1 Introduction

Eastern enlargements of the European Union to eight Central and Eastern European countries (EU8) together with Cyprus and Malta in 2004 and Romania and Bulgaria (EU2) in 2007 were unprecedented in many aspects. Leaving aside Cyprus and Malta, the large population size of the acceding block and substantial income differentials between the old and new EU members generated fears of a huge influx of Central and Eastern European migrants who would settle permanently in the old EU15 countries, leading to benefits shopping and negative impacts on the receiving countries’ labor markets.Footnote 1 However, a significant proportion of these migrants considered and indeed stayed abroad temporarily (Polard et al. 2008; European Commission 2008; Eurofound 2012). Overall, immigration from the EU8 and EU2 has increased the EU15 population by approximately 1 % after the enlargements, with around 1.8 % and 4.1 % of the respective regions’ population having moved into the EU15 (Holland et al. 2011; Brücker and Damelang 2009; Brücker et al 2009).

More details on post-enlargement migration are available also in the reports, such as the recent economic crisis and deterioration of the economic situation in many receiving countries suggests a possible slowdown in migration flows from the new EU members, as well as an increase in return migration or so-called accelerated returns, when return occurs earlier than planned. However, the economic crisis has affected both receiving and sending countries, with several countries of origin hit harder than the destinations. For example, while the unemployment rate was over 20 % in Spain and 14 % in Ireland in 2011, it was 8 % in the UK and around 6 % in Germany. On the other hand, unemployment reached double-digit figures in Bulgaria, Hungary, Slovakia and Estonia in 2011, was close to 10 % in Poland, and over 15 % in Latvia and Lithuania. Consequently, if the conditions at home are persistently worse than those in the destination, return migration might be delayed. Thus, the return migration of post-enlargement migrants is becoming an increasingly important issue, and one upon which this chapter aims to shed more light.

In general, migrants are usually highly responsive to economic cycles and act as “buffers” in the labor markets. In “good times”, they solve the excess labor demand problem in host countries by filling the available vacancies and contribute to ease the inflationary pressures; moreover, they also efficiently relieve sending countries’ labor markets of excess labor, and help to alleviate unemployment problems and raise wages. In times of trouble, they return to their countries of origin or move onwards to other destinations where work is available. Repeat or circular migration generally leads to a triple-win situation, with both countries of origin and destination gaining from highly flexible migration, while the migrants themselves also benefit.Footnote 2

Return migration is usually beneficial for the home countries, given that returnees bring new skills and competences, increase the overall human capital and enhance productivity and employment. Such “brain-gain” or “brain circulation” counter-acts the negative “brain-drain” effects of the emigration of highly skilled individuals, and is expected to generate positive effects for the country of origin’s labor markets and overall growth prospects, including indirectly via the so-called “incentive” effect, raising the average schooling of stayers in the home country. Sending countries also benefit from remittances, particularly from temporary migrants who plan to return. Finally, return (and repeat) migration might enhance trade and investment with host countries through network effects, especially in the case of skilled migration. Migration policies are also extremely important, with migrants being more likely to engage in return and circular migration when such policies are less restrictive (Constant and Zimmermann 2011).

Theoretical models of return migration postulate that if migration is temporary, the decision to move will not only be based on immediate and future incomes in the destination, but also the expected future returns in the country of origin. The main motives for return include a preference for consumption in the home country, family and other networks at home, taking advantage of differences in relative prices between host and home countries (high wages abroad and low prices at home), and the possibility to accumulate human capital abroad, the returns to which are higher at home. A return move is also realized when any pre-defined objectives are achieved, such as completing education or accumulating a certain amount of wealth.

A crucial issue in this context is the selection of returnees, or rather the double selection in the decisions to first migrate and subsequently come back, since any impact on the country of origin will generally depend on returnees’ characteristics. Mayr and Peri (2009) developed a theoretical model for the decisions to migrate, return and invest in schooling, simulated for the East–west case. The model predicts that if all returnees in the home country receive the same skill premium for having been abroad, they will be negatively selected but will still have intermediate education since migrants are usually positively selected on education. However, if the skill premium increases upon return with the level of education, only the most educated will return. Combined with the fact that some highly educated migrants return and other potential migrants end up staying in the East, the aforementioned “incentive” effect would offset the negative effect of –brain-drain accordingly.

Regarding post-enlargement migrants, some might prefer to return home due to economic reasons (such as job loss, given that many of them were employed in the sectors most affected by the crisis), and others due to initially temporary intentions to stay abroad or emerging social pressures exacerbated by economic difficulties in some destination countries. In addition, highly qualified individuals might find it easier to find adequate jobs at home. Moreover, as most of these migrants have been overrepresented in sectors and jobs below their qualifications, such workers might prefer to return home and work in jobs that match their qualifications. Finally, with the exceptions of Bulgaria and Romania, no restrictions on movement exist for these migrants,Footnote 3 rendering a potential repeat move a feasible and certain future option and thus enhancing the return. However, on the other hand, the worsening economic situation at home reflects an important push factor that might prevent the return, while established networks abroad and improved language could act as further return-deterring factors. Indeed, anecdotal evidence and mass media in some new member states suggest a decrease in return migration in recent years, and rather a “highest ever emigration”. Overall, East–west migration flows are likely to continue in the future, although their composition might change.

The remainder of this chapter is organized as follows. Section 2 reviews existing studies on return migration in the new EU member states. Section 3 presents an attempt to measure the return migration in the post-enlargement Europe. Characteristics of returnees, their selection patterns and determinants of return migration are analyzed in Sects. 4, and 5 outlines potential future East–west migration by investigating the migration intentions of both return migrants and stayers in the new EU member states. Finally, Sect. 6 presents some concluding remarks.

2 Return Post-enlargement Migration: Empirical Studies

Return migration in Central and Eastern Europe is a relatively recent phenomenon, and has attracted increasing academic and policy attention in the last few years. Iara (2008) and Martin and Radu (2009) are probably the first systematic cross-country studies of return migration in Eastern Europe (although the former only focuses on young men in 2003). Martin and Radu (2009) employ the European Social Survey data for 2006 to document shares of return migrants in the Czech Republic, Hungary, Latvia, Poland, Romania and Slovakia. Furthermore, they use EU Labour Force Survey (LFS) data for Poland, Hungary, Latvia, Lithuania and Romania for 2002–2007 to investigate the characteristics and performance of the return migrants. Returnees in their study are defined as individuals who were born in the country of their current residence but resided abroad the year before the survey, thus only capturing recent returns. The proportion of returnees was found to be the largest in Poland (7.97 %) and smallest in Hungary (2.61 %). Regarding characteristics, consistent with theoretical literature and empirical studies of return migration for other countries, they find that returnees are predominantly male, young, not married and with a medium or high level of education, thus suggesting a positive selection on education (albeit not in Romania). Their estimated income premium for returnees ranges from 10 % to 30 %, and some evidence was also found for negative selection on unobservables for returnees. Finally, returnees were estimated to participate less in the labor market, and to switch to self-employment; however, these results were not robust to the estimation method employed.

Two recent reports have focused on return migration in the post-enlargement Europe. Eurofound (2012) studies the impact of the recent return migration in the EU, including case studies for Hungary, Latvia, Poland and Romania. However, the report acknowledges that measuring return migration is challenging, since no comparable data across countries exist due to different data collection methods and definitions (stocks vs. flows, age range of migrants, time spent abroad, etc.).Footnote 4 Despite this major difficulty, the authors summarize evidence from several existing studies, as well as their own estimates, concluding that no mass return migration took place during the recent crisis.

They also show that outmigration from the EU countries and Norway only increased substantially for Romanian nationals among the countries in the analysis, according to the OECD outflows data in 2009. Indeed, as also mentioned in Holland et al. (2011) and Koehler et al. (2010), since the economic situation in home countries also deteriorated during the crisis, emigration rates from the destination countries did not “rocket”, with migrants rather adopting a so-called “wait-and-see” strategy. Based upon the increased outflows from some destination countries, the crisis has accelerated planned returns, if anything, given that some of them took place earlier than planned, as well as the likely onward migration of EU8 and EU2 migrants and growth in emigration from the EU8 and EU2 countries (Koehler et al. 2010; Eurofound 2012). There was also some heterogeneity across countries: emigration slowed down in Poland and Romania, and return and circular moves increased; emigration increased and return migration diminished in Latvia; while the impact of the crisis was minor in Hungary. In terms of characteristics, returnees are mid-age (under 45), single, male and employed in low-skilled jobs abroad, including those with higher education (ibid), thus suggesting that overqualified individuals prefer to come back, pointing towards the potential danger of brain-waste. Case studies also suggest that the most important return-deterring factors are higher incomes abroad, access to unemployment and other benefits, as well as the higher probability of finding a job abroad, while family reasons were named as the main motivation for return (either to join family left behind, or with family as a tied-mover or due to a child’s education). In addition, the improving economic situation and better career opportunities at home, social networks that facilitate job-finding, and particularly working according to one’s qualification and positive public perceptions also facilitate return. Finally, many returnees did not rule out the possibility of repeat emigration, with a significant proportion considering possible repeated periods of work abroad in the future, pointing towards a potential circular nature of future East–west migration.

The second report, Re-Turn (2012), is based on case studies of return migration for Czech Republic, Hungary, Poland, Slovenia and Slovakia (as well as Austria, Germany and Italy). The authors document that while returnees constitute a majority among all immigrants in Poland, non-nationals immigrate most in other new member states, thus importantly suggesting that EU8 countries are becoming new immigration countries. The authors employ the EU LFS data for 2005–2008 to document characteristics of the returnees, showing that recent returnees (as above, those who resided abroad 1 year before the survey) are younger and positively selected in terms of education, compared to stayers and migrants. However, they are also more often unemployed and have a higher probability of not participating in the labor market. As in Martin and Radu (2009), this result might be due to the analysis only including recent returnees, for whom the probability of finding a job during 1 year might be lower. If employed, returnees are more likely to work in the service sector (apart from the case of Poland). The authors also document that returnees do not lose connection with their home country while abroad, which could help them to re-integrate upon return. Interestingly, the majority of returnees in Poland state that work abroad has not changed their career path or even enhanced its fragmentation, with only 8 % reporting being able to enhance their career upon return; however, the majority among returnees with tertiary education did make use of their experience abroad (Vavreckova 2009; Grabowska-Lusinska 2010, both mentioned in Re-Turn 2012).

A number of studies have examined recent return migration in a single country context. For example, Hazans (2008) provides econometric analysis of the return migration premium for Latvian returnees using the LFS data, while Hazans and Philips (2010) analyze returnees in Estonia, Latvia and Lithuania as part of their study. After having controlled for demographic characteristics, human capital and experience abroad, Hazans (2008) finds a significant earnings premium for returnees in Latvia, which is larger for men than women (more than 20 % vs. 6 %). Furthermore, he also shows that this earnings premium is caused by foreign work experience. Hazans and Philips (2010) document that most migrants in the Baltics return within the first 2 years, and even within a year for more than half of Latvian movers, while the intended duration of further stay abroad increases with the time spent there. The authors also find higher inactivity and unemployment among returnees than stayers, arguing that this is not necessarily related to their weak labor market performance, but might rather reflect that returnees can afford to search longer due to accumulated savings, or perhaps simply plan to work abroad again and thus take no job at home. Comparing the characteristics of migrants, stayers and returnees, the authors show that returnees are more educated than migrants (and stayers in Latvia). Moreover, they also work in more skilled occupations than migrants, but are less skilled than stayers and less frequently overqualified for their jobs than migrants. They further have a higher incidence of temporary and short-term contracts than stayers, although this incidence is lower than for current migrants. Pungas et al. (2012) analyze the return intentions of Estonian migrants in Finland, showing that those who work below their qualifications have a higher tendency to return. Based on data analysis, case studies and interviews, Barcevicius and Zvalionyte (2012) provide a comprehensive study of return migration in Lithuania, showing that family and cultural motives are most important for return decisions. Furthermore, the authors show that work experience abroad is beneficial for those who worked according to their qualifications: they have higher earnings upon return, along with those who received a degree abroad. The authors also document a very high incidence of repeat emigration, with 25 % of the Lithuanian respondents with migration experience showing “firm” intentions to move again. Grabowska-Lusinska (2010) studies returnees in Poland, showing that the largest share of returnees is from Germany, while the shares of emigrants are the largest for the UK and Germany. Returnees are mainly of mid-age, with primary vocational education, with a high share not employed. Interestingly, the number of returnees applying for unemployment benefits in Poland doubled over 2008–2009, suggesting that returning “into benefits” might reflect one strategy of coping with the crisis. However, most of the interviewed Polish migrants in the UK did not intend to return due to the crisis, with the main reasons for return also being family-related in this case.

It is also possible to gain useful insights into the situation of return migrants from the receiving countries’ perspective. The deterioration of the Irish economy and a sharp increase in unemployment during the crisis mostly affected industries and sectors with large concentrations of migrants from the new member states (NMS), such as construction, agriculture, wholesale and retail trade (see, among others, Barrett and Kelly 2012; Papademetriou et al. 2010; Koehler et al. 2010, for the analysis of the Irish situation). Consequently, migrants from the new EU countries have experienced the largest increase in unemployment. Moreover, controlling for other socio-economic characteristics, this is only immigrant group that appears more likely to be employed compared to natives, and the only one whose employment prospects were negatively affected by the downturn, particularly for men (Barrett and Kelly 2012). The large increase in immigrants’ benefits claims during a recession was also documented for Ireland, with immigrants from the new member states representing the largest group. Outflows of migrants from the new EU members from Ireland have increased due to the deteriorating economy. By contrast, existing evidence suggests that only migrants from the new members states have experienced no significant rise in unemployment in the UK to date, while less educated, younger immigrants, those from poor countries, non-English speaking and ethnic minorities have generally faced an increase in unemployment (Papademetriou et al. 2010; Koehler et al. 2010). An increase in out migration in the UK started in 2008, and was the largest for the NMS migrants, although it subsequently fell in 2009. The situation was somewhat different in Spain: despite the sharp increase in unemployment, especially in construction and for immigrants from Asia and Africa, immigration from the EU2 countries and particularly Romania did not decrease (at least until 2009), although Spain experienced a slow-down in overall immigration (Papademetriou et al. 2010; Koehler et al. 2010). In fact, these trends possibly contributed to the re-introduction of the restrictions on free movement for the EU2 countries in July 2011.

In Germany, the crisis hit companies with relatively highly-qualified workers, and particularly export-oriented and skill-intensive manufacturing companies (Rinne and Zimmermann 2012; Kim 2010; Papademetriou et al. 2010; Koehler et al. 2010) with lean labor cost structures. Job losses were rather moderate, since crisis-affected companies used time-accounts and labor hording supported by government-sponsored reduction in working hours as strong measures of internal flexibility. Consequently, there was almost no increase in unemployment for migrants in Germany. Regarding post-enlargement migration, immigration from the EU10 countries continued to increase.

Overall, emerging empirical studies do not suggest massive return migration during the crisis, especially if home countries are affected relatively more by the crisis. Moreover, some migrants might have chosen to migrate onwards to other destinations less affected by the recession (such as Norway or Germany). There is also some evidence from receiving countries, particularly the UK and Ireland, that migrants from the EU emigrated in larger numbers than non-EU migrants, despite unemployment often being higher in the latter group, thus reflecting the role of restrictive migration policies (Koehler et al. 2010). Most importantly, the remaining transitional arrangements for the EU2 migrants might actually backfire, given that these migrants will possibly chose to remain in the host country after becoming unemployed due to uncertainty about the future possibility of re-entering the destination country (Koehler et al. 2010; Holland et al. 2011).

3 How Large Is Return Migration in the New Member States?

While return migration is difficult to measure, making cross-country comparisons is even more complicated. We start by documenting immigration into the UK and Ireland, two countries that have attracted a substantial proportion of immigrants from the new EU member states. As seen from Fig. 1, there was a slowdown in immigration to these countries during the crisis; however, while immigration continued to decline in Ireland, the trend in the UK was reversed in 2009, with immigration from the EU8 and EU2 countries starting to increase again. Moreover, while immigration to Ireland declined from all NMS (apart from Romania), it increased in the UK, especially from Lithuania and Latvia (with a slowdown in 2011), even surpassing their post-accession peaks.

Fig 1
figure 1

Immigration to the UK and Ireland of the EU8 and EU2 nationals (Source: http://statistics.dwp.gov.uk/asd/asd1/niall/index.php?page=nino_allocation and http://www.welfare.ie/EN/Topics/PPSN/Pages/ppsstat.aspx). Notes: NINOs refer to National Insurance Numbers (in thousands). PPSNs refer to Personal Public Service Numbers (in thousands)

Full size image

Turning now to the outmigration, Fig. 2 shows the outflows of EU10 nationals from the EU15 and Norway over 2004–2010, based on the data from the OECD International Migration Database. There is some heterogeneity across countries, with outflows from all EU15 countries having increased during the crisis for Bulgaria, Latvia and Lithuania and slowed down or decreased for the remainder of the new EU members.

Fig 2
figure 2

Outflows of the EU10 nationals from the EU15 and Norway, 2004–2010 (Source: OECD International Migration Database, Outflows of foreign population. OECDStats. Notes: In thousands. Data is missing for outflows from France, Greece, Ireland, Portugal, and generally Belgium and the UK. For Italy, data for 2010 is missing. PL: with UK for 2007–2010; with Belgium for 2004–2009; BG: with Belgium for 2008, 2009; CZ: with Belgium for 2008, 2009; HU: with Belgium for 2008, 2009; LT: with Belgium for 2008, 2009; RO: with Belgium for 2008, 2009; SK: with Belgium for 2008, 2009; SI: with Netherlands in 2009, 2010)

Full size image

However, it is important to note that these outflows do not necessarily reflect return migration, since it is not known where these migrants went to. Moreover, increased outflows might also reflect the increased inflows and onward migration.

We now turn to the sending countries data to document the return migration, plotting the proportion of returnees in the population aged 24–65 in Fig. 3. We employ two datasets containing information on return migrants, specifically the European Social Survey (ESS) rounds 4 and 5 for 2008 and 2010 and the Eurobarometer EB73-3 for March–April 2010. In the ESS data, returnees are defined as persons who were born in the country and spent at least 6 months working abroad over the last 10 years, before returning. In the Eurobarometer data, it is possible to distinguish between return labor migrants, students who went to study abroad and came back, and other returnees. For comparative purpose, we retain return labor migrants who were born in the country and have worked (including volunteering and traineeships) for at least three consecutive months in a country other than their country of birth. Employing these “stock” definitions and first considering the ESS data suggests that the share of returnees increased over the 2 years in Bulgaria, Hungary and Poland, and declined in the Czech Republic, Estonia, Slovenia and Slovakia. As expected, the proportion of returnees is larger in all countries using the Eurobarometer data (apart from the Czech Republic, where the difference is negligible), which is attributable to several factors. First, since the duration of stay abroad is shorter (3 months vs. 6 months), it is more likely to also include seasonal workers. Second, the time horizon over which migration and return happens is not limited in the Eurobarometer dataset (it is 10 years in the ESS). Consequently, the proportions of returnees in 2010 ranges from around 3 % in Slovenia to more than 10 % in Estonia and Poland if using the ESS data, and from around 6 % in the Czech Republic to around 18 % in Lithuania and 16 % in Slovakia if using the Eurobarometer data (see Fig. 3).

Fig 3
figure 3

Share of return labour migrants in the population aged 24–65, 2008 and 2010 (Source: European Social Survey, rounds 4 and 5; Eurobarometer EB73-3, March–April 2010. Notes: In the ESS data, return migrants are persons born in the country who spent at least 6 months working abroad over the last 10 years and returned. In the Eurobarometer data, return migrants are persons born in the country who have worked (including volunteering and traineeships) for at least three consecutive months in a country other than their country of birth. Estimates are weighted averages)

Full size image

Overall, despite return migration being inherently affected by substantial differences in measurement concepts, the available cross-country data suggest that some heterogeneity exists. On aggregate, there was a slow-down in inflows of the EU8 and EU2 countries’ nationals in the UK and Ireland, although the trend in the UK was reversed again in 2009. Moreover, while immigration from all new member states (apart for Romania) decreased in Ireland, in the UK it increased for a number of countries after 2009, reaching the peaks for Lithuania and Latvia. On the other hand, the outflows from all EU15 countries increased for Bulgaria, Latvia and Lithuania, also partially reflecting the increased inflows and potential onward migration, and slowed down or decreased for the remainder of the new EU members during the crisis. Regarding the stock of return migrants and depending on the definition, the Baltic States as well as Slovakia and Poland report the largest proportions of returnees.

4 Who Is Returning?

This section draws on the data from the EU LFS for 2009 and 2010 to document the profile of a typical returnee who returns during the crisis. The data contain information on the country of residence 1 year before the survey, thus recording individuals who returned to their country over the last year. However, it is important to note that these data are likely to underestimate the true level of return migration, since they only capture most recent returnees who came back over the last year, and thus no longer-run estimates can be calculated. It is also possible to identify current work migrants (or commuters) in these data, i.e. nationals of the new EU members whose current country of place of work is not their country of origin. Finally, a stayer is defined here as a national of the country who resides and works in his/her country of origin (i.e. is neither a migrant nor a recent returnee).Footnote 5

Figure 4 depicts the economic activity of recent returnees to the new member states 1 year before the survey, i.e. while abroad. However, the figure should be interpreted with caution, given that the sample size is small for many countries, which might lead to unreliable estimates. Nevertheless, Fig. 4 shows that the majority of migrants were employed while abroad in all countries, apart from Slovenia, where students constitute the largest group among returnees. The proportion of students who studied abroad a year before and returned is also substantial in Estonia. On the other hand, the shares of returnees who were unemployed while abroad are largest in Bulgaria, Latvia, Romania and Slovenia, while the proportions of those inactive are largest in the Czech Republic, Latvia and Slovakia. Accordingly, these simple tabulations suggest that employment was by far the main activity abroad of recent returnees from all new member states (probably apart from Slovenia).Footnote 6 Regarding the rest of the economic activities abroad, two scenarios become evident: in Estonia and Slovenia, returnees are also those individuals who went abroad to study and return after the completion of this goal, importantly suggesting a potential additional brain-gain for these countries; while in Bulgaria, Romania or Latvia, those who became unemployed also come back.

Fig 4
figure 4

Activity of the recent returnees from the EU15 1 year before the survey, age 15–64 (Source: EU Labour Force Survey 2009–2010). Notes: *stands for less reliable estimates due to the negligible sample size of the returnees from the EU15. A recent returnee is a national of the country who resided in a country other than his/her own 1 year before survey. Estimates are weighted averages

Full size image

In this context, it is vitally important to understand whether those who return are the “best” or “worst” compared to stayers and also migrants, i.e. whether those who come back are “the worst of the best” or vice versa (see, among others, Borjas and Bratsberg 1996). Figure 5 shows such selection in terms of education for migrants in the EU15 countries, returnees from these countries, and stayers in the countries of origin. Bulgaria and Romania stand out as having the highest proportion of individuals with low educational attainment, while both countries have higher shares of such individuals among returnees than stayers. Despite being smaller in magnitude, the shares of the low-skilled are also larger for returnees than stayers in Latvia, Hungary and Lithuania, and these shares are also relatively large among migrants in Latvia and Slovenia. At the other end of the skills distribution, skilled individuals are overrepresented among returnees in the Czech Republic, Hungary, and particularly Slovakia and Slovenia, suggesting a potential positive selection in terms of skills for returnees and a potential brain-gain for these countries. In Latvia, the shares of the highly educated among returnees are almost the same as among stayers, and both are substantially higher than for migrants. In contrast, although the proportions of highly skilled among stayers and migrants in Lithuania are similar, they are larger than for returnees.

Fig 5
figure 5

Distribution of education categories for stayers, migrants in and recent returnees from the EU15, age 25–64 (Source: EU Labour Force Survey 2009–2010). Notes: *stands for less reliable estimates due to the negligible sample size of the returnees from the EU15. A recent returnee is a national of the country who resided in a country other than his/her country 1 year before survey. A migrant is a national of the country whose current country of place of work is not his/her country. Estimates are weighted averages

Full size image

We have undertaken the following modelling strategy to gain further insights into the determinants of return migration. We again use the 2009 and 2010 annual EU LFS data to identify return migrants from the EU15 countries and current migrants who reside in the EU15 countries for at least 1 year as follows. In each of the new member states, we retrieve the returnees from the EU3 (the UK, Ireland and Sweden) and EU12 (the rest of the old EU15) countries, i.e. those individuals who resided in the EU3 or EU12 countries 1 year before the survey.Footnote 7 Subsequently, in each of the EU3 and EU12 countries, we identify nationals from the new member states who are present in the old member states in 2009 and 2010, i.e. current migrants. We merge returnees back to their countries of residence 1 year before the survey to obtain a dataset on current migrants and returnees from the old EU15 countries. Finally, we only retain those migrants who report residing in the country for more than 1 year in our sample, and drop children younger than 15 years old. Thus, our dependent variable is equal to one if an individual returns “home” from the EU15 countries between 2008 and 2009, or 2009 and 2010. Note that we assume that characteristics such as marital status did not change during 1 year in the analysis below, and use information on labor market status 1 year before the survey for both migrants and returnees. Table 1 shows the marginal effects from the probit regressions for the determinants of the return migration decision.

Table 1 Determinants of return migration from the EU15 for the EU8 nationals, marginal effects from Probit
Full size table

This table indicates that males, singles and those with higher or middle education degrees are more likely to return, relative to migrants in the destination countries. The impact of education is also economically significant and implies that having middle educational attainment increases the probability of return by 2.7 percentage points, while the probability increases by 3.5 percentage points for those with higher education (see column 1). In our model, being married is associated with a lower likelihood of returning; however, this might be due to families living abroad together with a migrant, or the larger responsibility that migrants with a family possibly feel with respect to family finances.Footnote 8 Regarding labor market status, although the sample size drops substantially, being inactive in the previous year appears to be positively correlated with the probability of returning to the country of origin, relative to the employed individuals, while no significant relation is found for the unemployed.Footnote 9 Interestingly, there is a lower likelihood of returns from the EU15 countries (both EU3 and EU12) in 2010, and relative to 2009, suggesting a potentially negative impact of the crisis on return migration. Finally, we have also experimented with including macro push-factors that might affect return probability, particularly relative changes in GDP per capita and the unemployment rate between 2008 and 2009 (relative to 2008) in the destination countries. When including both these factors, the increase in unemployment between 2008 and 2009 appears to significantly push EU8 migrants to return from both the EU3 and EU12 countries. In contrast, a higher (lower) growth rate in GDP per capita is associated with lower (higher) return migration from the EU12. However, it should be noted that apart from potential collinearity issues, the time span is possibly still too short to analyze the impact of the crisis on return migration. Moreover, aggregate country-level unemployment rate and GDP per capita might not effectively capture the impact of the crisis on individual return migration decisions, given that local labor markets are probably more appropriate.Footnote 10, Footnote 11

Summarizing this section, our tabulations from the EU LFS data show that returnees were overwhelmingly employed while abroad. Several new EU members are also likely to benefit from positive selection of returnees in terms of skills and potential brain circulation, given that the shares of highly educated among returnees are larger than among both stayers and migrants in the Czech Republic, Hungary, Slovakia and Slovenia. The brain circulation hypothesis is also confirmed by regressions, due to the higher probability of returning for individuals with higher educational attainments. The positive correlation with higher education is consistent with the theoretical implications that returnees, especially those who are skilled, are expected to earn a premium after returning, as well as the observation that recently graduated students constitute an important share of returnees (see also Fig. 4).Footnote 12

On the other hand, this evidence is also in line with the observation that overqualified migrants from the new member states tend to come back. In order to measure overqualified migrants and returnees, we would ideally observe highly educated returnees’ occupations before returning and compare them to the respective migrants, although this is unfortunately not possible with the dataset at hand. Tabulations of the sector of employment 1 year before the survey suggest that most of the returnees with higher educational degrees worked in manufacturing 1 year before the survey. Despite perhaps not being conclusive, this evidence is at least suggestive that overqualified migrants constitute an important share among the returnees, which is also consistent with both the anecdotal evidence and the available evidence from the case studies mentioned in this paper.

This suggests that a migration-induced brain-waste cannot be ruled out for some returnees at some stage. They might have worked abroad in jobs below their educational level, or will do so after their return (while they could have worked more profitably in other parts of the economy); moreover, they might also be unemployed after the return, or their human capital could have depreciated due to the underqualified work in the foreign country. However, even if overqualified for the jobs they performed abroad, these migrants usually chose jobs with larger payoffs than they would have earned working at home according to qualifications, and they could have also worked underqualified or been unemployed before migration. Furthermore, (over-) education might also act as a screening or signaling during the job search process. Accepting such jobs might also be a part of the initial strategy whereby a migrant plans to work temporarily in the host country in order to accumulate wealth, which can subsequently be invested in the country of origin, for example. Moreover, they might still acquire some skills while working temporarily in such jobs, including basic language skills or team-working, which could prove useful when returning to the country of origin. In the long run, relative returns to human capital, including additional skills, might be higher for such migrants at home than abroad. In addition, there could also be a “waste” from the receiving country’s perspective, owing to possible mismatches between migrants’ skills and jobs that they undertake, if, for example, migrants could use their human capital more efficiently in other jobs or sectors and receive higher returns to it. Overall, despite being potentially possible, the phenomenon of brain-waste requires more careful investigation, both with better data allowing for better measurement and involving a longer time period to disentangle potential assimilation effects.

5 Is This the End of the Story?

In today’s globalized world, migrants can be increasingly characterized as “people constantly on the move”. Having experienced migration once, individuals face substantially lower costs for their potential subsequent moves, with improved language or established networks abroad also making such a move easier. On the other hand, the situation in the home country after returning might not be as expected, given that migrants possibly overestimate the potential premiums at home, or the overall situation might be disappointing. In addition, the human capital of individuals who worked in jobs below their qualifications might have depreciated over their stay abroad, thus resulting in a brain-waste. Therefore, it is likely that some returnees would move abroad again.Footnote 13

To investigate the possibility of such repeat and circular migration for the new member states, we employ the Eurobarometer EB73_3 for March–April 2010, which contains the following question regarding future migration intentions: “How likely do you think it is that you will move to another country within the next 10 years, to live there?” It is particularly relevant in the context of our study that it also contains information on past migration experiences, and especially whether an individual had worked, studied or lived for another purpose in a country other than his/her own.Footnote 14

As with any study of migration intentions, it is important to note that intentions might not necessarily correspond to actual realizations (Manski 1990; Bertrand and Mullainathan 2001), leading to an overestimated migration. However, several studies confirm that migration intentions are indeed closely related to the subsequent move (see also Zaiceva and Zimmermann 2008 and the references therein). It is thus advisable to focus on the so-called “firm” migration intentions, actions such as learning a foreign language or applying for visas, although they are not available in our data. In addition, the time horizon in our data is 10 years, which is quite large and might also lead to overestimated migration. To reduce this problem, while presenting tabulations below for the likelihood to move abroad we only focus on answers of “very likely”, which best approximate a “firm” willingness to move.

Figure 6 plots the replies to the question on migration intentions for both stayers and return labor migrants, i.e. those individuals who have worked in a foreign country for at least three consecutive months. Several interesting facts emerge from this figure. First, the proportion of those replying that they are “very likely” to move to another country is consistently larger among returnees than stayers in all new member states, suggesting the circular nature of the potential future East–west migration in Europe. Second, the largest proportion of those willing to emigrate, for both stayers and returnees, are in countries that were affected by the crisis, with nearly 20 % of return labor migrants in Lithuania and 25 % of returnees in Latvia “very likely” to move again within the next 10 years. Third, apart from these two countries, the majority of returnees in the remainder of the EU8 and EU2 countries do not plan to emigrate in the next 10 years (“are not very likely” or “not at all likely”). Regarding the EU2, despite all remaining restrictions on mobility soon coming to an end, the proportions of respondents intending to move abroad are relatively small (around 9 % among returnees in Bulgaria and 5 % in Romania, and the figures are even lower for stayers, at around 2 %).

Fig 6
figure 6

Migration intentions of stayers and returnees in the EU10 (Source: Eurobarometer EB73-3, March–April 2010). Notes: Simple averages. Age: 15–64. Return labour migrants are those who replied “Yes” to the question: “You have worked (including volunteering and traineeships) for at least three consecutive months in another country than yours”

Full size image

For those willing to move abroad, the data also contain a question on a potential country of emigration, distinguishing between another EU country, another non-EU country, USA, Canada, Japan, Australia or New Zealand, or another country outside Europe. In all new member states (apart from Slovenia), the majority of respondents, both returnees and stayers, reply that they are willing to move to another EU country, while the shares of those intending to move to USA, Canada, Japan, Australia or New Zealand are substantial (close to or over 20 %) in Slovenia, Czech Republic, Slovakia and Hungary. However, these results should be interpreted with caution due to the small sample size for individuals intending to move abroad in a number of countries.

Finally, the results from probit regressions that control for individuals’ demographic and socio-economic characteristics, such as gender, age, education, marital status, children, labor market status and country fixed effects, indicate that return labor migrants are significantly more likely to move abroad relative to stayers (not presented here due to space constraints). Moreover, the effect holds after having controlled for networks abroad (having friends or relatives abroad) and language knowledge. There is also some heterogeneity across countries, with the largest marginal effects for returnees, relative to Slovenia, being for Latvia and Lithuania, while the coefficients are not significant for Hungary and Romania (and only marginally significant for Czech Republic, Poland and Bulgaria).

Overall, the results from this section point towards a potential for repeat and circular migration in the post-enlargement Europe. Individuals with work experience abroad are more likely to move abroad again, relative to those without such experience, in the majority of the new EU member states, and most intend to move to another EU country.

6 Conclusions

Return migration is crucial for sending countries, given that it might alleviate the potential negative consequences of the brain-drain and care-drain, and it is becoming increasingly important in post-enlargement Europe. This chapter provides a systematic analysis of its emerging evidence. While the on-going economic crisis might have accelerated the returns of some of these migrants, given that several new member states, particularly the Baltics, were hit relatively more by the crisis than several destination countries, return migration might be delayed.

Summarizing the findings from the literature and our own analysis, we find no evidence of “mass return” (Eurofound 2012). Instead, while there is a slowdown in overall post-enlargement immigration into the two major destination countries, the UK and especially Ireland, immigration to the UK from the two Baltic countries has increased to unprecedented levels since their accession. There is also some evidence of increased outflows from the EU15 countries for several new member states; however, this might reflect some migrants choosing to migrate onwards to other destinations less affected by the recession (such as Norway or Germany). Analyzing return migration is inevitably affected by substantial measurement challenges, and making the cross-country comparisons is even more complicated, due to variations in data collection methods and definitions (stocks vs. flows, age of migrants, time spent abroad, etc.). Despite this major difficulty, depending on definition, the Baltic States and also Slovakia and Poland emerge as countries with the largest proportions of returnees.

Regarding returnees’ characteristics, we find that return migrants were overwhelmingly employed while abroad. In addition, simple tabulations suggest that several new EU member states are likely to benefit from the positive selection of returnees in terms of skills and potential brain circulation, as the shares of highly educated among returnees are larger than among both stayers and migrants in the Czech Republic, Hungary, Slovakia and Slovenia. Furthermore, highly educated returnees also include students with tertiary degrees who return, including those recently graduated, which re-enhances the potential positive brain-gain effects. The positive selection on education is also confirmed by regressions, as the probability to return is larger for individuals with higher educational attainments relative to the lower ones. However, this result might also arise because overqualified individuals, namely those employed in jobs below their qualifications, return disproportionally more. Even if overqualified for the jobs they perform abroad, these migrants usually chose jobs with larger payoffs than what would have they earned at home working according to qualifications, and are also expected to have higher returns to skills at home in the long run. Thus, although potentially possible, the phenomenon of brain-waste requires more careful investigation, both with better data and over a longer time period.

Finally, regression results also suggest that males and singles are more likely to return, ceteris paribus, relative to migrants in the destination countries, and that the likelihood of return is lower in 2010 relative to 2009, which might reflect an impact of the ongoing economic downturn.

However, returning home at times of trouble may not be the end of the story. We find that individuals with work experience abroad are more likely to move abroad again, relative to those without such experience, in the majority of the new EU member states, and most intend to move to another EU country. This result points towards an important potential repeat and circular migration in the post-enlargement Europe.

Overall, enjoying free mobility (albeit still with some restrictions for the EU2 workers), post-enlargement migrants constitute a highly flexible and valuable resource that may act as a “buffer” during the economic crisis. Potential repeat and circular migration is expected to alleviate the negative impacts of the crisis, leading to brain circulation rather than brain-drain and a more efficient allocation of resources within the EU. Most importantly, remaining transitional arrangements for the EU2 migrants may actually backfire, since these migrants may chose to remain in the host country after becoming unemployed, towing to uncertainty about the possibility of re-entering the destination country in the future. Altogether and bearing in mind potential brain circulation in the enlarged Europe, this suggests important implications for migration policy.