Migration and development at home: Bitter or sweet return? Evidence from Poland

Abstract

The existing economic literature focuses on the benefits that return migrants offer to their home country in terms of entrepreneurship and human and financial capital accumulation. However, return migration can have modest or even some detrimental effects if the migration experience was unsuccessful and/or if the migrant fails to re-integrate into the home country’s economy. In our paper, we empirically show which factors – both individual characteristics and features related to the migration experience – influence the likelihood of a sub-optimal employment of returnees’ human capital employing an original dataset on a representative sample of return migrants in Silesia (Poland).

Keywords

Economic performance international migration regional development return migration

Introduction

Return migration is one of the most important channels through which sending countries might benefit from emigration, and return migrants are widely recognized as important agents of change, inducing modernization processes of their home countries (King, 1978). The returnees can be an important source of financial (remittances sent and savings brought upon return), social (transnational networks developed and maintained with the former destination countries, transfer of social norms, democratic values) and human capital (additional education, skills and professional experience acquired during the migration period; see Thomas-Hope, 1999). Yet, return migration does not automatically translate into the development of the source country or region, as there are complex migrant–home development interactions (De Haas, 2010). Moreover, the returnees do not constitute a homogenous group but return migration – as for initial emigration – is characterized by self-selection.¹ Looking at employment patterns of the returnees might be misleading when it comes to assessing their pro-development potential. Many returnees start businesses in their home country using the human and financial capital gained overseas (Mayer et al., 2015). However, McCormick and Wahba (2001) suggest that some unskilled return migrants become self-employed because of the lack of other alternative employment possibilities at home, and therefore their job-creating and development potential for the source countries can be rather small (Piracha and Vadean, 2010). The analyses of special programmes encouraging the return of diaspora members show that such policies can encourage more productive uses of remittances, but often fail to exploit the skills of the returnees, which is often caused by the shallowness of the labour markets in the home economy (Diatta and Mbow, 1999).

The main aim of this paper is to analyse the factors that prevent successful return migration to the home country (we label the occurrence of this event as a ‘bitter return’). The present work shows that a full understanding of the re-integration processes of the returnees in the home country should duly take into consideration the entire migration experience of the individuals (and not only their characteristics).

We firstly discuss the evidence on return migration and on the migrants’ performance upon return. Secondly, we carry out an empirical analysis of the determinants of unsuccessful return migration using an original and rich dataset on the migrants returning from several European countries to the Silesia region in Southern Poland. We explore three different dimensions: (i) the probability of employing the skills and qualification acquired during the migration spell; (ii) economic inactivity upon return; (iii) the probability of experiencing difficulties and hardships upon return (i.e. a ‘bitter return’). Given the current migration history of Poland, the most important source country within the European Union (EU), we believe that this case study has significant general implications.

In this study, a return migrant is a person who had an international migration experience for a period of at least 3 months and at the moment of the survey was present in a home country.² We find that younger and more qualified migrants are less likely to experience an unsuccessful return, in particular if the initial outmigration was motivated by educational and training purposes. A longer migration spell reduces the likelihood of a skill waste upon return but, in general, is associated with more intense difficulties and hardships in re-entering the home country society and labour market. We find strong evidence of a ‘repeated skill waste’: individuals experiencing a skill mismatch during the migration experience, ceteris paribus and controlling for the endogeneity of this outcome, are more likely to be economically inactive and less likely to use the newly acquired human capital once back in the home country. Finally, we find evidence that those individuals who managed to accumulate savings during the migration spell are more likely to be economically inactive and experience a skill waste upon return. The latter result seems to suggest that capital and skills accumulated abroad during the migration spell are not necessarily used in a complementary way in the home country.

The recent global financial crisis has boosted the academic and policy interest in the nexus between return migration and development. The deterioration of economic opportunities in several migrant destination countries after 2007 not only has contributed to a decrease in migration intensity, but also has spurred a large return migration (Tilly, 2011). These new developments are particularly intensive in the case of the Central and Eastern European (henceforth CEE) countries, which benefit from the principle of free mobility of people within the EU. Martin and Radu (2012) suggested that in the case of the recent East–West European migration, it was reasonable to expect a return ratio in the range of 20–50 per cent of the initial flows, which means at least half a million migrants returning to CEE countries in the period 2004–2009. The actual magnitude of return migration is likely to be even bigger than expected by these studies. For instance, the data from Polish Labour Force Survey (BAEL) shows that between January 2004 and June 2008, 580,000 persons returned from international migration, with 213,000 alone in 2007 (Anacka and Fihel, 2012), although for many individuals such homecoming was not permanent, and could be followed by subsequent new migration (White, 2014b). Therefore, many returnees are described as ‘intentionally unpredictable’, which means ‘that at some point they will have the need to migrate somewhere’ (Anacka et al., 2013: 302). The Polish case is notable, as Poland was one of the few EU Member States that remained relatively unharmed by the economic crisis, retaining a positive economic growth in the period 2007–2009. So, while a vast number of Poles remained abroad in countries harmed by the crisis (Krings et al., 2009) many others returned to Poland, due to personal motivations, achieved economic goals (including accumulated savings for car or apartment purchase) or lured by the relatively favourable economic situation at home (Brzozowski, 2011; Fihel et al., 2008).

In spite of the intensive return waves of migrants – in particular from Central and Eastern Europe – there exists limited evidence on the economic implications of such labour movement back to source countries (see De Coulon and Piracha, 2005; Iara, 2006; Fihel and Grabowska-Lusińska, 2014; Grabowska, 2016; Martin and Radu, 2012). Moreover, the knowledge on whether and under which circumstances the returnees are able to productively exploit the skills acquired during the migration spell is generally scarce in the migration studies literature. Our goal is to contribute in filling the existing gap in the literature.

The structure of our paper is as follows. In the second section we briefly review the literature on return migration and, in particular, on the economic contribution of returnees upon return. In the third section we describe the dataset and the empirical methodology used. The results of the empirical analysis are presented in the fourth section. The fifth section reports concluding remarks.

Bitter return? A conceptual discussion

When is return migration an optimal strategy from the point of view of the migrant? Do migrants contribute to their home country’s economy when they return? The economic research on return migration can be broadly divided into two main strands surrounding these two questions. The first strand of literature has focused on the economic determinants of return migration, and in particular on the role of push and pull factors, respectively in the host and home locations, which might influence the homecoming decision. In the standard neoclassical theory, return migration is generally seen as the result of the failure of the migration project due to the fact that the migrant has not succeeded in integrating economically and/or socially in the host country (De Haas et al., 2014) or, in the case of irregular migration, if the migration spell is forcefully interrupted by the host country’s authorities. In the latter case, the migration spell is sub-optimal if compared with the pre-migration expectations.

Using the ‘lenses’ of the new economics of labour migration, the unit of analysis is not the single individual but the household. The migration project is a collective venture of the migrant’s household, which should lead – via remittances and capital accumulated abroad – to the betterment of the living conditions of the migrant’s entire family residing in the home country, including the migrant. Therefore, in this framework the successful migration project is usually temporary, as the migrant intends to return in order to employ the resources accumulated during the migration spell in a productive activity for the benefit of his/her household (Dustmann, 1997; Stark et al., 1997). In this framework, the return decision might be postponed when the migrant is not successful in achieving the goal set for the migration project (De Haas et al., 2014). Several studies that have tested those contrasting theses find mixed results (Bijwaard et al., 2014; Bijwaard and Wahba 2014; De Haas et al., 2014; Fokkema and De Haas, 2015; Jensen and Pedersen, 2007). As argued by De Haas et al. (2014), there exist several theoretical explanations for return migration and the approaches highlighted above should be treated as complementary, not rival concepts to describe the overall process that leads to the patterns observed in migration flows (De Haas et al., 2014).

The second strand of the existing literature deals with the returnees’ economic performance in the home country. These studies generally focus on the occupational choice of migrants returning to the home country (wage employment, self-employment and entrepreneurial activities, but also unemployment and non-participation in the labour market) and/or their personal characteristics (whether they are positively or negatively self-selected compared to migrants and non-migrants). The theoretical explanation for these studies is provided mostly by the structural approach in migration studies. The structuralists consider how the returnees’ expectations are influenced by the socio-institutional factors in their home countries (De Haas et al., 2014).

The seminal work of Cerase (1974) defines a useful categorization of return migration: ‘return of failure’, ‘return of conservatism’, ‘return of retirement’ and ‘return of innovation’. Only the last type, the return of innovation, opens the possibility for migrants to become agents of change and helpers of the modernization process in their source country. Still, the author is rather pessimistic about the effective ‘innovative’ impact of the returnees, as they become constrained by the local power structures, which are resilient to socio-economic changes (Cerase, 1974). King is even more unenthusiastic in this regard: by providing a set of unsuccessful cases of return migration, he argues that ‘return of innovation is a myth’ (King, 1978: 178). Moreover, as Gmelch (1980) adds, the returnees tend not to be well-prepared for return, usually making a very naive and optimistic assumption about the socio-economic situation at home, which in turn explains their poor performance upon homecoming. This preparation for return is an issue that has been theoretically developed by Cassarino (2004). He distinguishes two main dimensions of preparation for an individual to return to his/her home country, namely resource mobilization (i.e. intangible and tangible resources and social capital accumulated before leaving the home country and during the migration spell) and preparedness (i.e. the willingness and readiness to return). Consequently, ‘returnees differ in terms of levels of preparedness and patterns of resource mobilisation’ (Cassarino, 2004: 272) and only those individuals with a high level of preparedness are able to reap new economic opportunities upon return and constitute a real gain for the home country. On the other hand, returnees with a low level of preparedness are bound to experience difficulties in re-integration at home and might decide to re-emigrate.

The theoretical concepts on the performance of returnees have been also developed in a Polish context. The first strand of research was devoted to return migration in the transition period (1989–2003), when the country was experiencing transformation from a centrally planned system to capitalism and a free-market economy (Fihel and Górny, 2013; Iglicka, 2002). The second wave of research includes the post-accession migration – which was started in 2004, when Poland joined the EU. Within this second strand, there are studies that need closer attention in the context of our analysis. A recent study by Krystyna Iglicka (2010) suggests that, in particular, young tertiary-educated migrants who left the country with very little or no working experience are threatened by the brain waste/over-qualification problem and subsequent deskilling of their human capital assets. Therefore, for those individuals it is generally difficult to return and re-integrate successfully in the Polish labour market. According to the author, they are threatened by a possible ‘migration trap’ – forced to work below formal qualification and constantly on the move between Poland and Western European countries, remaining at the socio-economic margins of both economic systems. This hypothesis is supplemented by the work of Anne White, who proposes the term, ‘double return migration’ (White, 2014b): many post-accession migrants actually try to go back to Poland, but their return is not successful in economic terms. Therefore, they seek to return to previous destinations, such as the UK, as they want to gain some economic stability. In such cases the return to Poland is only temporary, while the double return to Western European destinations might have a more permanent character (White, 2014a). On the other hand, Grabowska-Lusińska and Okólski formulate a more positive ‘crowding-out’ hypothesis; according to them, the post-accession migration might play a positive role: migrants leave underdeveloped regions, settle temporarily abroad, but might return with new skills and experience (Grabowska-Lusińska and Okólski, 2009). This kind of movement – emigration from rural and peripheral areas and return to industrial and developed regions in the home country – is described by King et al. (2008) as ‘J-turn’ migration, which offers better perspectives for economic integration of a migrant due to greater likelihood of employment. Consequently, this return should take place in a specific framework: returnees should find new destinations in Poland that offer better working possibilities, such as the fast-developing metropolises of Warsaw, Poznań, Wrocław or Kraków, instead of homecoming to the poor neighbourhoods of former residence.

These hypotheses have been tested in a number of empirical studies. Kaczmarczyk and Tyrowicz provide some empirical support for Iglicka’s (2010) ‘migration trap’ hypothesis, as they discover that tertiary-educated immigrants from Poland in the UK are relatively deprived in terms of remuneration when they are compared with counterparts who remained in Poland. Moreover, those individuals are increasingly employed in unskilled positions and likely to experience a brain waste (Kaczmarczyk and Tyrowicz, 2015). Consequently, they are endangered by the deskilling process and gradual depreciation of their human capital assets, which puts at question their successful return to the home country. Diatta and Mbow (1999: 254) focus on the Senegalese diaspora, claiming that the skills learned by migrants ‘have been put to little use’ upon return, constraining them to salaried employment. Co et al. (2000) find that the migration experience has a positive impact on earnings of the returnees in Hungary, but this effect is only significant for women migrants returning from Organization for Economic Cooperation and Development (OECD) countries. De Vreyer et al. (2010) find that return migrants from OECD countries get a substantial wage premium upon return to West African countries, but the returnees – probably due to lost social capital as compared to non-migrants – show smaller participation rates in the formal wage sector. These findings suggest that the characteristics of both origin and destination countries matter in explaining (un)successful return migration.³

De Coulon and Piracha (2005) find that returnees in Albania are negatively self-selected and typically underperform the non-migrant population in the home country labour market. However, they also observe a premium stemming from the migration experience on the labour market, suggesting that returnees have acquired some skills that make them more productive in the home country labour market. Iara (2006) finds that young male returnees from Western Europe receive a substantial wage premium in the labour markets in Central and Eastern Europe. However, her analysis shows the existence of a positive self-selection of both out-migrants and returnees. A positive self-selection in return migration is also found in the study of Martin and Radu (2012) for Poland, Hungary, the Czech Republic and Slovakia. These authors find evidence of an earning premium associated with the migration experience. However, their study points to the fact that returnees – due to weak social networks and/or to lack of experience related to the local labour market – often choose self-employment or remain economically inactive upon return. Heterogenous results have been found in studies carried by Centrum Doradztwa Strategicznego (Centre for Strategic Advice – CDS) at the regional level in Poland: while Silesia (Szymańska et al., 2012) and Lower Silesia (Bieńkowska et al., 2010b) regions were mostly able to exploit the economic potential of returnees, the returnees in Lesser Poland (Bieńkowska et al., 2010a) were experiencing more difficulties in economic re-integration and many of them declared a will to re-emigrate again, confirming the predictions of double return migration hypothesis proposed by White (2014b).

Most of the studies on the performance of return migrants do not fully investigate how the characteristics of the migration experience of the returnees account for the successful economic re-integration in the home country. In this regard, few notable exceptions need closer attention. Williams and Balaz (2005) study the process of human capital acquisition during the migration spell of three distinct groups of returnees in Slovakia: professionals and managers, students and au pairs. The authors argue that less than half of their respondents make use of the skills and competences acquired abroad. However, the process of human capital acquisition is much more nuanced and heterogeneous than expected, and even migrants exposed to semi-routine and unskilled jobs like au pairs were able to learn valuable competences in the UK. Therefore, they introduce a more comprehensive concept of human capital that also includes ‘soft’ competences, such as self-awareness or self-confidence, which were found crucial for the subsequent economic success of returnees in Slovakia. The importance of ‘soft’ (social) competences in the successful economic re-integration of returnees was also stressed in the more recent contribution of Grabowska and Jaźwińska (2015), who have analysed the professional histories of return migrants in Poland. However, the productive use of this form of human capital is dependent on opportunity structures existent at home, as those ‘who go to work in an international environment, are able to make a fuller and a better use of their skills acquired during migration’. Moreover, a returnee usually needs some time before these skills can be properly activated; thus, this process might be ‘postponed in time’ (Grabowska, 2016: 192).

In a related study, Lianos and Pseiridis (2009) investigate the performance of return migrants in Bosnia, Bulgaria, Georgia, Kyrgyz Republic, Romania and Tajikistan, taking into account the acquisition of qualifications and skills during the migration spell. In their study, the migrants who worked in the host country in the same sector as before migration were more likely to obtain valuable on-the-job skills; on the contrary, a job-mismatch was generally less conducive to human capital acquisition. Interestingly, the authors find that migrants whose economic situation before migration was unfavourable were also more likely to acquire on-the-job skills compared to migrants with a university education. Moreover, the skills learned during the migration spell result in a substantial (on average by 16 per cent) increase in earnings upon return. These two studies highlight the important (and heterogeneous) role played by human capital acquisition during the migrants’ stay abroad on their performance upon return.

The two strands of the literature briefly outlined above are generally detached from each other. Most studies on the economic performance and contribution of return migrants do not explicitly take into account the characteristics of the migration spell. The working experience of the migrant in the host country, the learned skills and acquired human capital, but also one’s failures, traumas and mistakes are seldom included when accounting for the variation of economic performance at home upon return. A ‘salt water curtain’⁴ between these two fundamental phases of the return migration experience seems to be present in the current literature. In what follows we empirically analyse how the experiences acquired during the migration spell influence the probability of succeeding once the migrant returns to the home region.

Understanding the roots of bitter return: Data description and methodology

Data description

The dataset for our study is based on a household-level survey conducted by CDS (a private research institution) in Silesia voivodeship, an administrative region in southern Poland. The CDS study was aimed at identifying the magnitude of out-migration and return migration flows at the regional level, and the socio-economic impact of these processes on the analysed region (Szymańska et al., 2012). In addition, this dataset contains a set of information on return migrants’ skills and their actual use upon return in the regional labour market, an issue that so far has been largely unexplored in empirical analysis.

Silesia is a region with a rich history of migration, especially to the USA and Germany. However, Silesia is also a traditional industrial region, where the coal mines, chemical and automobile industries still remain an important part of the local economy. In addition the region, as compared to national Polish economy, is relatively well-developed, with average monthly earnings above the national average (103.7 per cent in 2013; see CSO, 2015). This is an added value of our approach, as most of the studies on return migration focused on developing and emerging/transforming economies. Consequently, the choice of this region for our analysis is interesting, as we investigate the effects of return migration to a high-income and fast-growing economy, which has a relatively high potential to exploit the skills and competences of the returnees.

The household survey was carried out in April 2011 in two stages (Szymańska et al., 2012). In the first one, the aim was to identify the magnitude of migration and return migration at the regional level. In the first stage, an initial random sample of 26,000 households was selected based on the landline telephone list as a sampling frame. Using the CATI (Computer-assisted Telephone Interviewing) technique, interviews with respondents were carried out, yielding 17,772 responses (response rate 68.4 per cent), which accounts for circa 1 per cent of the Silesia region’s households. Out of this initial sample, all the households reporting at least one return migrant were identified (1227 observations). The second stage of investigation consisted of structured interviews based on the questionnaire.⁵ Out of 1227 households, 528 complete questionnaires were collected (a response rate of 43 per cent).⁶ One possible shortcoming related to the data collection method is the fact that our sample underestimates the magnitude of international migrants, as households entirely composed of migrants are not included. However, for the purpose of the present study this is not a major problem, as our focus is on returnees only. Consequently, we can argue that our sample is fully representative for the returnee population in the region.

In Table 1 we report the main characteristics of return migrants to the Polish region under analysis. We distinguish between migrants who face significant difficulties and hardships upon return to Poland (labelled as bitter return,⁷ column 2) and those who do not experience such difficulties (labelled as sweet return, column 3).

Table 1.

Return migrants in the Silesia region, Poland: some descriptive statistics.

Variable name	(1) All returnees	(2) Bitter return	(3) Sweet return	Mean difference
Age (years)	35.2	34.1	36.2	−2.12 ***
Female (%)	40.3	39.0	42.4	−2.64
Married (%)	45.5	43.2	47.4	−4.43
Migrant has at least 1 child (%)	37.1	34.9	39.0	−4.78
Vocational education (%)	44.7	49.4	40.8	8.53 ***
Secondary education (%)	13.8	13.8	13.7	0.01
Tertiary education (%)	38.8	35.6	41.9	−6.3 **
Years passed upon return	2.6	2.35	2.77	−0.42 ***
Destination country (%):
United Kingdom	31.3	31.4	30.6	0.75
Germany	17.2	13.4	20.4	−7.0 ***
The Netherlands	12.8	15.1	10.9	4.1 **
Ireland	9.2	10.9	7.7	3.2 **
Italy	5.0	7.1	3.2	3.9 ***
USA	4.0	2.9	4.9	2.0
No.	528	241	287

Source: authors’ calculations.

***

and ** represent statistical significances of mean differences at respectively 5 and 10 per cent.

The returnees in our sample are – on average – rather young, mostly male and with a rather large share of those who have completed vocational education. The most popular country of destination was the UK, followed by Germany. The statistics are basically reflecting the general trend in post-accession migration from Poland after 2004 and subsequent return waves (Anacka and Fihel, 2012). There exist some substantial differences between migrants experiencing sweet and bitter return, as reported in the last column of Table 1. Individuals that are more successful upon return tend to be slightly older, more likely to have children and tertiary rather than vocational education. Individuals who have returned in more recent years – most likely ‘pushed’ by the financial crisis – are slightly more likely to experience difficulties upon return. Finally, the country where individuals have migrated seems to matter: bitter returnees are less likely to be among those who returned from Germany and more likely to be among those returning from The Netherlands, Ireland and Italy. Clearly, the country of migration might matter both in terms of the initial self-selection process into alternative destinations and in terms of the heterogeneous possibility to accumulate experiences and human capital during the migration spell. For instance, for Silesia, Germany represents a more traditional country of destination with presumably a denser diaspora and, in turn, reduced costs of economic and social integration. On the contrary, migration in Italy is often characterized by rather low socio-economic integration and often the migration spell is conducted irregularly.⁸

Empirical methodology: Probit and seemingly unrelated bivariate probit models

What determines an unsuccessful return, both from the point of view of the returnee and of the home country society? In what follows we analyse the individual- and household-level factors that are associated with the probability of experiencing a sub-optimal use of the skills and human capital accumulated during the migration spell.

In particular, we explore four dimensions of unsuccessful return, employing the following alternative dependent variables: (i) skill waste upon return, a dummy variable equal to 1 if the returnee is not currently able to use the experiences and skills learned abroad; (ii) economic inactivity, a dummy variable equal to 1 if the returnee is not currently economically active/employed; (iii) bitter return, a dummy variable equal to 1 if the returnee experiences significant hardships/difficulties upon return in the home country socio-economic context; (iv) job-related difficulties, a dummy variable equal to 1 if the returnee experiences significant difficulties in the home country labour market. More details concerning the construction of these variables are reported in the appendix.

Our estimation strategy is based on the following baseline equation:

\begin{array}{l} Prob (U n s u c c e s s f u l_R e t u r n) = α + β * X_{i} + γ * \\ S k i l l_W a s t e_H o s t C o u n t r y_{i} + ε_{i} \end{array}

(1)

where the probability of an unsuccessful return (using the above-mentioned dependent variables) depends on a vector $X_{i}$ of individual (age, sex, type of educational background, employment status before migration), household (children, economic situation before migration, relative importance of remittances for household consumption) and migration experience characteristics (study-related migration, migration spell, year of return in Poland, irregular migration, use of social assistance during the migration spell, country of migration, skills acquired during migration). Table 2 reports the descriptions and summary statistics of all the dependent variables and covariates used in this study.

Table 2.

Summary statistics and variable descriptions.

Variables	Description	Mean (st. dev)
Dependent variables:
Skill waste (home)	(Dummy variable) Skill waste upon return: individual is not able to use the experience and skills learned abroad in current occupation at home	1.4449 (1.8979)
Economically inactive (home)	(Dummy variable) Individual is currently economically inactive (retirees and pensioners excluded)	0.2184 (0.4136)
Bitter return	(Dummy variable) Some hardship/difficulties upon return in the home country reported (no access to information, no support to find a job, no psychological support, other hardships)	0.4570 (0.4986)
Job-related difficulties	(Dummy variable) Migrant reported to have lost orientation about working conditions at home and/or finding a satisfactory occupation upon return	0.4646 (0.4992)
Individual- and household-level variables
Age	Age of the migrant at the time of the interview (min 20, max 80)	35.26 (12.0974)
Female	(Dummy variable) Equals 1 when the individual is female	0.4034 (0.4911)
Children (dummy)	(Dummy variable) Equals 1 when the individual has children	0.3690 (0.4830)
Vocational education	(Dummy variable) Equals 1 when the highest level of education is obtained (a vocational/professional degree)	0.4474 (0.4977)
Tertiary education	(Dummy variable) Equals 1 when the highest level of education is obtained tertiary level degree (e.g. university degree or master’s)	0.3901 (0.4882)
Household economic situation before migration	Economic situation of migrant’s household before migration on 1–5 scale, where 1 equals very bad, and 5 very good economic situation	3.3024 (0.9884)
Household living on remittances	(Dummy variable) Remittances sent by migrant constituted important or main source of income for migrant’s household during migration spell	0.1396 (0.3469)
Employed before migration	(Dummy variable) Migrant has been employed at home before migration	0.4245 (0.4947)
Main reason to migrate: study	(Dummy variable) Migrant indicated educational aim as the main motivation of migration	0.1606 (0.3675)
Year of return	Ordered year of return, where 1 accounts for 2004 and 8 for 2011	5.4261 (1.7347)
Social assistance in the host country	(Dummy variable) Equals 1 when the individual has benefited from welfare assistance during the migration spell, such as unemployment benefits, poverty-alleviation measures, etc.	0.1456 (0.3531)
Migration spell (in months)	Duration of the migration spell measured in months (min 3, max 240)	24.3182 (32.9589)
Skill mismatch during migration	(Dummy variable) Work abroad below one’s formal qualifications reported	0.3690 (0.4830)
Irregular migration	(Dummy variable) Illegal/undeclared work abroad reported	0.0688 (0.2534)
Accumulated savings	(Dummy variable) Accumulated savings reported as benefit of migration	0.4990 (0.5005)
Learned a language	(Dummy variable) Learned foreign language reported as benefit of migration	0.6979 (0.4596)
Acquired soft skills	(Dummy variable) Acquired soft skills reported as benefit of migration	0.5354 (0.4992)
Acquired technological knowhow	(Dummy variable) Acquired technological knowhow reported as benefit of migration	0.3843 (0.4869)
Germany	(Dummy variable) The migrant has returned from Germany	0.1721 (0.3778)
United Kingdom	(Dummy variable) The migrant has returned from the United Kingdom	0.3098 (0.4628)
Ireland	(Dummy variable) The migrant has returned from Ireland	0.09178 (0.2890)
Scandinavia	(Dummy variable) The migrant has returned from a Scandinavian country	0.0535 (0.2253)
Italy	(Dummy variable) The migrant has returned from Italy	0.0497 (0.2176)

Our model includes the variable Skill mismatch in the host country, a dummy variable equal to 1 if the return migrant was overqualified for the type of occupation in which he/she was employed abroad. Our main hypothesis is that the sub-optimal employment of skills and human capital during the migration spell might translate into a persistent sub-optimal use of human capital also at the end of it. In fact, if a migrant is not fully using his/her skills and qualifications in his/her employment abroad he/she might be induced to acquire little additional human capital and, most likely, of the type that is not necessarily of high potential upon return to the home country. Clearly, a skill mismatch during migration might be endogenous with respect to our dependent variable, since some characteristics of the migrant might explain the skill waste experienced both during and after migration. As discussed by Baum et al. (2012), the estimation of binary models with the endogenous binary regressor is not straightforward. Following Baum et al. (2012) and Angrist and Pischke (2009), we estimate a bivariate probit model as follows:

\begin{array}{l} Prob (U n s u c c e s s f u l_R e t u r n) = α + β * X_{i} + γ * \\ S k i l l_W a s t e_H o s t C o u n t r y_{i} + ε 1_{i} \end{array}

(2)

\begin{array}{l} Prob (S k i l l_W a s t e_H o s t C o u n t r y_{i}) = α + \\ θ * Z_{i} + ε 2_{i} \end{array}

(3)

(ε 1, ε 2, X, Z) N (0, 0, 1, 1, ρ)

(4)

where the two probit models (2), (3) are simultaneously estimated through maximum likelihood estimation and the model permits to estimate the parameter $γ,$ that is, the effect of the skill waste during migration on the subsequent unsuccessful return, allowing for the correlation of the error terms $(ρ) .$

This empirical approach allows us to investigate in a rigorous way the different factors that might hamper the ability of a return migrant to fully and profitably re-integrate in the home country economy and society.

Empirical results

In order to investigate the unsuccessful or successful return in the home country, one should consider several dimensions that are not necessarily related to each other. In the current paper we consider those aspects that are likely to be strongly related to the human capital potential embedded in migration flows. We first consider the factors that prevent individuals effectively using the newly acquired skills. Then we consider a rather objective measure of skill use, the status of economic inactivity. Thirdly, we consider the determinants of a subjective measure of difficulties and hardships upon return, that is, bitter return. Finally, in the last part of our analysis we focus on difficulties that are specific to the re-entry (or new entry) in the Polish labour market.

Starting from the determinants of a skill waste upon return, we report in Table 3 the marginal effects of both probit and bivariate probit models. The probability of experiencing a skill waste upon return is between 10.7 and 12.6 per cent higher when the migrant was overqualified during the migration spell when we do not control for potential endogeneity (columns 1–3, probit models). The likelihood of skill waste goes up to 38.1 per cent when accounting for endogeneity (column 1a).⁹ Younger returnees are less likely to experience a skill waste upon return; moving from mean age (approximately 35 years old) to one standard deviation above it (circa 47 years old) increases the probability of not using the acquired skills and experiences by 6–7 per cent. We also find mild evidence of a higher probability of skill waste upon return for female migrants.

Table 3.

The probability of skill waste upon return migration (marginal effects).

	(1)	(1a)	(2)	(2a)	(3)	(3a)
Dependent variable: skill waste upon return	Probit	Bivariate probit (*)	Probit	Bivariate probit (*)	Probit	Bivariate Probit (*)
Skill mismatch during migration	0.126**	0.381***	0.120**	0.317**	0.107**	0.357***
Skill mismatch during migration	(0.0506)	(0.099)	(0.0509)	(0.149)	(0.0526)	(0.115)
Age	0.0066***	0.0052***	0.00602**	0.0052***	0.00545**	0.0044**
	(0.00230)	(0.0019)	(0.00235)	(0.0019)	(0.00247)	(0.0018)
Female	0.107*	0.077*	0.0957*	0.078*	0.0904	0.068
	(0.0554)	(0.0445)	(0.0561)	(0.047)	(0.0587)	(0.044)
Vocational education	−0.00534	0.028	−0.0102	0.017	−0.0386	0.0040
	(0.0752)	(0.061)	(0.0756)	(0.066)	(0.0776)	(0.063)
Tertiary education	−0.136*	−0.129**	−0.133*	−0.129**	−0.152*	−0.136**
	(0.0782)	(0.061)	(0.0785)	(0.064)	(0.0811)	(0.063)
Household economic situation before migration	−0.0486*	−0.032	−0.0414	−0.032	−0.0454*	−0.030
Household economic situation before migration	(0.0265)	(0.0223)	(0.0268)	(0.0232)	(0.0275)	(0.022)
Employed before migration	−0.119**	−0.078*	−0.113**	−0.085*	−0.105*	−0.067
	(0.0530)	(0.046)	(0.0531)	(0.048)	(0.0541)	(0.045)
Main reason to migrate: study	−0.224**	−0.115	−0.215*	−0.135	−0.166	−0.0685
	(0.114)	(0.101)	(0.115)	(0.108)	(0.121)	(0.099)
Year of return	−0.00994	−0.0057	−0.00736	−0.0052	0.000346	−0.00097
	(0.0152)	(0.0113)	(0.0153)	(0.0124)	(0.0157)	(0.0113)
Social assistance in the host country	0.0952	0.098*	0.106	0.105*	0.127*	0.116**
Social assistance in the host country	(0.0668)	(0.056)	(0.0661)	(0.059)	(0.0663)	(0.057)
Migration spell (in months)	−0.0022***	−0.0014**	−0.0022***	−0.0017**	−0.0021***	−0.0014**
	(0.00071)	(0.00065)	(0.0007)	(0.00069)	(0.00071)	(0.00064)
Irregular migration			0.175*	0.113	0.155	0.069
			(0.0941)	(0.113)	(0.0982)	(0.101)
Accumulated savings					0.114**	0.0888**
					(0.0525)	(0.039)
Learned a language					−0.0890	−0.071
					(0.0595)	(0.0448)
Acquired soft skills					−0.0819	−0.0631*
					(0.0531)	(0.0392)
Acquired technological knowhow					−0.159***(0.0555)	−0.110**(0.046)
Rho		−0.592*(0.275)		−0.434 (0.356)		−0.592 (0.310)
LR test Rho = 0 (chi²(1))		2.839*		1.627		2.139
Pseudo R²	0.089		0.094		0.129
Observations	431		431		431

Note: Marginal effects from probit estimates are reported; for dummy variables marginal effects are calculated as discrete change from 0 to 1. Standard errors in parentheses.

***

p < 0.01, **p < 0.05,* p < 0.1.

The second equation in the bivariate model is the following: skill mismatch during migration (dependent variable) = Germany, UK, Ireland, Scandinavian country, Italy, vocational education, tertiary education, economic situation before migration, employed before migration, main reason to migrate is study, social assistance in the host country, migration spell, irregular migration.

Individuals with tertiary education are between 13 and 15 per cent less likely to experience a skill waste once they return home. A similar effect is found for those who were working before migrating and for migration related to study and training reasons. The latter result suggests the importance of cross-boundary migration for study and training reasons for the home country socio-economic development. This type of return migration is well identified by the above-mentioned Cerase’s category of ‘return of innovation’ (Cerase, 1974).

A longer migration spell is, ceteris paribus, associated with a lower probability of not using the skills accumulated while abroad: an increase of the migration duration of one standard deviation above the mean decreases the probability of skill waste upon return of circa 7 per cent. We control for the year of return and find a negative but not significant effect on the occurrence of skill waste.

Our results show that the type of migration experience significantly affects the ability to use skills and competence acquired abroad. In column 2 we introduce a dummy that is equal to 1 when the status of the migrant was not regular and find some support for an increase in the probability of skill waste. In fact, irregular migrants are severely constrained in terms of ability to learn and accumulate human capital and by the type of skills and experience available to them. Our probit estimates suggest that illegal migrants returning home are 17.5 per cent more likely to be unable to use the experiences and skills acquired during the migration spell.

Finally, it is interesting to note that return migrants who have accumulated savings abroad are less likely to put their skills to use upon return. This result might be explained by a higher reserve wage of those returnees who have accumulated sufficient savings, which translates into a lower participation rate in the home country labour market. On the contrary, when technological knowhow is gained abroad there is a significant reduction of the probability of skill waste when returning to the home country.

In the next step of our analysis, we analyse the probability of the returnees being currently economically inactive. From the point of view of the home country society this is probably the worst case scenario, as it severely limits the potential socio-economic benefits of return migration. Marginal effects obtained by estimating a probit model as in Equation (1) above are reported in Table 4.¹⁰ As expected, older immigrants are more likely to be economically inactive. At mean value using the model in column 1, a 30-year-old returnee has the probability of being currently inactive of 21.3 per cent compared to 33.3 per cent for a 50 year old. Female return migrants appear to be more likely to be economically inactive when returning home.

Table 4.

Economic inactivity upon return in the home country: marginal effects (probit model).

Dependent variable: currently economically inactive	(1)	(2)	(3)	(4)
Skill mismatch during migration			0.0862*	0.0970**
Skill mismatch during migration			(0.0459)	(0.0466)
Age	0.00594***	0.00583***	0.00650***	0.00615***
	(0.00207)	(0.00204)	(0.00204)	(0.00206)
Female	0.113**	0.112**	0.0758	0.0903*
	(0.0479)	(0.0487)	(0.0486)	(0.0501)
Vocational education	−0.0465	−0.0597	−0.0655	−0.0735
	(0.0564)	(0.0571)	(0.0564)	(0.0565)
Tertiary education	−0.0625	−0.0876	−0.106*	−0.104*
	(0.0574)	(0.0575)	(0.0557)	(0.0568)
Household economic situation before migration	−0.0557**	−0.0549**	−0.0485**	−0.0491**
Household economic situation before migration	(0.0221)	(0.0228)	(0.0225)	(0.0225)
Employed before migration	−0.0862**	−0.0872**	−0.105**	−0.0963**
Employed before migration	(0.0429)	(0.0435)	(0.0433)	(0.0435)
Main reason to migrate: study	−0.145**	−0.159***	−0.144**	−0.123*
Main reason to migrate: study	(0.0590)	(0.0552)	(0.0591)	(0.0700)
Year of return	0.0699***
	(0.0138)
Year of return: 2008		−0.0199	−0.0310	−0.0272
		(0.0663)	(0.0646)	(0.0648)
Year of return: 2009		0.0749	0.0706	0.0872
		(0.0693)	(0.0690)	(0.0704)
Year of return: 2010		0.190**	0.189**	0.193**
		(0.0743)	(0.0754)	(0.0760)
Year of return: 2011		0.483***	0.469***	0.506***
		(0.103)	(0.108)	(0.107)
Social assistance in the host country			0.174**	0.180**
Social assistance in the host country			(0.0706)	(0.0714)
Accumulated savings				0.0881**
				(0.0417)
Learned a language				−0.0497
				(0.0527)
Acquired soft skills				−0.0558
				(0.0448)
Pseudo R²	0.14	0.16	0.19	0.20
Observations	415	421	418	418

Note: for dummy variables marginal effects are calculated as discrete change from 0 to 1.

Standard errors in parentheses.

***

p < 0.01, **p < 0.05, *p < 0.1.

We find evidence of a positive effect of household wealth before migration on the likelihood of being active in the job market. The probability of inactivity is approximately 22 per cent higher for the returnees whose families were in debt and constantly borrowing in order to make a living before migration compared to those whose families were able to make substantial savings. This finding might be related to differences in the density and efficiency of social networks that are not cancelled out or mitigated by the migration experience. In line with the previous exercise, we find that holding a job before migration and migration for study purposes are associated with a lower probability of being inactive.

The probability of being economically inactive is higher for those individuals returning at the onset of the financial crisis (after 2009). The studies that investigate the reaction of workers to recession show that immigrants are more responsive to labour shocks than native workers (Boman, 2011). Therefore, during a strong recession foreign born individuals are more likely to spatially relocate (both internally and across borders), including the possibility of return to the home country. Our result suggests that individuals often do not return with ‘a job in their hands’¹¹ and that it is likely that the crisis that affected many destination countries acted as a ‘push’ factor for return migration to Poland, as documented in other studies (Zaiceva and Zimmermann, 2012).

We find evidence that individuals experiencing a skill mismatch during migration are more likely to be economically inactive upon return (between 8.6 and 9.7 per cent according to the model specification). Besides, when a returnee has been supported by the welfare state in the destination country (e.g. unemployment benefits, child benefits), the likelihood of being unemployed is more than 17 per cent higher.

Finally, it is interesting to note that migrants who have accumulated savings during the migration spell are significantly more likely to be economically inactive once they return. From Figure 1, it is evident that the effect of accumulated savings is positive and significant at all age levels, although, as expected, it is stronger for older returnees.

Figure 1.

Age and the probability of being inactive upon return in Poland.

Return migrants might experience difficulties upon return that go beyond the inability to employ the newly acquired human capital or economic inactivity.

As a final step of our analysis, we estimate Equation (1) [and the joint estimates of the endogenous regressor version, Equations (2) and (3)] using as a dependent variable ‘bitter return’, a dummy equal to 1 when the returnee experiences significant hardship and/or difficulties upon return (and in particular those related to an entry into the job market). In Table 5 we report the marginal effects of five alternative specifications. The first model (1 and 1a) is estimated using the full sample, while model 2 is restricted to returnees who were employed at the moment of the survey. Model 3 includes a set of dummies for the main countries of destination of the returnees. The last two specifications employ as a dependent variable a dummy equal to 1 when job-related difficulties are experienced by returnees. The main results highlighted above are partly confirmed when using a broad measure of unsuccessful return, but new findings also emerge. A skill mismatch during migration is a strong predictor of further hardship and difficulties upon return, in particular in getting back into the home country labour market. Our preferred estimates, that is, those correcting for the endogeneity of the skill mismatch during migration, imply an increase in the likelihood of difficulties upon return of more than 40 per cent at mean values.

Table 5.

Bitter return? The probability of facing hardship/difficulties upon return in the home country (marginal effects).

	(1)	(1bis)	(2)	(3)	(4)	(4bis)	(5)	(5bis)
Dependent variable	Bitter return	Bitter return (bivariate probit)	Bitter return (only employed individuals)	Bitter return	Job-related difficulties	Job-related difficulties (bivariate probit)	Job-related difficulties	Job-related difficulties (bivariate probit)
Skill mismatch during migration	0.117**	0.413***	0.124**	0.110**	0.135**	0.444***	0.139***	0.453***
Skill mismatch during migration	(0.0525)	(0.0874)	(0.0618)	(0.0543)	(0.0523)	(0.0456)	(0.0524)	(0.046)
Age	−0.008***	−0.006***	−0.0061	−0.009***	−0.0047**	−0.003**	−0.0034	−0.0024*
	(0.0024)	(0.002)	(0.004)	(0.0026)	(0.0024)	(0.0015)	(0.00235)	(0.0014)
Female	−0.0134	−0.0157	0.0257	−0.0277	0.0175	−0.0031	0.0277	−0.0016
	(0.0563)	(0.0384)	(0.0683)	(0.0576)	(0.0561)	(0.03445)	(0.0565)	(0.0331)
Vocational education	0.0638	0.0604	0.0355	0.0809	−0.0593	−0.008	−0.0538	−0.008
Vocational education	(0.0756)	(0.0576)	(0.0937)	(0.0770)	(0.0754)	(0.0564)	(0.0750)	(0.0554)
Tertiary education	−0.0158	−0.051	−0.0667	0.00649	−0.145*	−0.136**	−0.119	−0.106**
Tertiary education	(0.0776)	(0.0587)	(0.0911)	(0.0792)	(0.0766)	(0.0562)	(0.0768)	(0.0545)
Children (dummy)	−0.141**	−0.082*	−0.166**	−0.141**	−0.143***	−0.094**	−0.129**	−0.082**
Children (dummy)	(0.0548)	(0.0448)	(0.0671)	(0.0558)	(0.0551)	(0.037)	(0.0556)	(0.0354)
Household living on remittances	0.244***	0.185***	0.247***	0.259***	0.158**	0.111**	0.165**	0.1103**
Household living on remittances	(0.0717)	(0.0449)	(0.0937)	(0.0730)	(0.0741)	(0.0488)	(0.0744)	(0.0474)
Year of return	0.0223	0.0179*	0.0190	0.0317**	0.0610***	0.0374***
	(0.0155)	(0.0105)	(0.0191)	(0.0160)	(0.0158)	(0.0113)
Migration spell (in months)	0.0019**	0.0014**	0.000770	0.0018**	0.00121	0.009*	0.00112	0.00085
Migration spell (in months)	(0.0008)	(0.0006)	(0.00108)	(0.0009)	(0.0008)	(0.0005)	(0.0008)	(0.0005)
Main reason to migrate: study	−0.146	−0.0048	−0.114	−0.117	−0.197*	0.0037	−0.228**	−0.0215
Main reason to migrate: study	(0.103)	(0.098)	(0.111)	(0.106)	(0.103)	(0.0884)	(0.0968)	(0.086)
Social assistance in the host country	0.204***	0.168***	0.248***	0.203***	0.221***	0.185***	0.211***	0.169***
Social assistance in the host country	(0.0678)	(0.0578)	(0.0878)	(0.0690)	(0.0664)	(0.0536)	(0.0682)	(0.053)
Accumulated savings	0.151***	0.115***	0.159***	0.175***	0.0120	0.0184	0.00624	0.0167
Accumulated savings	(0.0506)	(0.0399)	(0.0598)	(0.0516)	(0.0517)	(0.0314)	(0.0524)	(0.0298)
Learned a language	0.112*	0.080*	0.115	0.0984	−0.0434	−0.0273	−0.0332	−0.0199
Learned a language	(0.0589)	(0.0487)	(0.0727)	(0.0602)	(0.0603)	(0.0366)	(0.0604)	(0.0347)
Acquired soft skills	−0.0640	−0.047	−0.0928	−0.0820	−0.104**	−0.0677**	−0.133**	−0.079**
Acquired soft skills	(0.0528)	(0.037)	(0.0632)	(0.0537)	(0.0528)	(0.0364)	(0.0532)	(0.0397)
Germany				−0.205***
				(0.0719)
United Kingdom				−0.0867
United Kingdom				(0.0639)
Ireland				0.00573
				(0.0903)
Scandinavia				−0.199*
				(0.112)
Italy				0.196*
				(0.117)
Year of return: 2008							0.164**	0.111**
Year of return: 2008							(0.0750)	(0.0466)
Year of return: 2009							0.343***	0.224***
Year of return: 2009							(0.0636)	(0.0568)
Year of return: 2010							0.293***	0.177***
Year of return: 2010							(0.0667)	(0.051)
Year of return: 2011							0.193**	0.132**
Year of return: 2011							(0.0949)	(0.0624)
Rho		−0.73 (0.27)				−0.82 (0.17)		−0.87 (0.18)
LR test Rho = 0 (chi²(1))		3.738**				6.535***		7.754***
Pseudo R²	0.097		0.083	0.125	0.107		0.127
Observations	446		318	446	446		454

Note: Marginal effects from probit estimates are reported; for dummy variables marginal effects are calculated as discrete change from 0 to 1.

Standard errors in parentheses.

***

p < 0.01, **p < 0.05, *p < 0.1.

Individuals from poor households (as proxied by the importance of the migrants’ remittances as a source of family income) and those using the welfare state in destination countries are more likely to experience a bitter return and greater difficulties in getting a job upon return. This result might be related to weaker social capital in the home country (Szymańska et al., 2012).

The social capital necessary for re-entering the home country labour market and society might be eroded as the migration spell increases. An additional year of migration increases the probability of a bitter return by approximately 3 per cent at mean values.

Although younger individuals are less likely to be economically inactive or unemployed, they are more likely to face job-related difficulties and more general hardships upon return. Job market search difficulties are approximately 10 per cent less likely to happen for the returnees with tertiary education (model 5bis).

In support of the hypothesis of negative effects of savings accumulated abroad on the labour supply of the returnees, we find that this variable is positively related to a bitter return, but not to job market difficulties. This finding suggests that the higher probability of economic inactivity is likely to be (at least to a great extent) voluntary.

It is interesting to note that a bitter return is less likely for migrants returning from Germany and Scandinavian countries, but more likely for those returning from Italy (model 3). Note that immigrants in Italy are relatively more likely to have an illegal status and are often employed in low-skilled and precarious occupations. More research in this respect is needed in order to disentangle the effects of an initial self-selection of immigrants into alternative destinations versus the effects of the characteristics of migration spells, which might as well be highly heterogeneous across countries.

Finally, we note that difficulties, in particular the job-related ones, are more likely for the immigrants who returned in 2009 and 2010, that is, during the most negative phase of the current crisis. This evidence suggests that the crisis might have substantial international spillovers that go beyond those related to the transmission of shocks through capital and financial markets.

Conclusions

International migration might represent an important channel through which countries and regions boost their economic potential, in particular when migrants return to their home location with new financial and human resources. In this study we investigate the factors that might hamper the link between return migration and development using evidence from a representative sample of returnees in a large and relatively developed Polish region, Silesia. Although we focus on a specific region, we believe that the analysis has important implications that might be arguably extended to several low- and middle-income regions across the world that are characterized by a rather intensive return migration.

Our analysis sheds light on the importance of the returnees’ characteristics as well as on the context within which migration takes place in destination countries. We show that a skill mismatch during migration is often associated with a skill waste also when the migrants return home, thus leading to a reduced economic potential embedded in return migration flows. In fact, the occurrence of brain waste/over-qualification during migration is a serious issue, influencing strongly the future fate of the returnees and lowering their chances of successful re-integration. The empirical results support Iglicka’s (2010) hypothesis of a ‘migration trap’ for some groups of returnees that are more vulnerable to skill waste and overall negative effects from the migration experience. This is the case with older return migrants, females and individuals from less wealthy households and with lower educational achievements.

On average, returnees to the Silesia region are rather successful in terms of economic re-integration, and new skills acquired abroad do contribute to a more successful return, in particular soft skills and technological knowhow (Grabowska-Lusińska and Okólski, 2009). In the light of the theoretical framework of Cassarino (2004), we find that often successful integration occurs in spite of the relatively moderate or even low level of the average returnees’ preparedness, as individuals often do not return with ‘a job in their hands’.

Although Silesia is a relatively a well-developed and fast-growing region compared to other Polish regions – a feature that facilitates the inclusion of return migrants in the labour market – the human resources waste related to the difficulties of re-adjusting is substantial (more that 45 per cent of returnees experience a ‘bitter return’). Reducing hindrances to a better use of human resources by the returning diaspora might boost not only the individual returns from migration, but also the social returns. Policies aimed at providing assistance to returnees – and in particular those who are likely to be more vulnerable – might be justified on these premises. Public institutions at the regional level might play a role in facilitating burdensome administrative practices (for instance, the certifications of diplomas and skills acquired abroad) and in matching skills of the returnees with those required by the local labour market.

One final remark is related to the time needed for a successful return. In line with Grabowska’s (2016) predictions, we find evidence that the skill waste tends to be reduced over time as the returnees will re-adjust to the home country labour market but, admittedly, in this respect the cross-sectional nature of our dataset is a limitation of our analysis. A longitudinal and dynamic analysis of the performance of the returnees on the Silesian regional labour market would have been preferable in this respect. In particular, some soft skills acquired abroad might become a source of a competitive advantage of the returnee over non-migrating workers only after a certain period of time. Future studies on the labour market integration of return migrants should particularly explore this direction of investigation.

Footnotes

Appendix

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

TThe author(s) received no financial support for the research, authorship, and/or publication of this article.

1.

For the USA, Borjas and Bratsberg (1996) find that return migration reinforces the same skill-selectivity of the initial migration flows. Whether return migrants are positively or negatively self-selected is highly context- and country-specific. For instance, Coniglio et al. (2009, ) find evidence of a positive self-selection of returnees when considering irregular immigrants residing in Italy. The authors argue that the ‘skill waste’ associated with the status of irregular migrants generates a higher opportunity cost of returning for the best and brightest among the migrants. On the other hand, given the high risks associated with emigration – in particular during the current financial crisis – one can expect that the higher the rate of ‘failed’ migration, the higher the risk that return migration flows will be negatively self-selected.

2.

In the Polish context the recent work by Anne White suggests that migrants are becoming increasingly mobile during their lifetime, which means that the homecoming event should not necessary be treated as a definite one, as more emigration (and subsequent return) episodes may take place (White, 2014a). Due to the nature of the data employed, we are not able to distinguish between definite/permanent and short-term return.

3.

The country of emigration from which return migration takes place is important for at least two reasons. Firstly, given the high heterogeneity in migration policies and barriers to migration, it is likely that individuals will positively (or negatively) self-select some destinations. This will obviously translate into a subsequent self-selection also in terms of return migration. Secondly, the success of the migration experience – for instance, in terms of accumulation of financial capital – clearly depends upon the destination country. Rich destinations and those destinations where migrants are more easily integrated will likely offer better opportunities. As we argue in this paper, the failure of the migratory experience – in particular in terms of using one’s skills and abilities – is an important determinant of an unsuccessful (or ‘bitter’) return to the home location.

4.

The ‘salt water curtain’ has become a symbolic description in migration studies since the British historian highlighted the high degree of impermeability between researches on the determinants of migration to the Americas from the studies that focused on the settlement process of newcomers in host countries. The remarks of Thistlethwaite still remain relevant for migration studies in the second decade of the 21 century, and for the economists researching return migration in particular.

5.

Detailed information on the questionnaire for return migrants are available upon request.

6.

Note that in each household only one return migrant was interviewed; in the case of the households with multiple returnees, the interview was conducted with the individual who had the longest migration experience.

7.

The definition of bitter return is based on self-declared difficulties faced by the returnees, such as the lack of/poor access to reliable information on formalities required upon return and no or weak support in the process of finding opportunities in the labour market.

8.

In our sample, more than one third of the returnees who had an irregular status during their migration spell were residing in Italy.

9.

Note that we find only weak evidence of endogeneity of the skill mismatch in the host country on skill waste upon return. reports the LR test on $ρ = 0,$ which is significant at the 10 per cent level only when considering the most parsimonious model reported in column 1.

10.

Based on LR and Wald tests on error terms correlation, we reject the hypothesis of endogeneity of regressors – in particular on the dummy variable measuring skill waste during the migration spell – and hence we report only probit estimates in .

11.

According to the so-called ‘contracted migration view’, individuals typically migrate with ‘a job in their hands’. This view is opposed to that of ‘speculative migration’, where geographical mobility is an integral part of the job search strategy. See for a survey of the empirical literature concerning the complex interrelationships between (un)employment status and migration. To our knowledge, the research on this issue is limited with respect to return migrants.

References

Anacka

Fihel

(2012) Return migration to Poland in the post-accession period. In: Galgóczi

Leschke

Watt

(eds) EU Labour Migration in Troubled Times. Skills Mismatch, Return and Policy Responses. Farnham, UK: Ashgate Publishing Limited, pp. 143–167.

Anacka

Matejko

Nestorowicz

(2013) Ready to move. Liquid return to Poland. In: Glorius

Grabowska-Lusińska

Kuvik

(eds) Mobility in Transition: Migration Patterns after EU Enlargement. Amsterdam: Amsterdam University Press, pp. 277–307.

Angrist

Pischke

(2009) Mostly Harmless Econometrics. Princeton, NJ: Princeton University Press.

Baum

Dong

Lewbel

Yang

(2012) Binary Choice Models with Endogenous Regressors. San Diego, CA: Stata Users Group.

Bieńkowska

Ulasiński

Szymańska

(2010a) Kierunek Małopolska. Charakterystyka powracających i rekomendacje jak nie zmarnować ich kapitału. Kraków: Centrum Doradztwa Strategicznego.

Bieńkowska

Ulasiński

Szymańska

(2010b) Warto wracać? Strategie zachowań reemigrantów i rozwiązania służące wykorzystaniu ich potencjału. Kraków: Centrum Doradztwa Strategicznego.

Bijwaard

Wahba

(2014) Do high-income or low-income immigrants leave faster? Journal of Development Economics 108(2014): 54–68.

Bijwaard

Schluter

Wahba

(2014) The impact of labor market dynamics on the return migration of immigrants. The Review of Economics and Statistics 96(3): 483–494.

Boman

(2011) The mobility of immigrants and natives: evidence from internal migration following job displacement. Regional Studies 45(3): 283–297.

10.

Borjas

Bratsberg

(1996) Who leaves? The outmigration of the foreign-born. Review of Economics and Statistics 78(1): 165–176.

11.

Brzozowski

(2011) Return migration of young Poles: implications for the regional labor markets. In: Larsen

Hasberg

Schmid

Bittner

Clement

(eds) Measuring Geographical Mobility in Regional Labour Market Monitoring: State of the Art and Perspectives. München u. Mering: Rainer Hampp Verlag, pp. 142–151.

12.

Cassarino

(2004) Theorising return migration: the conceptual approach to return migrants revisited. International Journal on Multicultural Societies (IJMS) 6(2): 253–279.

13.

Cerase

(1974) Expectations and reality: a case study of return migration from the United States to Southern Italy. International Migration Review 8(2): 245–262.

14.

Gang

Yun

(2000) Returns to returning. Journal of Population Economics 13(1): 57–79.

15.

Coniglio

De Arcangelis

Serlenga

(2009) Intentions to return of clandestine migrants: the perverse effect of illegality on skills. Review of Development Economics 13(4): 641–657.

16.

Coniglio

De Arcangelis

Serlenga

(2010) Return decisions of undocumented migrants: do network effects help the high-skilled overstay? Labour 24(1): 93–113.

17.

CSO (2015) Local Data Bank, data-set provided by Central Statistical Office of Poland. Available at: http://stat.gov.pl/bdlen/app/strona.html?p_name=indeks

18.

De Coulon

Piracha

(2005) Self-selection and the performance of return migrants: the source country perspective. Journal of Population Economics 18(4): 779–807.

19.

De Haas

(2010) Migration and development: a theoretical perspective. International Migration Review 44(1): 227–264.

20.

De Haas

Fokkema

Fihri

(2014) Return migration as failure or success? Journal of International Migration and Integration 16(2): 415–429.

21.

De Vreyer

Gubert

Robilliard

(2010) Are there returns to migration experience? An empirical analysis using data on return migrants and non-migrants in West Africa. Annals of Economics and Statistics/Annales d’Économie et de Statistique 97–98: 307–328.

22.

Diatta

Mbow

(1999) Releasing the development potential of return migration: the case of Senegal. International Migration 37(1): 243–266.

23.

Dustmann

(1997) Return migration, uncertainty and precautionary savings. Journal of Development Economics 52(2): 295–316.

24.

Fihel

Górny

(2013) To settle or to leave again? Patterns of return migration to Poland during the transition period. Central and Eastern European Migration Review 2(1): 55–76.

25.

Fihel

Grabowska-Lusińska

(2014) Labour market behaviours of back-and-forth migrants from Poland. International Migration 52(1): 22–35.

26.

Fihel

Tyrowicz

Kaczmarczyk

(2008) Migracje powrotne Polaków (Biuletyn nr. 5). Warsaw, Poland: Fundacja Inicjatyw Społeczno-Ekonomicznych (FISE).

27.

Fokkema

De Haas

(2015) Pre-and post-migration determinants of socio-cultural integration of African immigrants in Italy and Spain. International Migration 53(6): 3–26.

28.

Gmelch

(1980) Return migration. Annual Review of Anthropology 9: 135–159.

29.

Grabowska

(2016) Movers and Stayers: Social Mobility, Migration and Skills. Frankfurt am Main: Peter Lang.

30.

Grabowska

Jaźwińska

(2015) Migracje poakcesyjne Polaków i kapital ludzki: transfer wiedzy, umiejętności, kompetencji. Studia Migracyjne-Przegląd Polonijny 156(2): 53–80.

31.

Grabowska-Lusińska

Okólski

(2009) Emigracja ostatnia? Warsaw: Wydawnictwo Naukowe Scholar.

32.

Herzog

Schlottmann

Boehm

(1993) Migration as spatial job-search: a survey of empirical findings. Regional Studies 27: 327–340. 4

33.

Iara

(2006) Skill Diffusion by Temporary Migration? Returns to Western European Work Experience in Central and East European Countries. Wiiw Working Papers 46. Vienna, Austria: Wiener Institut für Internationale Wirtschaftsvergleiche.

34.

Iglicka

(eds) (2002) Migracje powrotne Polaków. Powroty sukcesu czy rozczarowania? Warszawa: Wydawnictwo Instytutu Spraw Publicznych.

35.

Iglicka

(2010) Powroty Polaków po 2004 roku: W pętli pułapki migracji. Warsaw, Poland: Wydawnictwo Naukowe Scholar.

36.

Jensen

Pedersen

(2007) To stay or not to stay? Out-migration of immigrants from Denmark. International Migration 45(5): 87–113.

37.

Kaczmarczyk

Tyrowicz

(2015) Winners and losers among skilled migrants – the case of post-accession Polish migrants to the UK. IZA Discussion Paper No. 9042. Bonn, Germany: Institute for the Study of Labor.

38.

King

(1978) Return migration: a neglected aspect of population geography. Area 10(3): 175–182.

39.

King

Skeldon

Vullnetari

(2008) Internal and international migration: bridging the theoretical divide. Sussex Centre for Migration Research Working Paper No. 52. Brighton: University of Sussex.

40.

Krings

Bobek

Moriarty

Salamonska

Wickham

(2009) Migration and recession: Polish migrants in post-Celtic Tiger Ireland. Sociological Research Online 14(2). http://www.socresonline.org.uk.sci-hub.cc/14/2/9.html. Doi:10.5153/sro.1927

41.

Lianos

Pseiridis

(2009) On-the-job skills and earnings of returned migrants. International Migration. 51(1): 78–91

42.

McCormick

Wahba

(2001) Overseas work experience, savings and entrepreneurship amongst return migrants to LCDs. Scottish Journal of Political Economy 48(2): 164–178.

43.

Martin

Radu

(2012) Return migration: the experience of Eastern Europe. International Migration 50(6): 109–128.

44.

Mayer

Harima

Freiling

(2015) Network benefits for Ghanaian diaspora and returnee entrepreneurs. Entrepreneurial Business and Economics Review 3(3): 95–122.

45.

Piracha

Vadean

(2010) Return migration and occupational choice: evidence from Albania. World Development 38(8): 1141–1155.

46.

Stark

Helmenstein

Yegorov

(1997) Migrants’ savings, purchasing power parity, and the optimal duration of migration. International Tax and Public Finance 4(3): 307–324.

47.

Szymańska

Ulasiński

Bieńkowska

(eds) (2012) Zaraz wracam. albo i nie. Skala powrotów, motywacje i strategie życiowe reemigrantów z województwa śląskiego. Kraków: Centrum Doradztwa Strategicznego.

48.

Thistlethwaite

(1960 [1991]) Migration from Europe overseas in the nineteenth and twentieth centuries (XIe Congrès International des Sciences Historiques, Rapports (Uppsala 1960)). In: Vecoli

Sinke

(eds) A Century of European Migrations, 1830–1930. Urbana, IL: University of Illinois Press, pp. 17–57.

49.

Thomas-Hope

(1999) Return migration to Jamaica and its development potential. International Migration 37(1): 183–207.

50.

Tilly

(2011) The impact of the economic crisis on international migration: a review. Work, Employment and Society 25(4): 675–692.

51.

White

(2014a) Double return migration: failed returns to Poland leading to settlement abroad and new transnational strategies. International Migration 52(6): 72–84.

52.

White

(2014b) Polish return and double return migration. Europe-Asia Studies 66(1): 25–49.

53.

Williams

Balaz

(2005) What human capital, which migrants? Returned skill migrants to Slovakia from the UK. International Migration Review 39(2): 439–468.

54.

Zaiceva

Zimmermann

(2012) Returning home at times of trouble? Return migration of EU enlargement migrants during the crisis. Discussion Paper Series No. 7111. Bonn: Forschungsinstitut zur Zukunft der Arbeit.