countries and that these numbers have increased over time. Countries with low GDP also have the largest average numbers of refugees residing abroad relative to their populations, although the pattern fluctuates over time (Figure 8). The majority of the world's refugees are also hosted by low‐income countries. This was particularly the case in the 1990s when approximately half of the world's refugees resided in the 30 poorest countries. Relative to their populations, countries in the low and medium GDP group host most refugees as well. Particularly in the 1990s and after 2010, medium GDP countries host significant shares of refugees. Both figures, however, clearly show that high‐income countries share the lowest “burden” in terms of hosting the world's refugee populations, although differences across country income groups have decreased since the early 1990s.

FIGURE 8.

Refugees by origin and destination country, GDP per capita (absolute numbers)NOTES: Refugee data are derived from the UNHCR Population Statistics Database. Data on GDP per capita is derived from the World Bank Development Indicators. Countries were grouped into three equally sized groups (low, medium, and high GDP) each year. Refugee statistics are presented in absolute numbers

FIGURE 9.

Refugees by origin and destination country, GDP per capita level (%)NOTES: Refugee data are derived from the UNHCR Population Statistics Database. Data on GDP per capita is derived from the World Bank Development Indicators. Countries were grouped into three equally sized groups (low, medium, and high GDP) each year.

Conclusion

This paper has studied the long‐term trends and patterns of global refugee migration, with particular emphasis on the post‐WWII period. Importantly, the analysis did not detect a long‐term increase in global levels of refugee migration—refuting conventional wisdom, the long‐term levels have been quite stable over time. Primarily depending on levels of conflict, refugee numbers have fluctuated at between 0.1 and 0.3 percent of the world population. The analysis also showed that apparent increases in the numbers of globally displaced are driven by the inclusion of populations and countries that were previously excluded from the dataset. Besides a strong increase in the number of countries included in the UNHCR database over the last half‐century (from 76 in 1970 to 214 in 2018), recent figures have been artificially inflated through the inclusion of previously unregistered or underreported IDPs as well as people in “refugee‐like” situations in refugee statistics. This shows the dangers of using refugees and, more generally, migration data uncritically. Our analysis also highlights the bias and politics involved in choices around defining, collecting, and presenting displacement statistics. Finally, our analyses are based on stock data, which allowed us to study long‐term trends, but which is less suitable to study short‐term trends, circular flows, or the average duration of displacement (see, e.g., Devictor and Do ²⁰¹⁷).

While refugee populations continue to be concentrated in countries with low‐to‐medium income levels, the analysis reveals several geographic shifts in refugee migration. Such shifts primarily reflect the idiosyncratic occurrence of conflicts rather than being part of some broad, inevitable trend toward intensified conflict—in fact, the long‐term trend has been toward reduced levels of violent conflict. While levels of refugee migration have remained stable, refugees have tended to come from a shrinking number of origin countries and to move to an increasing variety of destination countries over recent decades. This trend seems to reflect a concentration of recurrent conflict cycles in a relatively small number of countries and a parallel increase in the number of safe destination countries. Although the vast majority of refugees continue to stay near their origin countries, the average distance between origin and destination countries has increased over time, presumably linked to the greater ease of travel and the migration‐facilitating function of networks.

Yet, overall, these findings challenge the assumption that we are experiencing a global “refugee crisis.” From a historical perspective, the claim of “unprecedented levels” of forced migration seems even more unjustified, especially when we compare current levels to earlier periods. Rather than a general increase, we observe fluctuations in refugee migration over time, as well as geographic shifts in the spread and distance of refugee migration. The fact that refugee records only started after the mid‐century peak and that, until the 1990s, the data were only partial, means that recent numbers may seem inflated if they are not put into a historical context. This suggests that recent surges in refugee numbers as well as asylum applications in Western countries do not reflect a structural change in the trends and patterns of global refugee migration but, rather, reflect a “normal” and therefore temporary response to recent increases in conflict levels in particular countries, with refugee numbers usually going down again after the conflicts subside.