Emigration rates in the EBRD regions have been rising since 1990. Single, male, young and highly educated individuals from urban areas are more likely to migrate than their peers. Low satisfaction with the quality of local amenities (such as air and water, education, healthcare, housing, and roads and transport) is strongly associated with intentions to migrate. Many countries in these regions also serve as transit and/or host countries for refugees and irregular migrants from other parts of the world, who tend to be single, male, young and reasonably well educated (relative to the average level in their country of origin). Well educated refugees and irregular migrants often select their intended country of destination with a view to maximising the return on their skills, while those with lower levels of education are more likely to head for countries that have better integration policies and faster asylum processes.
The total number of international migrants around the world – a figure that includes both voluntary migration and forced displacement – has continued to grow rapidly. It reached 258 million in 2017, up from 153 million in 1990, according to figures compiled by the United Nations Department of Economic and Social Affairs (UN DESA).1 People residing in a country where they were not born, or of which they do not hold citizenship, totalled 3.4 per cent of the world’s population in 2017, up from 2.9 per cent in 1990. Most of those people are economic migrants – people who have travelled abroad voluntarily in search of economic opportunities. Such migration can be short-term, long-term or circular (see Box 3.1 for a discussion of circular migration).
The total number of refugees – people who have been forcibly displaced as a result of natural disasters, political persecution or war – has also continued to rise, standing at 20 million in 2017, up from 17 million in 1990, according to the Office of the United Nations High Commissioner for Refugees (UNHCR). However, the number of refugees has fallen as a percentage of the world’s population over that period, declining from 0.33 per cent to 0.26 per cent.
While almost two-thirds of all economic migrants reside in high-income countries, developing countries host nearly 80 per cent of the world’s refugees and asylum seekers.2 To be granted refugee status, an asylum seeker must face a well-founded fear of persecution on the grounds of race, religion, nationality or membership of a particular social group.
Emigration rates in the EBRD regions have consistently been higher than the global average. Indeed, in 2017, 9.7 per cent of all people who were born in the EBRD regions or held citizenship of a country in these regions lived outside their country of birth or citizenship, up from 8.1 per cent in 1990.
This chapter looks at the factors that are contributing to the large numbers of migrants leaving economies in the EBRD regions. The first section of the chapter provides a snapshot of current migration trends. The second section then looks at the profile of a potential economic migrant – an individual who is considering moving abroad – before investigating the factors that shape migration decisions and examining the ways in which these factors differ across specific groups of individuals. The last section looks at recent flows of refugees from conflict areas and examines the socio-demographic characteristics of migrants who have made their way to Europe.
Migration trends in the EBRD regions
What do we know about the geography of migration in the EBRD regions? Russia and Ukraine are both in the top 10 economies of origin worldwide in terms of absolute numbers of migrants (see Chart 3.1), with Russia ranked third, after India and Mexico. The West Bank and Gaza have the highest rate of emigration in the EBRD regions (and the 10th highest in the world) as a percentage of total population, with their 3.8 million emigrants equating to 82 per cent of the remaining local population.
The main destinations for migrants leaving economies in the EBRD regions are Russia, Germany, Ukraine, the United States of America (USA) and Kazakhstan (see Chart 3.2). Indeed, almost 98 per cent of migrants living in Russia come from within the EBRD regions, with more than 6 million coming from eastern Europe and the Caucasus (EEC) and almost 5 million coming from Central Asia. Conversely, more than 4 million people who were born in Russia or hold Russian citizenship are now living in the EEC region, while almost 4 million live in Central Asia.
Around 3 million people originating from the southern and eastern Mediterranean (SEMED) currently live in a different economy in the SEMED region. Another 9 million are living outside the EBRD regions, mainly in Europe and Gulf Cooperation Council (GCC) countries. European countries are also the main destination for the roughly 8.9, 10.4 and 3.4 million migrants originating from central Europe and the Baltic states (CEB), south-eastern Europe (SEE) and Turkey, respectively, playing host to 7.3 million migrants from CEB countries, 8 million from SEE countries and 2.8 million from Turkey.
These figures represent stocks of migrants, so they reflect past migration flows, but they can also help us to understand future migration patterns. Indeed, existing networks of migrants abroad may be an important pull factor for people who are considering migrating.3 The presence of diasporas abroad and the extent to which migrants are concentrated in certain geographical areas may also have an impact on how much money migrants send home to their country of origin (see Box 3.2).
Russia is ranked fourth worldwide in terms of absolute numbers of immigrants, after the USA, Saudi Arabia and Germany (see Chart 3.3), with more than 11 million foreign-born individuals or foreign citizens living in the country. Jordan and Lebanon have the highest numbers of immigrants in the EBRD regions as a percentage of their local populations.
- Number of emigrants living abroad (left-hand axis)
- Emigrants living abroad as a percentage of the remaining local population (right-hand axis)
Source: UN DESA (2017a) and authors’ calculations.
Note: Data relate to 2017. “Emigrants” are defined here as people who are no longer living in their economy of birth or citizenship. The numbers in parentheses indicate the position of each economy in the global rankings in terms of the absolute number of emigrants and emigrants as a percentage of the remaining local population respectively. The dark green bars and diamonds indicate the top 10 economies in absolute and percentage terms respectively.
- Number of migrants from EBRD regions living in country (left-hand axis)
- Migrants from EBRD regions as a percentage of the local population in the country of destination (right-hand axis)
Source: UN DESA (2017a) and authors’ calculations.
Note: Data relate to 2017. “Migrants” are defined here as people who are no longer living in their economy of birth or citizenship. The dark green bars and diamonds indicate the top 10 economies in absolute and percentage terms respectively.
- Number of immigrants (left-hand axis)
- Immigrants as a percentage of the local population (right-hand axis)
Source: UN DESA (2017a) and authors’ calculations.
Note: Data relate to 2017. “Immigrants” are defined here as people who are no longer living in their economy of birth or citizenship. The numbers in parentheses indicate the position of each economy in the global rankings in terms of the absolute number of immigrants and immigrants as a percentage of the local population respectively. The dark green bars and diamonds indicate the top 10 economies in absolute and percentage terms respectively.
Potential migrants: in search of more than just wages?
As regards the factors shaping migration decisions, previous studies have highlighted the roles played by (i) income differentials between countries of origin and destination, (ii) the corresponding differences in unemployment levels between countries of origin and destination, (iii) the cost of migration and (iv) the presence of migrant networks abroad.4 Less attention, however, has been devoted to understanding how the quality of life and the quality of amenities in the home country may be linked to decisions to move abroad. This section looks at the profile of a typical migrant and compares monetary and non-monetary incentives to leave one’s home country.
In general, individuals with higher levels of subjective well-being tend to report having less desire to emigrate.5 Indeed, immigration rates in Russia are positively correlated with expected income differentials, but they are also lower where expectations regarding improvements in the quality of life at home are higher.6 Similarly, a person’s propensity to migrate to another area (including other areas of the same country) has been shown to be lower where satisfaction with the amenities in the current place of residence is higher.7
At the same time, however, other research suggests that quality of life – as captured by demographic and environmental indicators – had no impact on migration to high-income countries in the period 1991-2000.8 Thus, evidence on the links between local amenities and international migration decisions is far from conclusive. The analysis presented in this section seeks to provide fresh insight into the impact that quality of life has on plans to move abroad.
Data and methodology
The analysis in this section is based on data from the Gallup World Poll – a comprehensive annual survey completed by people living in more than 160 economies around the world (including all economies in the EBRD regions). The survey collects information on people’s demographic and socio-economic characteristics, as well as their attitudes. Crucially, between 2010 and 2015 respondents were also asked about their satisfaction with a range of amenities – air and water, education, healthcare, housing, and roads and transport. They were also asked whether they would ideally like to move permanently to another country if they had the opportunity and whether they intended to migrate in the next 12 months. Other questions concerned people’s preferred countries of destination and any social networks they had at home and abroad. Given that the survey mainly covers non-conflict areas, respondents can predominantly be thought of as potential economic migrants.9
Stated intentions to migrate have been found to be strongly associated with actual migration.10 At the same time, intentions to migrate are also interesting in their own right as a reflection of people’s desire to leave the country, irrespective of whether they do actually leave. The following analysis of people’s intentions to migrate accounts for the cost of moving from the country of origin to the stated destination country on the basis of the geographical distance and language databases prepared by the Centre d’Études Prospectives et d’Informations Internationales (CEPII). Annex 3.1 describes the relevant regression analysis in greater detail.
Intentions to migrate on the rise
Intentions to migrate rose around the world between 2010 and 2015 (see Chart 3.4). In the EBRD regions, more than 3 per cent of respondents indicated an intention to migrate in 2015, compared with 1.8 per cent in 2010. A similar trend was observed in countries with comparable levels of income, with 3.9 per cent of respondents in those economies reporting an intention to migrate in 2015, up from 2.5 per cent in 2010. In absolute terms, those figures suggest that around 20 million people in the EBRD regions may have been considering a move abroad in 2015.
The five countries with the largest percentages of people reporting an intention to move abroad are all in sub-Saharan Africa, with Liberia topping the list at 14 per cent. Albania is ranked sixth in the world, with almost 10 per cent of respondents in that country indicating an intention to migrate within a year. Meanwhile, rates in excess of 5 per cent can be observed in Armenia, Kosovo and FYR Macedonia (see Chart 3.5). At the same time, these economies already have large numbers of emigrants living abroad (see Chart 3.1). Between 2010 and 2015 the strongest increases in intentions to migrate were observed in the EEC region, the SEE region and Turkey, while Central Asia and Russia saw small declines in the percentage of respondents intending to leave their respective countries.
The profile of potential migrants
Across the world, young men from urban areas are more likely to report an intention to migrate (see columns 1-3 of Table 3.1). These results also apply to potential migrants from the EBRD regions and comparator economies (see columns 4 and 5 of Table 3.1), and are confirmed when a similar analysis is undertaken on a subset of economies in central and south-eastern Europe using a different dataset (see Box 3.3). Married individuals are less likely to want to migrate, while having children has a positive effect on intentions to migrate but is only marginally statistically significant. People who have completed secondary education are more likely to migrate than their less-educated counterparts.
People in full-time employment and with a tertiary degree are around 1 percentage point less likely to report an intention to migrate. The effect of having a full-time job is stronger in the EBRD regions than in countries with comparable income levels (see columns 4 and 5 of Table 3.1).
At the same time, highly educated individuals in the EBRD regions are more likely to seek to emigrate than people who have only completed secondary education. This is true across all of those regions, with the exception of the CEB region. This may point to large skills mismatches in many EBRD economies, negatively affecting returns to skills (see also Box 3.4 for a discussion of the issue of “brain drain”).
Unsurprisingly, intentions to migrate are also more likely to be reported where the wage differential between the country of destination and the country of origin is larger. They are also more likely where respondents have social networks abroad (that is to say, friends or relatives who are already living abroad who can be counted on in times of need). Conversely, the likelihood of migration is lower where individuals have strong social networks at home and where the distance between the country of destination and the country of origin is greater.
People who have greater confidence in their own personal safety and greater trust in the national government are less likely to report an intention to migrate, while perceptions of corruption in business are associated with an increase in the likelihood of intentions to migrate. When respondents’ views on corruption and personal safety, social networks and other likely determinants of life satisfaction are all controlled for, life satisfaction itself does not have a statistically significant impact on intentions to migrate.
People who intend to migrate report lower levels of satisfaction with amenities
On average, satisfaction with the quality of air and water, education, healthcare, housing, and roads and transport remained broadly stable in the EBRD regions over the period 2010-15, standing at around 0.54 on a scale of 0 to 1 in 2015 (see Chart 3.6). However, that average masks substantial variation across individual regions. In the SEMED region, for example, average satisfaction with amenities declined by 20 per cent between 2010 and 2015, while satisfaction levels in Turkey fell by 5 per cent. In Central Asia and the EEC region, by contrast, average satisfaction with amenities improved by around 9 per cent over that period.
Across regions, people who intend to migrate tend to report lower levels of satisfaction with local amenities than people who intend to stay. Those differences are statistically significant at the 1 per cent level and continue to be observed after controlling for individual characteristics such as age, education and economy of residence (see Chart 3.7).
Results at the individual level confirm that lower levels of satisfaction are also associated with intentions to migrate (see Table 3.2). Indeed, being satisfied with the quality of local amenities reduces the likelihood of seeking to migrate by nearly 1.6 percentage points (see column 3). This is a sizeable effect, considering that the average likelihood of intending to move abroad is 3.5 per cent in the sample. In order to obtain a similar reduction in the likelihood of seeking to migrate, wages would need to rise in a way that reduced the wage differential between the countries of origin and destination by approximately 70 per cent. In a country such as Albania, this would correspond to a wage increase of roughly US$ 477 per month.
At the level of individual amenities, there is a particularly strong correlation between lower levels of satisfaction with air and water, education and housing and an increased likelihood of seeking to migrate. The link between satisfaction with healthcare and intentions to migrate is also statistically and economically significant. While the coefficient for satisfaction with roads and transport is not statistically significant, the effect appears to be driven by economies in the sample that are not part of the EBRD regions.
Satisfaction with individual amenities and intentions to migrate are negatively correlated in the EBRD regions (see Chart 3.8). Moreover, the link between satisfaction with roads and transport and decisions to migrate is stronger in the EBRD regions than in comparator economies, and the difference is statistically significant.
Regression analysis (unreported) involving interaction between levels of satisfaction with amenities and a variable capturing the respondent’s highest qualification shows that people who have been educated to tertiary level attribute greater importance to education, healthcare, and roads and transport than peers with no education or only primary education when thinking about emigrating. Similar analysis based on interaction between satisfaction with amenities and variables for different age groups indicates that satisfaction with education is more important to people aged between 18 and 24 than it is to people aged between 25 and 64.
Source: Gallup World Poll and authors’ calculations.
Note: This chart reports the percentage of respondents who indicated that they wanted to move to another country on a permanent basis and were planning to do so within the next 12 months. The comparator economies used for this purpose are those with GDP per capita (at PPP) between the minimum and maximum values observed in the EBRD regions.
Source: Gallup World Poll and authors’ calculations.
Note: This map reports the percentage of respondents who indicated that they wanted to move to another country on a permanent basis and were planning to do so within the next 12 months. Data represent six-year survey-weighted averages. The last four intervals in the legend are half-open: they include the upper endpoints but not the lower endpoints.
|Dependent variable||Intention to migrate (0/100)|
|Global sample||Economies in the EBRD regions||Comparator economies|
|Married or in partnership||-1.55***||-1.60***||-1.64***||-1.88***||-1.30***|
|Has children below age of 15||0.10*||-0.02||0.19**||0.44**||0.03|
|Living in urban area||1.06***||1.20***||1.15***||1.02***||1.06***|
|Education and labour market outcomes||No education or only primary education||-1.15***||-0.51***||-0.47***||0.08||-0.54***|
|Full-time employment and tertiary education||-0.75***||-0.94***||-0.95***||-0.81*||-0.72|
|Wage differential and cost of migration from country of origin to country of destination||Wage differential (log)||1.26***||1.28***||0.76***||1.05***|
|Distance between most populous cities (log)||-1.46***||-1.40***||-0.23||0.07|
|Once part of same country||-1.39||-1.64||-2.90||-5.90**|
|Once in colonial relationship||1.74**||1.42||15.94***||-0.37|
|Respondent has network abroad||3.71***||4.07***||4.15***||3.81***|
|Respondent has network at home||-0.86***||-0.84***||-0.74***||-0.76***|
|Attitudes and perceptions||Personal safety||-0.84***||-0.60***||-0.72***|
|Trust in the national government||-1.17***||-1.20***||-1.10***|
|Corruption widespread in business||0.34***||0.77***||0.12|
|Corruption widespread in government||0.15||-0.00||-0.09|
|Number of observations||774,175||469,670||325,004||77,375||166,251|
SOURCE: Gallup World Poll, CEPII database and authors’ calculations.
NOTE: Estimated using a linear probability model with survey-weighted observations. Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. Columns 1-3 report results for all economies included in the Gallup World Poll, while columns 4 and 5 show results for economies in the EBRD regions and for economies with comparable income levels only. All specifications take account of demographic characteristics, education and labour market outcomes, country of origin fixed effects and survey year fixed effects. Specifications in columns 2-5 also control for the wage differential and the cost of migration from the country of origin to the country of destination, whereas specifications in columns 3-5 include variables that capture the respondent’s attitudes and perceptions. The results are robust to the inclusion of country of destination fixed effects. The comparator economies used for this purpose are those with GDP per capita (at PPP) between the minimum and maximum values observed in the EBRD regions.
Source: Gallup World Poll and authors’ calculations.
Note: The comparator economies used for this purpose are those with GDP per capita (at PPP) between the minimum and maximum values observed in the EBRD regions.
Source: Gallup World Poll and authors’ calculations.
Note: Estimates calculated on the basis of regression analysis controlling for demographic characteristics, education and labour market outcomes, country of origin fixed effects and survey year fixed effects, using survey-weighted observations. The comparator economies used for this purpose are those with GDP per capita (at PPP) between the minimum and maximum values observed in the EBRD regions.
Source: Gallup World Poll, CEPII database and authors’ calculations.
Note: These estimates of the impact that changes in satisfaction with amenities have on intentions to migrate have been obtained using a linear probability model. The specification controls for country of origin fixed effects and survey year fixed effects, demographic characteristics of the respondent, education and labour market outcomes, the wage differential and the cost of migration from the country of origin to the country of destination, as well as variables capturing the respondent’s attitudes and perceptions. The regression also includes interaction terms between each of the amenity variables and a dummy variable identifying economies in the EBRD regions. Robust standard errors have been used to construct the 95 per cent confidence intervals shown.
Refugees and irregular migrants from other parts of the world
At the end of 2017, there were 68.5 million forcibly displaced persons worldwide. The total number of people seeking sanctuary in a foreign country as a refugee stood at 20 million, with more than half of all refugees coming from Afghanistan, South Sudan and Syria.11 The present refugee crisis differs from that of the 1990s in three crucial respects. First, it has had a much stronger impact on politics in Europe on the back of the rise of populist parties and the increase in economic hardship that was brought about by the financial crisis of 2008-09 and the subsequent eurozone debt crisis. Second, the current crisis involves multiple actors, over which Western nations have much less influence than they did in the 1990s. Third, the refugees arriving in Europe are regarded by many as being more culturally distant than those of the previous wave.12
This makes it all the more important that we understand the socio-demographic characteristics of such refugees and irregular migrants, many of whom are likely to apply for asylum in their destination countries. Indeed, achieving a better understanding of those people’s profiles will help with the development of more effective integration policies in transit and host countries, contributing to improved social cohesion and better economic outcomes. It may also help their economies of origin to develop policies that address the skills gaps left behind by refugees. This section examines the socio-demographic characteristics of refugees and irregular migrants, breaking them down on the basis of their economies of origin and destination. Analysis in this section is based not only on Gallup World Poll data, but also on the Flow Monitoring Survey (FMS) conducted by the International Organization for Migration (IOM), which is carried out in Europe as part of the IOM’s Displacement Tracking Matrix and surveys refugees and irregular migrants aged 14 and over (see Annex 3.2 for details).
Many economies in the EBRD regions act as transit points for refugees and irregular migrants
Many economies in the EBRD regions host refugees and irregular migrants in transit. In particular, from 2015 the Western Balkans migration route – formally “closed”, as announced by the EU in March 2016 – saw sharp increases in the numbers of people on the move. This was a key route for migrants who entered the European Union via Bulgaria and Greece – travelling via Turkey by both land and sea – with the aim of reaching various countries in the Schengen area (see Chart 3.9).
Chart 3.10 shows the countries where the IOM has interviewed refugees and irregular migrants, indicating that Italy is Europe’s main transit country, followed by Greece, FYR Macedonia, Croatia and Bulgaria. Other countries in the EBRD regions also host large numbers of refugees, such as Turkey (3.5 million according to the UNHCR), Lebanon (1 million), Jordan (700,000) and Egypt (300,000). In addition, the Western Balkans route was also used by migrants from other economies where the EBRD invests, such as Albania, Bosnia and Herzegovina, FYR Macedonia, Kosovo, Montenegro and Serbia. Box 3.5 discusses the impact that such influxes of refugees have had on host countries’ labour markets, using the example of Turkey.
Where do refugees and irregular migrants come from?
Refugees and irregular migrants predominantly originate from low or lower-middle-income countries. Indeed, Syria, Afghanistan and Iraq account for 55 per cent of all respondents interviewed by the IOM as part of its FMS surveys (see Chart 3.11). The overwhelming majority of respondents are male (83 per cent) and single (70 per cent), with an average age of 26. Around half of the respondents report having been employed before migrating, with 17 per cent being educated to tertiary level.
Most respondents have been displaced by war and conflict (77 per cent of the total; see Chart 3.12). This is very much the main reason for migrating among survey respondents originating from Afghanistan, Eritrea, Iraq, Somalia, Sudan and Syria.
At the same time, 17 per cent cite economic factors as the main reason for being on the move, including the vast majority of respondents from Morocco, Algeria and Pakistan. There are no real differences between men and women or between people of different ages when it comes to the reason for leaving their home country.
Nearly two-thirds of respondents cite Germany or Italy as their final destination (see Chart 3.13), followed by France, Sweden and the United Kingdom (see Box 3.6 for an overview of the impact that immigration has had on populist voting in those countries). More than 80 per cent of people heading for Austria, Denmark, Finland, Germany or Norway have left their home country because of conflict. That figure is less than 60 per cent among people heading for Italy, Belgium and France, with more respondents migrating for economic reasons in the case of those destination countries.
Socio-demographic characteristics of refugees and irregular migrants
Regression analysis indicates that people who are educated to secondary or tertiary level are significantly more likely to migrate than people with lower levels of education, particularly when people are fleeing a major conflict (see columns 1 and 2 of Table 3.3). In countries with only a minor conflict or no conflict at all (see column 3), education plays less of a role (and only tertiary education remains a statistically significant predictor of becoming a migrant). Thus, refugees and irregular migrants escaping major conflicts tend to be highly educated relative to the national average in their country of origin, possibly because better-educated individuals will be in a better position to finance their trip, while liquidity constraints and immigration restrictions will prevent the poorest people from migrating.13 Full details of this regression analysis can be found in Annex 3.3.
Like respondents to the Gallup survey who report an intention to migrate, refugees and irregular migrants who are interviewed by the IOM are also more likely to be single, male and young. Being in employment increases the likelihood of migration where a person lives in an area with a major conflict (perhaps because only people with sufficient financial resources can afford to escape a conflict zone), but otherwise it has no impact. There is no real difference between the findings for men and women.
Since the FMS surveys did not ask respondents about their income (which may be an important determinant of people’s willingness and ability to leave their home country), this analysis uses Gallup data to estimate earnings for those respondents on the basis of their characteristics and the earnings profiles in their respective countries of origin, as discussed in Annex 3.3.
Higher levels of estimated pre-migration income strongly increase the probability of emigration – both in countries with major conflicts and in other countries, and for both men and women (see Table 3.4). However, the impact is stronger in countries with a major conflict, and it is also stronger for men. A similar pattern is observed if we estimate household income, rather than personal income. Thus, better-educated people with jobs and higher levels of income are more likely to be able to leave countries affected by severe conflicts.
For many refugees and irregular migrants, education and experience are their only assets. It is perhaps not surprising, then, that people try to maximise their returns on those assets. As Table 3.5 shows, highly educated refugees and irregular migrants are more likely to target countries where returns to education tend to be higher. In this regression analysis, the sample is restricted to respondents covered by FMS surveys and the dependent variable is the potential return to education in the intended country of destination. Potential returns to education have been estimated on the basis of wages in the relevant destination country for individuals with the same level of education and the same socio-demographic characteristics as the refugees and irregular migrants in question (see Annex 3.3 for details). Returns to primary education in intended destination countries tend to be relatively low, while returns to higher levels of education vary. These findings do not imply causality, as other factors may affect the choice of destination and be linked to a person’s education. Nonetheless, these results indicate that education may have an important role to play when refugees and irregular migrants choose their destination.
As the regression analysis in Table 3.6 shows, a number of other factors also appear to influence refugees and irregular migrants’ choice of destination. As with the previous analysis, the sample is again restricted to respondents covered by FMS surveys, while outcome variables represent various characteristics of the intended destination country of each respondent.
Refugees and irregular migrants who are educated to primary level (or less) and secondary level are more likely to head for countries that have lower unemployment rates and more comprehensive migrant integration policies. The nature of integration policies is captured by the Migrant Integration Policy Index (MIPEX), which ranges from 0 to 100 and is based on 167 policy indicators covering the following eight policy areas: labour market mobility, reunification of families, education, political participation, long-term residence, access to nationality, measures tackling discrimination and health.
Refugees and irregular migrants who are educated to primary (or less) and secondary level are also more likely to choose destination countries where asylum applications are considered faster and where work permit applications, once asylum has been granted, take less time to process. More highly developed social safety nets also make a destination country more attractive for migrants with primary and secondary education. In other words, refugees and irregular migrants coming to Europe respond to incentives at all stages of the migration process.14
|Dependent variable||Intention to migrate (0/100)|
|Satisfied with:||Air and water||-0.53***||-0.47***||-0.36***|
|Roads and transport||-0.29***||-0.27***||-0.09|
|Wage differential (log)||1.24***||1.28***||1.33***||1.33***|
|Number of observations||755,121||462,339||324,733||594,064||378,739||281,227|
|Wage differential and cost of migration from country of origin to country of destination||No||Yes||Yes||No||Yes||Yes|
|Attitudes and perceptions||No||No||Yes||No||No||Yes|
SOURCE: Gallup World Poll, CEPII database and authors’ calculations
NOTE: Estimated on the basis of a global sample using a linear probability model with survey-weighted observations. Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. All specifications take account of demographic characteristics, education and labour market outcomes, country of origin fixed effects and survey year fixed effects. Specifications in columns 2, 3, 5 and 6 also control for the wage differential and the cost of migration from the country of origin to the country of destination, whereas specifications in columns 3 and 6 include variables that capture the respondent’s attitudes and perceptions. The results are robust to the inclusion of country of destination fixed effects.
Source: IOM and authors’ calculations.
|All||Major conflict||Minor/no conflict|
|Country fixed effects||Yes||Yes||Yes|
|Number of observations||44,272||29,794||14,478|
SOURCE: FMS (2015-16) and Gallup World Poll (2009-14).
NOTE: Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. Data relate to people from the following nine countries: Afghanistan, Algeria, Iran, Iraq, Libya, Morocco, Nigeria, Pakistan and Syria. The dependent variable is equal to 1 for refugees and irregular migrants in the FMS data and 0 for participants in Gallup World Polls. Reference categories are as follows: education below secondary level, unemployed or out of labour force, female, aged 54+, and single. All specifications include dummies for widowed and divorced, although these are not reported above. A “major conflict” is defined as a country with 1,000 or more conflict-related deaths in any of the years in question.
|All||Major conflict||Minor/no conflict|
|Men and women|
|Log of estimated income||0.113***||0.137***||0.069***|
|Number of observations||44,272||29,794||14,478|
|Log of estimated income||0.150***||0.172***||0.106***|
|Number of observations||23,665||16,448||7,217|
|Log of estimated income||0.048***||0.058***||0.032***|
|Number of observations||20,607||13,346||7,261|
SOURCE: FMS (2015-16) and Gallup World Poll (2009-14).
NOTE: Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. Data relate to people from the following nine countries: Afghanistan, Algeria, Iran, Iraq, Libya, Morocco, Nigeria, Pakistan and Syria. The dependent variable is equal to 1 for refugees and irregular migrants in the FMS data and 0 for participants in Gallup World Polls. All specifications take account of country fixed effects. A “major conflict” is defined as a country with 1,000 or more conflict-related deaths in any of the years in question.
|All||Major conflict||Minor/no conflict|
|Number of observations||3,429||2,478||951|
SOURCE: FMS (2015-16) and Gallup World Poll (2009-14).
NOTE: Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. The dependent variable is the return to education in the intended country of destination (as calculated on the basis of Gallup World Poll data). Reference categories are as follows: less than secondary education, unemployed or out of labour force, female, age 54+, and single. A “major conflict” is defined as a country with 1,000 or more conflict-related deaths in any of the years in question.
|Outcome variable||Log of unemployment rate in destination country||MIPEX (0-100)||Average duration of asylum procedure (0-1)||Waiting time before accessing labour market (0-1)||Social expenditure (as a percentage of GDP)|
|Number of observations||3,492||3,509||3,484||3,509||3,423|
SOURCE: Eurofound, FMS (2015-16), MIPEX, OECD, World Bank and authors’ calculations.
NOTE: Robust standard errors are reported in parentheses, and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively. At least 100 nationals have been surveyed for each country included in this analysis. All specifications include survey country fixed effects. Reference categories are as follows: more than secondary education, unemployed or out of labour force, female, age 54+, and single. All specifications include widowed and divorced dummies, though not reported above.
International migration is on the rise, both globally and in the EBRD regions. Emigration rates in the EBRD regions have risen since 1990, with almost 10 per cent of people born there now living outside their country of birth or citizenship. Many economies in the EBRD regions have also become major destinations for migrants or important transit countries on migration routes.
The emigration of workers is a concern for many countries, particularly in central and eastern Europe. People who express an intention to migrate tend to be single, male and young and live in cities. They also tend to be better educated than the average person. Many of those people also believe that corruption is widespread in the local business sector and they are often dissatisfied with local amenities such as healthcare, education, air and water, housing, and roads and transport. Satisfaction with the local education system is particularly important for the young and the highly educated when it comes to deciding whether to move abroad.
These findings suggest that improving the business environment and the quality of public services may significantly reduce people’s desire to emigrate. Indeed, increasing satisfaction with local amenities can have the same impact in terms of lowering intentions to migrate as closing 70 per cent of the wage gap between the country of origin and the country of destination. At the same time, the impact of wage increases in the economy of origin is not straightforward: at lower levels of economic development, rising wages can actually increase emigration, as low-skilled workers find it easier to afford the cost of migration. Moreover, the findings presented in Chapter 2 suggest that emigration by skilled workers results in a decline in total factor productivity for firms in countries of origin, but emigrants do send back knowledge, thereby helping those countries to boost innovation and move towards the technological frontier.
In order to minimise the costs of emigration and maximise the associated gains, governments can work with firms to establish training programmes so as to foster skills that are widely sought after in their domestic labour markets. Policies aimed at attracting highly qualified migrants can also help to address specific labour market shortages in the short term. The EU’s Blue Card initiative is a good example of this kind of approach.15
Many economies in the EBRD regions also serve as transit and/or host countries for refugees and irregular migrants from other parts of the world, many of whom are likely to seek asylum. Like economic migrants from the EBRD regions, these people tend to be single, male, young and reasonably well educated (relative to the average level in their country of origin). It appears that they often select their intended country of destination with a view to maximising the return on their skills – typically the only assets they possess. Refugees and irregular migrants who are educated to primary and secondary level are more likely to head for countries with lower unemployment rates, better migrant integration policies, faster asylum processes, easier access to the labour market for people who have successfully claimed asylum, and stronger social safety nets.
While the vast majority of refugees and irregular migrants leave their country in order to escape conflict, the main motivation of a significant number of migrants from countries such as Algeria, Egypt, Morocco and Pakistan is a desire to seek out better economic opportunities abroad. While many of these migrants may ultimately be denied asylum, they can slow down asylum application procedures. This may, in turn, undermine popular support for a well-managed and fair asylum system.16 Ageing European economies – including some countries where the EBRD invests – could consider tackling this problem by increasing legal employment opportunities for African citizens on a selective basis, depending on local needs. Such initiatives could form part of a broader strategy aimed at containing irregular migration to Europe.17
Moreover, policies that support the integration of refugees and irregular migrants into the labour market need to be tailored to those people’s skills.18 Refugees escaping major conflicts (such as the fighting in Syria) may well benefit from receiving early access to language courses and other basic training while waiting for decisions on their asylum applications. Prompt access to employment will also help refugees to integrate better into society.19
BOX 3.1. Circular migration: key to economic development?
Multinational firms spend millions of dollars seconding their workers to foreign subsidiaries and branches for training and work experience. Evidence suggests that such largesse pays off, since hiring managers and workers with foreign experience increases productivity, profitability and innovation.20 These beneficial effects of international migration do not just apply to firms: migration is also an important driver of the spread of knowledge across economies.
Average productivity levels differ very considerably from country to country, and a large percentage of those differentials are arguably due to differences in the knowledge deployed in production. Such knowledge comes in different forms. It may, for example, be embedded in goods and referred to as “technology”. Somebody who possesses a calculator, for instance, does not need to understand the mechanics of addition or multiplication. Knowledge may also be codified – for example, if it is written down in the form of patents. Other forms of knowledge are confined to people’s brains. For instance, a worker’s productivity largely reflects his/her accumulated experience and ability to learn on the job. Knowledge that is embedded in goods or codified in patents can be easily traded, both within and across borders, but the knowledge that is held in the brain of a surgeon, a pilot or a senior manager can only be shared through personal interaction over the course of many years.
This is why migration plays such a crucial role in the transfer of knowledge: foreign workers and returning migrants act as a “revolving door” for the kinds of tacit knowledge that cannot be traded even when goods and capital move freely. Thus, migrants who move abroad – taking their acquired knowledge and know-how with them – and then subsequently return home again boost productivity in both economies.
A recent study documenting the benefits of circular migration looked at the experiences of refugees who fled the former Yugoslavia during the 1990s.21 In the early 1990s, more than half a million citizens of the former Yugoslavia travelled to Germany in order to escape the war. Most of them were integrated into the German labour force, with almost no restrictions on employment. Following the signing of the Dayton Peace Accords in 1995, most of those refugees were repatriated to their home countries. Data suggest that this led to a significant increase in exports from those former Yugoslav economies. Strikingly, the highest rates of export growth were seen in the sectors where those former refugees had tended to work while living in Germany (see Chart 3.1.1). This effect was particularly strong for professionals and people in management roles, where the transfer of knowledge has the potential to be most valuable.
Source: Bahar et al. (2018).
Note: The first quartile contains the 25 per cent of industries with the lowest numbers of returning migrants in 2000, the second quartile contains the next 25 per cent, and so on.
BOX 3.2. Concentration of migrants and remittance flows from Italy
According to Eurostat, there were more than 5 million foreigners living in Italy in 2017, up from 1.5 million in 2003. If we look at the economies where those migrants come from, we can see that 6 of the top 10 are in the EBRD regions: Romania, Albania, Morocco, Ukraine, Moldova and Egypt (in declining order of importance). Italy is also home to large numbers of Chinese migrants.
Migrants living in Italy tend to cluster together in specific regions in the richer, northern parts of the country (see Chart 3.2.1).22 Data point to migrants from the SEE region23 being relatively strongly concentrated in Piacenza in the Emilia-Romagna region and Asti in the Piedmont region. Meanwhile, migrants from the SEMED region are concentrated in specific provinces of the Emilia-Romagna and Lombardy regions, in Sicily’s Ragusa province, and in Aosta. Strikingly, there are very large numbers of Chinese immigrants in the province of Prato, where their density is 17 times higher than the national average. There are also large numbers of Serbs in Vicenza and Trieste, a significant cluster of migrants from Bosnia and Herzegovina in the province of Gorizia, and a large Tunisian population in Ragusa.
The presence of large numbers of migrants in particular geographical areas raises an interesting question: do migrants in such areas behave differently when it comes to sending money home, and does this vary by region and country of origin? To answer this question, this box uses bilateral panel data on remittances and migrant stocks, broken down by province and country of origin.
Those data, which come from the Bank of Italy and Italy’s National Institute of Statistics, cover the period 2005-16. In line with the approach adopted by Santos Silva and Tenreyro (2006), remittances (expressed in euros per migrant) are modelled as a function of the quotient value (see note accompanying Chart 3.2.1) in a regression framework using the Poisson pseudo-maximum likelihood. This analysis controls for province and year fixed effects, as well as province-level economic conditions that are likely to affect labour market outcomes, including income per capita, unemployment rates, the shares of agriculture, manufacturing and services in value added, the age structure of residents and the percentage of migrants who are women.
Larger numbers of migrants from a particular country of origin in a given province are associated with lower levels of remittances (see Chart 3.2.2). This pattern holds across various countries of origin, with the smallest effect being observed for migrants from Morocco and the largest being observed for migrants from Egypt. This may suggest that migrants who live in close proximity to large numbers of their compatriots establish stronger local financial links, resulting in weaker financial links with networks in their countries of origin.
Source: ISTAT (2018) and author’s calculations.
Note: The maps above show residential concentration quotients for 2016. The index is equal to 1 if the distribution of the migrant group across provinces is identical to the distribution across the general population, less than 1 if the migrant group is under-represented in a province, and greater than 1 if the group is over-represented.
Source: Bank of Italy (2018), ISTAT (2018) and author’s calculations.
Note: The chart shows the estimated coefficients from a regression framework using the Poisson pseudo-maximum likelihood. The error bars indicate the 95 per cent confidence intervals.
BOX 3.3. Migration intentions in central and south-eastern Europe: a socio-demographic profile
In 2017, the Austrian central bank, the Oesterreichische Nationalbank (OeNB) conducted another round of its regular Euro Survey, asking questions of 1,000 people aged 15 and over in 10 countries in central and south-eastern Europe (Albania, Bosnia and Herzegovina, Bulgaria, Croatia, the Czech Republic, FYR Macedonia, Hungary, Poland, Romania and Serbia). Just over 8 per cent of respondents of working age (25 to 64 years old) reported intentions to move abroad within one year of the survey. As in the Gallup survey, migration intentions were higher among younger individuals and among men. If those intentions are representative of the entire population and all those who intend to migrate do so, the demographic profile of the region will change considerably (see Chart 3.3.1). In particular, the median age of the labour force will increase further, as will the old-age dependency ratio.
Further econometric analysis of those intentions to migrate reveals that migration intentions vary considerably across regions within individual countries.24 People who live in regions with high average incomes, low levels of unemployment and a dynamic economy tend to be less likely to migrate. Moreover, it tends to be the level of regional income and unemployment, rather than the rate of economic growth, that influences migration intentions.
Source: OeNB Euro Survey (2017) and author’s calculations.
BOX 3.4. Economic development and “brain drain” in transition economies
International migration exhibits two general patterns. First of all, there is an inverted U-shaped relationship between migration and economic development, typically referred to as the “migration transition curve”.25 At income levels below approximately US$ 6,000 per capita, emigration increases as incomes rise (see Chart 3.4.1). If incomes continue to increase, emigration then subsides again. Various explanations for this relationship have been put forward, the most common being the existence of credit constraints preventing potential migrants in poorer economies from realising their aspirations. As financial constraints ease, emigration initially rises. It then falls again as income differentials between the economy of origin and potential destination economies decline.
Second, well-educated people exhibit a much greater propensity to emigrate than their less-educated compatriots, and they tend to cluster in countries or regions where skills are well rewarded.26 Highly skilled people tend to be more responsive to economic opportunities abroad and have more transferable skills. They may also find it easier to comply with the skills-based immigration policies that are in place in many potential destination countries. Skills-based selection into migration decreases with economic development. In low-income economies, university graduates are 20 times more likely to emigrate than their less-educated peers, whereas the ratio of the emigration rate for university graduates to the emigration rate for less-educated individuals is slightly above unity in high-income countries. Eight economies in the EBRD regions have levels of skills-based selection into migration that are higher than one might expect on the basis of their per capita incomes: Tajikistan, Egypt, Jordan, the Kyrgyz Republic, Uzbekistan, Azerbaijan, Latvia and Armenia (see Chart 3.4.2).
This gives rise to concerns that economic development could be accelerating the brain drain and slowing down the accumulation of human capital – which may, in turn, reduce the effectiveness of development policies. In order to examine this question, we can look at the links between per capita income growth and emigration using the migration accounting method proposed by Dao et al. (2018).
It turns out that the inverted U-shaped relationship only holds for low-skilled workers. The emigration rate of university graduates always decreases with economic development. However, the percentage of university graduates in the local population rises with economic development, and more highly skilled workers are more likely to emigrate than low-skilled workers. As a result, a country’s emigrants may become more highly skilled as incomes rise, even though better-educated people become less likely to emigrate.
Regression analysis can also shed light on other factors explaining emigration, such as financial incentives and constraints, the geographical proximity of high-income economies, the linguistic proximity of potential destination countries and the presence of existing migrant networks in destination economies. Increases in skill levels in economies of origin explain around a quarter of the increase in emigration where income per capita is below US$ 1,000 or between US$ 4,000 and US$ 6,000. Macroeconomic drivers – such as income differentials and proximity to existing migrant networks – play a major role in explaining emigration. The effect of financial constraints exceeds that of skill sets in economies where per capita incomes are less than US$ 1,000. However, at higher levels of income, changes in skill sets are more important than financial constraints when it comes to explaining increases in emigration rates.
Although it has a significant impact in the poorest economies, the relaxation of financial constraints leads to only a small rise in emigration in middle-income economies (including those in the EBRD regions). Of greater importance is the increase in education that is associated with rising incomes, which leads to a higher emigration rate overall. Nonetheless, the rate of emigration among the highly skilled continues to fall as countries develop. In that sense, the risk of economic development resulting in a brain drain is smaller than is commonly believed.
Source: OECD and authors’ calculations.
Note: Based on a sample comprising 123 economies which excludes periods of major conflict. The results have been obtained using non-parametric kernel density estimation.27
Source: OECD and author’s calculations.
Note: Based on a sample comprising 123 economies which excludes periods of major conflict. The results have been obtained using non-parametric kernel density estimation.28
BOX 3.5. Large-scale influxes of low-skilled labour and their impact on locals’ use of skills
The Syrian civil war, which has been raging since 2011, has seen more than 5 million refugees flee to neighbouring countries. The majority of those refugees are in the EBRD regions, with more than 3.5 million in Turkey and many others in Jordan and Lebanon. Understanding the impact that these large numbers of refugees have had on local labour markets can help governments to design effective policies aimed at addressing the relevant issues.29
Syrian refugees tend to be low-skilled relative to the population of the host country and are likely to represent a source of relatively cheap labour.30 Owing to a lack of documentation, they work almost exclusively in the informal sector. Consequently, economic research looking at this issue has focused on the impact that refugees have on the employment rates and wages of local workers.31 That research suggests, in line with most previous studies, that Syrian refugees have had a small negative impact on employment rates, particularly for unskilled locals and people working in the informal sector. Those studies do not typically find any large-scale displacement of local people by Syrian refugees. The limited nature of that impact, despite very large numbers of refugees, suggests that adjustment mechanisms beyond employment and wages have played an important role.
One such mechanism involves changes to the combination of production factors that firms use. Refugee labour is likely to complement some production inputs and replace some others. A recent study looked at differences between the numbers of refugees living in different Turkish regions, accounting for the fact that some regions are more attractive to refugees than others.32 It seems that large inflows of refugees have increased the complexity of the tasks performed by local workers. In particular, many locals have reduced the amount of time spent on manual tasks and focused more on abstract, routine and IT-intensive tasks. Turkish middle-school graduates have shifted to routine tasks and high-school graduates have moved to more abstract tasks, while people who are only educated to primary level (that is to say, people at the bottom of the skills distribution) and university graduates (people at the top of the skills distribution) appear to have been unaffected.
The lack of impact on the complexity of less-educated locals’ tasks is in line with the negative impact on their employment rates and wages. Thus, poorly educated Turks have not managed to switch to more complex forms of labour when confronted with large numbers of Syrian refugees. The increased competition they have faced in the labour market has resulted in lower wages.
In response to the sharp increase in the supply of low-skilled labour, firms have also reduced the amount of machinery and equipment they use relative to the amount of labour, making production less capital-intensive. This effect has been stronger in sectors where firms have traditionally relied more on informal labour (even before the arrival of the Syrian refugees).
While firms that have started using more low-cost labour in lieu of capital may outperform other firms in the short run, this rational adjustment may have undesirable long-term implications.33 Firms may start to rely too heavily on informal refugee labour and be left with a suboptimal combination of production inputs when refugees start to return to Syria.
In order to mitigate the long-term impact that the presence of refugees has on labour markets and productivity, vocational training could be given to disadvantaged locals who have lost their jobs. In addition, firms that are located in areas with large numbers of Syrian refugees need stronger incentives to invest in machinery and equipment. Syrian refugees can also be given training, helping them to provide more highly skilled labour.
BOX 3.6. Immigration and populist voting in Europe
Has the recent rise in populism in advanced economies been caused by immigration and the refugee crisis? This box surveys the evidence, looking at recent studies carried out in Austria, France, Greece, Italy, Sweden and the United Kingdom – as well as a Europe-wide study of immigration and emigration. These studies show that the impact that immigration has on political outcomes depends on the composition and intensity of immigration. In particular, some studies identify a non-linear relationship between the intensity of immigration and populist voting, with a limited increase in immigration reducing support for right-wing populists – consistent with the “contact theory” posited by Allport (1954) – and a large increase in immigration having the opposite effect.
Most of these studies use sophisticated empirical strategies to identify the causal effect that the presence of immigrants in specific regions or municipalities has on the voting patterns of local residents. It is essential, in this regard, to go beyond the correlation between populist voting and immigration and analyse the direction of causality.
Indeed, in certain European countries the global financial crisis resulted in populist parties winning elections even before the recent increase in immigration. Those populists then saw an opportunity to use anti-immigrant discourse to strengthen their hold on power and introduced strict anti-refugee policies. Consequently, there are currently very few immigrants (especially asylum seekers) in those specific countries. Some of those governments have also established firm control over public authorities, the media and the judiciary, allowing them to keep winning elections. If the direction of causality were not identified correctly, such countries would suggest a negative correlation between immigration and populist voting.
The most common way of addressing the issue of causality is to assume that new immigrants are most likely to head for areas where previous generations of immigrants settled. Consequently, an increase in immigration (for instance, owing to the recent refugee crisis) will have a disproportionate effect on areas that already have large immigrant populations. Other studies (such as recent research looking at Austria, Denmark and France) are based on central government’s random or quasi-random allocation of refugees to specific municipalities. All studies also control for trends in pre-influx election results.
In Denmark, the assignment of refugees to particular municipalities in the period 1986-98 led to increases in anti-immigrant voting in all but the largest and most urbanised municipalities (the largest 5 per cent).34 In smaller municipalities, each percentage point increase in refugees’ share of the municipality’s population resulted in a 1-2 percentage point increase in voting for anti-immigration parties. In the largest municipalities, however, the opposite was observed: each percentage point increase in refugees’ share of the municipality’s population resulted in a 1.5-3 percentage point decline in anti-immigrant voting.
In Upper Austria, the assignment of refugees improved attitudes towards refugees and reduced the vote share of the far-right Freedom Party (FPO) by approximately 3.5 percentage points in 2015.35 Interestingly, however, the FPO’s share of the vote increased by 2.7 percentage points in municipalities that refugees travelled through on their way to the German border. This result is in line with the contact theory, in the sense that locals who had regular contact with refugees that had settled in their municipalities developed sympathy for them. It is important in this regard that refugee resettlement programmes limit the number of refugees to 1.5 per cent of the local population (with refugees averaging no more than 1.35 per cent of the population across the municipality as a whole).
Similar results can be seen in France, where a recent study found that Marine Le Pen’s share of the vote fell in the 2017 presidential election in municipalities that had received refugees following the dismantling of the “Calais Jungle” refugee camp.36 In this case, the average municipality received 1.7 refugees per 100 inhabitants. That study also found that the pro-immigrant effect declines as the number of refugees increases, turning negative at 4 refugees per 100 inhabitants.
The importance of the intensity of exposure to refugees is confirmed by Dinas et al. (2017), who looked at the impact that refugees arriving on Greek islands had on the Greek elections in January and September 2015. Here, the intensity of immigration was far higher than in Austria or France. Indeed, among islands that were exposed to arrivals of refugees, the median island received 2.5 refugees per local resident (with one island receiving 125 refugees per local resident). In this case, the arrival of refugees increased the vote share of the far-right Golden Dawn party by between 2.5 and 4.5 percentage points.
Burkart (2018) identifies a similar effect in Sweden. Using a quasi-random allocation of refugees, he shows that an increase of 1 percentage point in refugees’ share of the population resulted in an increase of about 0.6 percentage point in the vote share of the anti-immigrant Sweden Democrats party in the 2014 election relative to the 2010 election.
Becker and Fetzer (2016) find that eastern European migrants who came to the United Kingdom following the opening-up of the UK labour market in 2004 had a similar effect. That migration wave was very substantial (totalling 1 million people – 3 per cent of the UK labour force). Their study finds that eastern European migration explains about 2 percentage points of the increase in UKIP’s share of the vote in the European elections in 2009 and 2014 relative to the 1999 and 2004 elections.
Other historical studies also point to a causal link between immigration and voting for the far right. Barone et al. (2016) examined Italy’s 2001, 2006 and 2008 elections, finding that a 1 percentage point increase in immigrants’ share of the population led to a 1.3 percentage point increase in the vote share of the anti-immigrant centre-right coalition led by Silvio Berlusconi. Halla et al. (2017) studied Austrian elections from 1983 to 2013, finding that a 1 percentage point increase in immigrants’ share of the population resulted in a 0.16 percentage point increase in the FPO’s vote share. And Edo et al. (2017) studied French presidential elections from 1988 to 2017, finding that immigration boosted support for the far right, but not the far left. What is more, that effect was a strong one: a 1 percentage point increase in immigrants’ share of the population resulted in a 2 percentage point rise in the far right’s share of the vote.
Thus, the results of such research tend to point to a positive correlation between immigration and anti-immigrant voting – although some studies do support the contact theory (especially when the number of refugees is sufficiently small). The effect of immigration may also depend on the skill sets of the immigrants in question. For example, Mayda et al. (2018) studied US elections from 1990 to 2010, showing that high-skilled immigration reduces the Republican vote share in presidential elections, while low-skilled immigration increases it. The magnitude of that impact is similar to those observed in other studies: a 1 percentage point increase in immigrants’ share of the population leads to a 1 percentage point change in the vote share in one direction or the other. The impacts on Senate and House elections have the same sign, but are smaller in magnitude.
In contrast to the studies above, which investigate the impact that migration has on populist voting in individual Western countries, Guriev et al. (2018) conduct Europe-wide analysis of 510 elections in 160 NUTS-2 regions in 19 countries since 2001. They follow the approach developed by Algan et al. (2017) when it comes to identifying right-wing, left-wing and other populist parties. Their dataset includes five post-communist countries (Bulgaria, the Czech Republic, Poland, Romania and the Slovak Republic). Given the demographic and migration challenges in eastern and southern Europe, their analysis considers gross cross-border emigration and gross cross-border immigration flows separately – also controlling for regional and time fixed effects, regional unemployment and regional intra-country migration. Their identification of the causal effects that gross immigration flows have on populist voting is based on the strategy detailed above, using immigrants’ pre-crisis population shares in the relevant subnational regions.
The results of their analysis can be found in Table 3.6.1. Immigration is associated with declines in the vote shares of left-wing and non-anti-immigrant populists. The magnitude of that impact is substantial: a 1 percentage point rise in immigrants’ share of the population since the previous election is associated with a 6 percentage point decline in the combined vote share for left-wing populist parties (and a 7 percentage point decline in the combined vote share for non-anti-immigrant populist parties) in a given election. While the impact observed on the vote share of right-wing and anti-immigrant populist parties is positive, it is not statistically significant. Meanwhile, emigration has a positive effect on the vote share of left-wing populist parties. Overall, these Europe-wide results confirm the mixed messages of the single-country studies detailed above, showing that the impact of immigration varies when it comes to the vote shares of right-wing and anti-immigrant parties.
|Dependent variable (vote share)||All populists||Right-wing populists||Left-wing populists||Other populists||Anti-immigrant populists||Non-anti-immigrant populists|
|External immigration rate||-6.12**||1.34||-5.65**||-1.80||0.64||-6.75**|
|External emigration rate||4.23||-1.84||4.93**||1.13||0.88||3.35|
|Net internal migration rate||0.94||-0.19||0.86||0.27||-0.00||0.94|
|Number of observations||510||510||510||510||510||510|
SOURCE: Guriev et al. (2018).
NOTE: These estimates are derived from a second-stage estimation, with immigration instrumented by immigrants’ share of the population at the beginning of the period, interacted with Europe-wide immigration rates. “Other populists” are populist parties that are neither right-wing nor left-wing. Fixed effects for NUTS-2 regions and year dummies are included in all specifications. The unit of observation is the NUTS-2 region. Robust standard errors are reported in parentheses (clustered at the country level), and *, ** and *** denote values that are statistically significant at the 10, 5 and 1 per cent levels respectively.
Annex 3.1. Estimating determinants of migration intentions for potential migrants
The following linear probability models are used to study determinants of intentions to migrate abroad:
The dependent variable, , takes a value of 100 if individual is planning to move permanently to another country within the next 12 months and 0 otherwise. and denote country of origin and year fixed effects respectively. is a set of demographic characteristics and education and labour market outcomes including gender, age bracket, marital status, level of education, employment, place of residence (urban or rural) and presence of children under the age of 15 in the household.
The second set of controls () includes the log differential between the individual’s current wage and the average expected wage in the stated destination country. The expected wage in the country of destination was obtained using a set of multivariate regressions, estimated separately for each country and each survey year, whereby a person’s annual log wage is explained by their age, gender and education. The coefficients from those models were then applied to each person on the basis of their preferred country of destination, assuming that returns to age, gender and education in the country of destination are the same for locals and migrants alike. Where respondents intended to migrate, but did not identify an intended destination country, the wage differential and the variables reflecting the cost of migration were calculated as weighted averages on the basis of the preferred destinations of individuals from the same country of origin with the same level of education in the same survey year.
Controls also include the log of the distance between the most populous cities in the two countries, dummy variables for a common border between the two countries and a common colonial history, and an index of linguistic proximity. Additional variables capture social networks abroad and at home.
The third set of controls () captures individual perceptions and attitudes. It includes an index reflecting life satisfaction (on a scale of 0 to 10), a dummy variable which takes the value of 1 if the respondent feels safe walking alone at night in their place of residence, a measure of trust in the national government, a dummy variable which is equal to 1 if the respondent believes that corruption is widespread within businesses located in their country, and a similar dummy for corruption in the government.
A set of satisfaction indicators measure whether an individual is satisfied with (i) the quality of air and water, (ii) the country’s education system and schools, (iii) the availability of high-quality healthcare, (iv) the availability of good affordable housing, and (v) the quality of roads, highways and public transport systems. The satisfaction index indicates the average level of satisfaction across all of these various aspects.
Annex 3.2. Data on refugees and irregular migrants, populations of countries of origin and country characteristics
Data on refugees and irregular migrants
Data on refugees and irregular migrants are taken from the three waves of FMS surveys conducted by the IOM. The FMS survey seeks to produce quantitative estimates of the numbers of third-country nationals originating from outside the European Union who are migrating towards Europe via the central and eastern Mediterranean routes. The first wave (October to December 2015) saw interviews being conducted in Bulgaria, Croatia, FYR Macedonia, Greece, Hungary, Italy, Serbia and Slovenia. The second wave (January to November 2016) covered Bulgaria, Croatia, FYR Macedonia, Greece, Hungary, Serbia and Slovenia. And the final wave (June to November 2016) covered Italy. All in all, this sample includes data on nearly 21,000 individuals.
FMS data contain a wealth of information on migrants’ demographic characteristics (age, gender, level of education and marital status), their employment status prior to migration, key transit points on their route, the cost of the journey, their reasons for leaving and their intended destination.
Those surveys, which are conducted in a total of 11 different languages, are administered by trained data collectors with a range of cultural backgrounds. Respondents are approached by IOM field staff, told about the objectives of the research and advised that participation will not influence their legal status in the country where the interview is conducted. Despite fast-changing conditions on the ground, FMS data do provide a good picture of migrant groups.37
Population data for refugees and irregular migrants’ countries of origin
Population data for refugees and irregular migrants’ countries of origin are taken from the Gallup World Polls conducted between 2009 and 2014. These nationally representative surveys are conducted every year in more than 160 economies around the world. Each survey round covers approximately 1,000 individuals in each economy. These surveys provide detailed information on people’s demographic characteristics (such as their age, gender, level of education and marital status), as well as labour market outcomes, income levels and intentions to migrate within the next 12 months.
The Gallup and FMS data have been combined in a single sample in order to provide information on the pre-migration populations of refugees and irregular migrants’ countries of origin (those with at least 100 respondents in the FMS dataset). Each variable in the Gallup dataset has been brought into line with the definitions used in the FMS data.
The Uppsala Conflict Data Program (UCDP) dataset on conflict-related deaths has been used to classify refugees and irregular migrants’ countries of origin in terms of the intensity of conflicts. Countries are regarded as experiencing a major conflict if 1,000 or more conflict-related deaths occur in a single year between 2009 and 2014, with all other countries being placed in the “minor/no conflict” category. The results remain broadly unchanged if a year-specific measure of conflict intensity is used, with few countries moving between the two categories from one year to the next.
Country-level unemployment rates have been taken from the World Bank’s World Development Indicators database. For details of the compilation of the MIPEX, see Huddleston et al. (2015). Information on the average duration of asylum procedures comes from Eurofound and represents the average number of months between the submission of an asylum claim and the initial decision. The waiting time before accessing the labour market comes from the OECD and represents the time, in months, that it takes to obtain a work permit after successfully claiming asylum. For ease of interpretation, both variables have been rescaled such that they range from 0 to 1. The social expenditure indicator comes from the OECD and is measured as a percentage of GDP.
Annex 3.3. Estimating socio-demographic characteristics of refugees and irregular migrants
This analysis uses a series of multivariate regressions to assess the ways in which the socio-demographic characteristics of refugees and irregular migrants differ from those of the general population in their economies of origin.38 The main explanatory variables of interest are age, gender, marital status and level of education. In some specifications, estimated incomes are also included in order to understand differences between the earning potential of refugees and irregular migrants from different economies.
The following linear probability model is estimated using a combined sample comprising individuals completing FMS surveys and Gallup World Polls:
where the dependent variable, , takes a value of 1 if individual from country is in the FMS sample and 0 if he/she is in the Gallup sample. The set of controls () includes age groups, levels of education, employment status prior to migration (employed or not) and marital status. A set of source country () dummies control for all time-invariant differences across countries of origin. The analysis focuses on individuals aged 25 and over, who are therefore likely to have completed their education.
Models are estimated separately for men and women and for different levels of conflict in countries of origin. Logit regressions (not reported) yield similar results.
The regressions reported in Tables 3.5 and 3.6 are based on a sample of refugees and irregular migrants whereby the outcome variables () represent various characteristics of the stated destination country (). The same set of control variables is included in these specifications.
Y.E. Akgündüz and H. Torun (2018)
“Two and a half million Syrian refugees, skill mix and capital intensity”, GLO Discussion Paper No. 186.
C.G. Aksoy and B. Ozcan (2018)
“The Impact of Immigration on Natives’ Fertility: Evidence from Syrians in Turkey”, EBRD working paper, forthcoming.
C.G. Aksoy and P. Poutvaara (2018)
“Self-Selection of Refugees and Other Migrant Groups”, EBRD working paper, forthcoming.
Y. Algan, S. Guriev, E. Papaioannou and E. Passari (2017)
“The European trust crisis and the rise of populism”, Brookings Papers on Economic Activity, Fall.
G. Allport (1954)
The Nature of Prejudice, Addison-Wesley.
D. Bahar, A. Hauptmann, C. Ozgüel and H. Rapoport (2018)
“Let their knowledge flow: the effect of Yugoslavian returning refugees on export productivity”, mimeo.
G. Barone, A. D’Ignazio, G. de Blasio and P. Naticchioni (2016)
“Mr. Rossi, Mr. Hu and Politics. The Role of Immigration in Shaping Natives’ Voting Behavior”, Journal of Public Economics, Vol. 136, pp. 1-13.
S.O. Becker and T. Fetzer (2016)
“Does Migration Cause Extreme Voting?”, CAGE Working Paper No. 306.
G.J. Borjas (1987)
“Self-selection and the earnings of immigrants”, American Economic Review, Vol. 77(4), pp. 531-553.
S. Burkart (2018)
“Asylum seekers and populist voting in Sweden”, background note for this Transition Report.
R. Cai, N. Esipova, M. Oppenheimer and S. Feng (2014)
“International migration desires related to subjective well-being”, IZA Journal of Migration, Vol. 3.
E. Ceritoğlu, H.B. Gürcihan-Yüncüler, H. Torun and S. Tümen (2017)
“The impact of Syrian refugees on natives’ labor market outcomes in Turkey: evidence from a quasi-experimental design”, IZA Journal of Labor Policy, Vol. 6.
P. Choudhury (2016)
“Return migration and geography of innovation in MNEs: a natural experiment of knowledge production by local workers reporting to return migrants”, Journal of Economic Geography, Vol. 16, pp. 585-610.
A. Clark (2007)
“Understanding community: a review of networks, ties and contacts”, ESRC NCRM Working Paper No. 9/07.
T.H. Dao, F. Docquier, C. Parsons and G. Peri (2018)
“Migration and Development: Dissecting the Anatomy of the Mobility Transition”, Journal of Development Economics, Vol. 132, pp. 88-101.
X.V. Del Carpio and M. Wagner (2015)
“The impact of Syrian refugees on the Turkish labor market”, World Bank Policy Research Working Paper No. 7402.
E. Dinas, K. Matakos, D. Xefteris and D. Hangartner (2017)
“Waking Up the Golden Dawn: Does Exposure to the Refugee Crisis Increase Support for Extreme-Right Parties?”, mimeo.
F. Docquier, B.L. Lowell and A. Marfouk (2009)
“A gendered assessment of highly skilled emigration”, Population and Development Review, Vol. 35, pp. 297-321.
F. Docquier, G. Peri and I. Ruyssen (2014)
“The cross-country determinants of potential and actual migration”, International Migration Review, Vol. 48, pp. S37-S99.
F. Docquier and H. Rapoport (2012)
“Globalization, Brain Drain, and Development”, Journal of Economic Literature, Vol. 50, pp. 681-730.
C. Dustmann, F. Fasani, T. Frattini, L. Minale and U. Schönberg (2017)
“On the economics and politics of refugee migration”, Economic Policy, Vol. 32, pp. 497-550.
C. Dustmann and A. Okatenko (2014)
“Out-migration, wealth constraints, and the quality of local amenities”, Journal of Development Economics, Vol. 110, pp. 52-63.
C. Dustmann, K. Vasiljeva and A. Piil Damm (2016)
“Refugee Migration and Electoral Outcomes”, CReAM Discussion Paper No. 19/16.
A. Edo, Y. Giesing, J. Oztunc and P. Poutvaara (2017)
“Immigration and electoral support for the far left and the far right”, CEPII Working Paper No. 2017-20.
V.A. Epanechnikov (1969)
“Non-Parametric Estimation of a Multivariate Probability Density”, Theory of Probability and its Applications, Vol. 14, pp. 153-158.
M. Erdoğan (2014)
“Syrians in Turkey: Social acceptance and integration research”, Hacettepe University Migration and Politics Research Centre (HUGO).
L. Foster, J. Haltiwanger and C. Syverson (2008)
“Reallocation, Firm Turnover, and Efficiency: Selection on Productivity or Profitability?”, American Economic Review, Vol. 98(1), pp. 394-425.
G. Friebel, M. Manchin, M. Mendola and G. Prarolo (2017)
“Human smuggling and intentions to migrate: global evidence from a supply shock along Africa-to-Europe migration routes”, DEMS Working Paper No. 375.
Y. Giesing and N. Laurentsyeva (2018)
“Firms Left Behind: Emigration and Firm Productivity”, CESifo Working Paper No. 6815.
S. Guriev, N. Melnikov and D. Treisman (2018)
“Immigration, Emigration, and Populist Vote in Europe”, background note for this Transition Report.
M. Halla, A. Wagner and J. Zweimüller (2017)
“Immigration and Voting for the Far Right”, Journal of the European Economic Association, Vol. 15, pp. 1341-1385.
T.J. Hatton (2017)
“Refugees and asylum seekers, the crisis in Europe and the future of policy”, Economic Policy, Vol. 32, pp. 447-496.
T. Huddleston, O. Bilgili, A.-L. Joki and V. Zvezda (2015)
“Migrant Integration Policy Index 2015”, Barcelona Center for International Affairs (CIDOB) and Migration Policy Group (MPG).
S.P. Kerr, W. Kerr, C. Ozden and C. Parsons (2016)
“Global talent flows”, Journal of Economic Perspectives, Vol. 30, pp. 83-106.
J.J. Lewer, G. Pacheco and S. Rossouw (2009)
“Do Non-Economic Quality of Life Factors Drive Immigration?”, IZA Discussion Paper No. 4385.
P. Mahajan and D. Yang (2017)
“Taken by storm: hurricanes, migrant networks, and U.S. immigration”, NBER Working Paper No. 23756.
A. Mansoor and B. Quillin (eds.) (2006)
Migration and Remittances: Eastern Europe and the Former Soviet Union, World Bank.
J. Markusen and N. Trofimenko (2009)
“Teaching locals new tricks: Foreign experts as a channel of knowledge transfers”, Journal of Development Economics, Vol. 88, pp. 120-131.
A.M. Mayda (2005)
“International migration: a panel data analysis of economic and non-economic determinants”, IZA Discussion Paper No. 1590.
A.M. Mayda, G. Peri and W. Steingress (2018)
“The Political Impact of Immigration: Evidence from the United States”, CEPR Discussion Paper No. 12848.
Flexible Solidarity: A comprehensive strategy for asylum and immigration in the EU – 2018 MEDAM assessment report on asylum and migration policies in Europe, Mercator Dialogue on Asylum and Migration, Kiel.
K. Munshi (2003)
“Networks in the modern economy: Mexican migrants in the U.S. labor market”, The Quarterly Journal of Economics, Vol. 118, pp. 549-599.
Engaging with Employers in the Hiring of Refugees, Paris.
D.A. Plane and P.A. Rogerson (1994)
The Geographical Analysis of Population: With Applications to Planning and Business, John Wiley.
A. Raggl (forthcoming)
“Migration intentions in CESEE – Evidence from OeNB Euro Survey data”, Focus on European Economic Integration, Oesterreichische Nationalbank.
J.M.C. Santos Silva and S. Tenreyro (2006)
“The log of gravity”, The Review of Economics and Statistics, Vol. 88, pp. 641-658.
A. Steinmayr (2018)
“Contact Matters: Exposure to Refugees and Voting for the Far-Right”, mimeo.
UN DESA (2017a)
Trends in international migrant stock: The 2017 revision, New York.
UN DESA (2017b)
International Migration Report 2017, New York.
Global Trends: Forced Displacement in 2017, Geneva.
P. Vertier and M. Viskanic (2018)
“Dismantling the ‘Jungle’: Migrant Relocation and Extreme Voting in France”, CESifo Working Paper No. 6927.
World Bank (2018)
Asylum seekers in the European Union: building evidence to inform policy making, Washington, D.C.
W. Zelinsky (1971)
“The Hypothesis of the Mobility Transition”, Geographical Review, Vol. 61, pp. 219-249.
N. Zoubanov (2004)
“Assessing general and country-specific determinants of migration in the European Union: a panel data approach”, paper presented at the IZA Summer School.
Subscribe to the EBRD Transition Report 2018-19 mailing list.
Sign up to receive important announcements and emails about the Transition Report.