Motivation in the dynamics of European youth migration

People often have goals that they cannot accomplish in the place where they reside. To achieve these goals, which can be seen as ‘desired end states’ (Toth–Bos et al.2019), people compare their current place of residence to other possible destinations and might decide to move to a destination that they believe can help them achieve their goals.

The goals of migration are rarely independent of one another (De Jong and Fawcett 1981). Rather, they are primarily clustered according to spheres of life (Austin and Vancouver 1996; Clark and Maas 2015; Sandu et al.2018). For example, the goal of improving one's education could go with the goal of living closer to a place with better job conditions, friends and relatives, or a certain natural environment. Job skills enhancement goals tend to be associated with goals of improving one's education through migration. Most quantitative studies tend to ignore these clusters. In this study, we posit the existence of such clusters. This is in line with the idea that migration decision-making is characterized by patterns of motivations (Tartakovsky et al.2001) and not by atomized motivations.

This study examines motivations associated with migration goals during the migration process. In addition to a clustering of motivations according to spheres of life, we also posit the existence of a process linking motivations in the migration process over time. Our purpose here is to introduce the novel idea that, over time, patterns of migration motivations at the individual level become part of what is called the ‘internal dynamic of migration’ (De Haas 2010). Literature has long held that former migrations are conducive to new migrations (Lee 1966; Petersen 1958). According to this tradition, emigrants play a critical role in changing the network capital of their communities of origin, with ‘network capital’ referring to the array of interpersonal relationships—including relationships developed and maintained through mobilities, contacts, and communications—that can generate other types of capital, such as social, economic, symbolic, and cultural capital (Urry 2012). Implicitly, emigrants play a role in promoting higher emigration rates from their origin communities as a feedback mechanism in the dynamics of migration. This is the approach taken in the cumulative causation of migration theory (Massey et al.1999). The internal dynamics of migration approach also considers that there are vicious circles, that is, negative consequences of migration (such as diminishing human and material resources in the communities of origin), that favor new emigrations (De Haas 2010).

The approach we adopt here is in line with the theory of cumulative causation. However, it does not involve structural changes in community social capital as a mechanism of the internal dynamics of migration. It refers to causal chains at the individual level, in the sense that some types of motivations at moment t1 may be positively or negatively related to some other types of motivations for the same individual later in the migration process. Chain effects of motivations work not only through aggregation at the locality level, but also at the personal level. For the same person at different stages of the migration process, past and future migration motivations are intertwined.

What are the key patterns of motivation at the individual level in the migration process of young returnees? This is the basic research question that we will consider, taking into account migration stages, sociodemographic characteristics, and places of origin. We further examine a series of subquestions. First, among the young returnees, to what degree do motivations for the first migration influence motivations for remigration abroad? Second, to what degree is the inertia in the patterns of motivation linked to places of origin? The focus of our analysis is on cumulative changes in migration motivations at the individual level. We look at returned migrants in our analyses because this group allows us to examine the links between reasons for past emigration abroad and future reasons for territorial mobility or stability.

Return was initially conceived as an endpoint, either as a failure by neoclassical theory or as a success by the theory of new economics of migration (Cassarino 2004; Massey et al.1999). Here, we are more in line with the approaches that consider return a sequence in the migration process (Gemi and Triandafyllidou 2021). Depending on the resources and context, return could be converted into successful reintegration, circular migration, or permanent remigration abroad. This seems to be particularly true in the case of young people, whose migration is especially dynamic. The advantage of focusing on returnees is that this is an area where one can find patterns of interconnections among the motivations for the first migration, the return, and the remigration of returnees (i.e. returnees migrating again, after returning to their country of origin) or their potential remigration (i.e. the intention of returnees to migrate again).

Why focus on returnees if one wants a better understanding of remigration as a key mechanism in migration dynamics? First, return seems to be a frequent event in international migration flows, with some estimates putting it at around a quarter of all migration events (Azose and Raftery 2019), but research on return is still scarce (Hagan and Wassink 2020; King and Kushminder 2022). Second, it is important to look at returnees’ intentions to remigrate. This is because, in order to design viable migration policies, it is necessary to understand why many young returnees wish to remigrate. Finally, we want to see if it is possible to identify similar patterns of motivations for remigration among returnees, a population that has been described as highly heterogeneous (Hagan and Wassink 2020).

In the new economics of labor migration, return is considered a success (Massey et al.1999). If this is the case, why do many young returnees wish to remigrate? What are the demographic and geographical contexts that favor remigration over residential stability for young returnees? Additional empirical data that address these questions could contribute to the development of theories in this field. It is also important in this case to avoid what is called the ecological fallacy of transferring conclusions from the aggregate to the disaggregate level or the atomistic fallacy of transferring conclusions from the disaggregate to the aggregate level (Hox et al.2010). For this reason, we use individual-level survey data in this study, adjusting for the aggregate level in a multilevel approach.

In this paper, we analyze only cross-border migration, and we hypothesize that motivations for the first migration have some influence on motivations for future remigration of returnees. The key interest in our analysis is the relation between returnees’ recalled motivations for their first migration abroad and their declared motivations after their return for potential future remigration versus residential stability. Here, we look at remigration intentions of returnees (returnees’ intention to migrate again, following a return to their country of origin), and we do not consider other forms of remigration, such as remigration from the destination back to the origin (which we refer to simply as ‘return migration’) or remigration of returnees within their origin countries.

In the first part of this paper, we introduce a theoretical framework for migration motivations and their structuring as social worlds, communities of places, and spheres of life. The second part provides details on the methodological framework of this study. In the third and last part, we present and discuss the results of our analyses.

In order to better understand any sequence in the migration process and its motivation, it is important to distinguish between external and internal motivation frameworks. The external frameworks refer to the time, place, and social spaces in which motivation is structured (see Table 1). The time framework has three facets. The first facet is which stage in the migration process the motivation refers to (i.e. the first emigration, the return, the second emigration, or the intention to remigrate). The second facet is the time at which the motivation is expressed in relation to the referenced migration sequence. For example, the motivation for the first migration could be recorded before the migration movement, after emigrating to the destination place, or after returning home from the former destination. Recollections of the motivations after the movement are considered reasons. They are reinterpretations of past motivations migrants had before their change of residence. Such reasons are proxy estimations of the premovement motivations (Winchie and Carment 1989). The third facet is the life course (Findlay et al.2015) and age. These two are external frameworks of migration that are highly related to time.

Place frameworks for migration motivations at origin are multilevel and cumulative, combining individual profiles, residence (i.e. local communities, regions, and countries), and types and levels of development.

The social space of the motivation is essentially characterized by the spheres of life involved in the migration decision making (Benson and O’Reilly 2009; Gosnell and Abrams 2011; Ullmann et al.2011). These spheres of life are usually identified in terms of the individual's occupation, education (human capital), personal and family networks (personal communities), desired residential communities, and perceived personal problems that could be solved by residential mobility, discourse patterns, or life worlds (Clarke 2008). Social stratification indicators contribute to explaining the adoption of different types of motivations.

From an internal point of view (see Table 1), migration motivation could be determined by referring to the degree of structuring in intentions to migrate, the importance or intensity of the motivations assigned by individuals, the goals and reasons for migrating, and how reasons to move are clustered into factors or latent variables. In addition, the causal relations among the motivations of the same persons in different sequences of the migration process can provide information on the internal structures of migration motivation.

Values are mentioned under different names or labels, in several approaches on determinants of migration. In a long-running tradition in the social sciences, a value is something that is considered desirable or targeted as a goal in a specific social world. In the founding sociological approach, a social value is ‘any datum having an empirical content accessible to the members of some social group and a meaning with regard to which it is or may be an object of activity’ (Thomas and Znaniecki 1918: 21). In anthropology, social values appear as virtues or traits that societies promote to achieve specific goals (Benedict 1960[1934]). Social psychologists underline the idea that values are transsituational goals. That is, they function as enduring criteria in a variety of evaluations and choice behaviors (Schwartz and Sortheix 2018).

An external view of migration (Paasi 2002: 807) implies considering it not only in terms of its relationship with places and social spaces but also in terms of its relationship with life cycles or the life course and age. The internal framework includes migration intentions, values and goals in the migration process, and motivations. A distinction similar to that between external and internal frameworks is found in De Haas’ (2010) analysis of migration processes. He makes the distinction between exogenous macro-level factors and community factors or meso-level factors for endogenous mechanisms (De Haas 2010).

In our approach, we use the migration process as the basic framework for integrating information. We consider the changes from an unstructured desire to relocate to structured intentions, to preparatory actions, and to successive acts of migration (first emigration, first return, and circular movements). The data we analyze allow for an approach centered on our research question: What are the key patterns of motivation in the migration process of young returnees in different European contexts? While we do not specify the stages in the subjective process at the individual level, the changes are reflected in the transition from the first migration to return, to circular migration, and to intentions to migrate, categorized by previous migration experiences (Sandu et al.2018).

Even if we are underlining internal aspects of international migration, we also consider the role of some external facets of migration. In the literature, the external framework of reference for the migration process considers the impact of the life course (Clark 2013), urbanization (Beenstock et al.2015), or modernization (Zelinsky 1971) on migration.

We derive our hypotheses from the cumulative causation theory and from the new economics of migration theory (Massey et al.1999). We employ a framework that focuses on time and space. The cumulative aspect is given by the assumption that the reasons for future remigration are related to the reasons for the first migration. These were the reasons at the start of the migration process, before the migrants left their country of birth for the first time. They are also shaping the migrants’ reasons for returning. Migrants, as reflexive agents, redefine life situations in each stage of the migration process. Some, after their return, consider that they have achieved their goals. Others decide to remigrate as a result of their low integration into their society of origin or because they believe that only further migration could lead to the fulfillment of their objectives. Their family members might also play a significant role in the process of deciding whether to remigrate or not.

First, we hypothesize a chaining of motivations over time: the motivations of returnees to remigrate are influenced by the motivations for their first migration (H1 of chain causation of motivations over time). The expectation here is based on the idea that migration experience contributes to the structuring of a specific culture of migration associated with different practices in employment, education, family, leisure, the use of network capital, etc. Life cultures circulate and are transported by migrants and influence their aspirations and behaviors, including their behaviors of remigration. A recollection of the reasons for the first migration or for the return to the home country could function as a proxy for understanding the different cultures that influence the migrants’ potential remigration.

To our knowledge, the hypothesis has not been formulated as such before, but it integrates into the cumulative causation theory (Massey 1986; Massey et al.1999). However, while the cumulative causation theory states this hypothesis in reference to the intensity of migration at the community level, our hypothesis is formulated in reference to the content of motivations at the individual level.

A strong test of H1 should examine whether these effects are statistically significant, even after controlling for an extensive set of other relevant predictors of motivations for future migration. We expect motivations for the first migration abroad to influence the degree of importance attached to various reasons for future migrations, even after one controls for an extensive set of other relevant predictors (more details regarding the sets of predictors we employed in our analysis are provided in the following paragraphs).

As motivations to return are part of the chain of motivations over time, we also estimate models in which we explicitly account for the impact of motivations for return on the returnees’ motivations to remigrate, although these relationships are not our main interest in our conceptual model. We expect motivations to return to be informed by reasonings and governed by processes that are unlike those of the motivations in the first migration and the motivations to remigrate. The latter two types of motivations accompany movements or potential movements that are in the same direction (departures from the origin country). In the case of migrations with multiple returns to the origin, the motivations for the first return are likely to be more closely linked to the motivations for a second return. We expect the effects of the motivations for the first migration on the motivations to remigrate (the effects stipulated in H1) to be preserved, even when controlling for the effects of the motivations to return, in addition to the rest of the control variables in our models (i.e. the place, sociodemographic, and satisfaction variables). At the same time, we expect the motivations for return to have little additional predictive power over and above that of the motivations for the first migration and the rest of the control variables employed in our models.

Our second hypothesis (H2) is that noneconomic motivations for the first migration have a higher inertia than economic motivations in influencing motivations for future migration. This expectation is generated by the idea that noneconomic targets associated with migration might be more difficult to meet than economic targets. Consequently, once the opportunity for remigration after returning to the origin arises, non-economic reasons for remigration might also re-emerge.

The third and fourth hypotheses refer to expected regularities in the geographical space: we expect economic or material motivations in the job sphere to be higher in emigration countries than in immigration or emigration–immigration countries (H3). This expectation is derived from the idea that higher levels of poverty foster higher levels of interest in meeting economic needs. We also examine profiles of motivations in regions of the European Union (EU) in relation to social development levels, on the one hand, and economic development levels, on the other hand. We use NUTS2 regions in the EU's Nomenclature of Statistical Territorial Units for this analysis. We expect that EU regions with higher levels of social development are more likely to foster non-economic reasons for potential remigration. On the other hand, we expect that EU regions with higher economic development levels are less likely to foster economic reasons of potential remigration (H4). Thus, H4 stipulates that regional social development and economic development influence the potential remigration of returnees in different ways. Social reasons for potential migration are expected to be more favored in socially developed regions, with social development indicated by life expectancy at birth. Higher regional economic development, as measured by the per capita gross domestic product (GDP), is expected to diminish the importance of economic motivations for potential emigration.

We also include controls for potentially important confounders in this relationship, such as place variables, life satisfaction variables, and individual status variables, in order to correctly specify the effects of motivations for the first migration on motivations to remigrate (Arnold et al.2019). We follow the recommendation in the literature discussing causal inference with observational data, according to which the effects of primary interest explanatory variables and effects of confounders in regression models have different interpretations (Westreich and Greenland 2013). Thus, we first examine the effects of just the primary-interest explanatory variables (i.e. the motivations for the first migration) in these models. All such effects have the same meaning and interpretation—they are interpreted as estimates of the direct effects of motivations for the first migration on the motivations of returnees to remigrate, while controlling for potential confounders.

To present a complete picture of the way we conceptualize the relationships we are studying, we include in the conceptual model in Figure 1 the influences of unmeasured migration culture variables [e.g, information on destinations, ideologies of migration (Sandu and De Jong 1998), or changes in values produced by former migration], which we omitted from our regression models as they were not available in our data. The presence of unmeasured possible confounders, such as the migration culture variables, may introduce biases in the causal estimates of effects of past motivations. We attempt to minimize such biases through the use of the wide range of possible confounders we control for in our models, controls that are likely associated with the omitted variables and partly account for their effects.

Our last two hypotheses (H3 and H4) switch our focus to the role of place variables and how returnees’ motivations to remigrate vary across geographical spaces. While these variables served as necessary controls for confounders in our examination of the effects of motivations for the first migration on motivations to remigrate, they become predictors of primary interest in our discussion of geographical patterns and influences. We focus on interpreting the direct causal effects (as seen in the path analysis conception—see, e.g. Alwin and Hauser 1975) of these variables within the framework of our proposed conceptual model since we are interested in seeing whether these effects of geographical spaces will persist (i.e. will remain statistically significant) even after we account for their indirect effect on future migration motivations through the motivations for the first migration. We use the same empirical model we previously described to estimate and interpret these direct effects.

The data set for the analyses we present in this paper came from a large online survey on Youth Mobility (YMOBILITY) that was conducted between the end of 2015 and the start of 2016 in nine European countries: the United Kingdom, Germany, and Sweden as immigration countries; Ireland, Italy, and Spain as emigration and immigration countries; and Latvia, Romania, and Slovakia as emigration countries.

In the survey, a country was categorized as an immigration country if it had a much larger number of immigrants compared to emigrants, and as an emigration country if the opposite was true. The 29,679 respondents to this online survey were young (16- to 35-year-old) returnees, nonmigrants, and immigrants.

To our knowledge, this survey is one of the very few surveys in which the same individuals were asked about their reasons for their first migration, return, and potential remigration or stay, covering several different European countries.

Previous studies that used these data focused on explaining migration intentions (Williams et al.2018) and motivations for first migrations of returnees (Sandu et al.2018). In this study, we analyze these data for a different purpose: to determine if and how effectively past motivations can explain motivations for potential migration versus stability, while keeping under control a large set of variables that are relevant predictors of motivations for potential future migration (i.e. sociodemographic variables, life satisfaction, and territorial variables at the community, regional, and national levels). In the survey, only the subsample of 3,307 returnees were asked questions regarding their past motivations for migration since the questions were relevant only to them. Thus, in this study, we analyzed only the responses of this subsample. The survey defined a returnee as a young person who had lived or worked abroad for at least six months continuously and then returned to the home country. The returnees in the survey returned once or multiple times to their home country.

Due to the mode of data collection used (online surveys), the samples might not be fully representative of the target population. To correct for possible biases due to this, we weighted the data using post-stratification weights to account for the target groups’ gender, age, urban or rural residence, and education distributions at the country level. The data on the country-level population distributions for these variables were sourced from Eurostat, the European Statistical Office (see Williams et al.2018 for details regarding the construction of the survey weights). To explore the impact of the weighting, we conducted sensitivity analyses (Treiman 2014) that compared the weighted and unweighted regression model results. In general, the weighting did not affect the substantive model results, and it increased the prediction power (i.e. the coefficients of multiple determination) of our regression models.

We present descriptive statistics for the subsample of returnees and some comparisons between the subsamples of returnees and nonreturnees in Tables A1a, A1b, and A1c in the Supplemental materials section. The descriptive statistics indicate that more migration experience was associated with a higher probability of future migration, in line with other findings in literature (Ciobanu 2015; Docquier et al.2014; Williams et al.2018). This was true for all nine countries in the survey (see Table A1b, Supplemental materials).

A matrix question was used in the survey to capture first migration motivations. Returnees were asked: ‘What were your reasons when you first decided to migrate by yourself (rather than to accompany your family)? Please indicate the importance of each reason by ticking one box only (1 = not at all important; 5 = very important; or N/A = not applicable).’ A list of 17 reasons was provided (see column 1 in Table 2 and Table A2, Supplemental materials). The reasons were related to the main spheres of life—job, human capital, personal communities (friends and relatives), local communities, lifestyles, public services, climate, and others.

The same issues were transformed into another matrix question on the importance of the given reasons for migrating abroad or continuing to live in the same community: ‘In any decision that you make about moving abroad or staying, what is the importance of the following reasons (1 = not at all important; 5 = very important; or N/A = not applicable)?’

We argue that these questions are measuring motivations for potential future migration, given their sequence in the questionnaire. The returnees were first asked how important each of the 17 reasons was in their first migration. Immediately after that, they were asked whether they intended to migrate in the next year or in the next five years. That was directly followed by the matrix question on the importance of the same 17 reasons for any decision they will make about moving abroad or staying.

Our empirical models, a series of multivariable Ordinary Least Squares (OLS) regression models, adjusting for the clustered sampling design of the data, include our primary-interest explanatory variables measuring the importance of various reasons for the first migration. This is the set of 17 motivations for the first migration. However, these variables are relatively independent of one another. Indeed, measures to assess multicollinearity in our models suggested low degrees of multicollinearity. Therefore, rather than attempting to artificially reduce the complexity of our data by grouping these variables into a smaller number of scales or discarding some of them, we opted to include the full set of 17 measures as predictors in our models instead.

To test the sensitivity of our results to variations in methods of analysis (Treiman 2014), we also used a multi-item approach to predicting motivations for potential migration abroad (Table A5, Supplemental materials). The procedure indicated a high stability of the results, regardless of whether we used a single item or a multi-item approach.

We are also particularly interested in the effects of place variables on the formation of patterns of motivations for future migration, so we included these variables in our models as predictors. Additional controls include sociodemographic status and life satisfaction, but their effects are not of primary interest in this study. Our focus on the reasons for the first migration as predictors of motivations for potential migration is consistent with the cumulative causation theory (Massey et al.1999) in the sense of a self-perpetuating tendency toward migration but also in the sense of a conditioning of motivations for future migrations based on former motivations for migration. Origin ‘place’ variables are also of particular interest in our design because they capture specific conditions for relative deprivation or for life strategies that are explicitly considered in the new economics of labor migration (Massey et al.1999) and in the theory of planned behavior in the case of migration (De Jong and Fawcett 1981).

A descriptive examination of the data (Table A1a, Supplemental materials) shows a clear distinction between the pattern of motivations for the first migration in the three Eastern European countries (Romania, Latvia, and Slovakia) and that in the six surveyed countries in the Old EU (Spain, Italy, Germany, Sweden, the UK—which was still a member of the EU in 2015—and Ireland). The salary motivation, as a motivation for the first migration, was much higher for young people in the New EU than in the Old EU. On another dimension, motivations for the first migration related to improving education and investments in human capital were much higher in the Old EU than in the New EU countries. These differences in motivation could also have been driven by the sociodemographic profiles of the respondents, as there was a much higher share of manual workers in Central and Eastern European countries than in the rest of the countries in the sample. Consistently, the mean education level was much higher in the Old EU than in the New EU. The differences in the motivations for the first migration and in the sociodemographic profiles between the immigration and emigration–immigration countries were less clear-cut.

The results shown in Table 2 fully support our first hypothesis (H1), that motivations for the first migration, as recalled after returning home, are still important in justifying or giving reasons for future decisions to remigrate or stay. We found consistency among all of the types of motivations we examined, except for the motivation to earn an educational degree. If a particular motivation was important in justifying the first temporary emigration, the respondent was statistically significantly more likely to regard that motivation as important when justifying future migration decisions. There might have been an exception for educational degrees because the migrants might have already completed their educational goals during their previous migrations. The returnees’ reasons for future potential migration or stability associated with, for example, having a professional career were mainly predicted by having a motivation linked to one's career for the first migration abroad (see Table 2). A much less important but still statistically significant predictor of career motivations for future potential migration or stability decisions was the motivation to improve language skills for the first migration. Those who considered friends an important reason for their first migration tended to consider career a less important reason for future decisions to remigrate or stay (see Table 2).

Thus, career-related motivations seemed to be highly stable over time: if they were considered important in the first migration decision, they were likely to be considered important for future migration decisions and were hardly influenced by other motivations. Family-related motivations were similar: wanting to be closer to one's family in future migrations was mainly influenced by having the same motivation in the first migration. The only other significant predictors of family proximity-related motivations for future migrations were health-related motivations for the first migration (see Table 2). Other motivations that had the same pattern were motivations related to company transfers and quality of life (results not shown but are available upon request). As a group, these motivations are unique in that they are largely influenced by the same types of motivations in the past without much input from other types of migration motivations.

At the other extreme are reasons for future migration that are highly rooted in several reasons for the past migration, not only in an identical type of motivation in the past. A motivation for past migration may be linked to a motivation for possible future migration that has a different nature. This is the case for employment, education, and friends as important reasons for future migration (see Table 2). For example, employment-related motivations for future migration were linked not only to past economic motivations but also to past lifestyle motivations. Motivations for future migrations related to lifestyle were similarly associated with an array of other types of motivations from the first migration (results not shown but are available upon request). These relations among past and future reasons for migration may be seen as tools for understanding future mobility reasons through past motivation patterns in migration. When young Europeans who were interviewed in the survey said that employment was a very important reason for their future mobility or stability decision, we can understand that what they had in mind (in a decreasing order of importance) were career aspects, avoiding a precarious job, or accessing the appropriate conditions for a certain lifestyle; but on the other hand, family reasons were less important in this decision.

The same strategy of comparing regression coefficients for dependent variables with correlated meanings could improve our understanding of future motivations for mobility decisions. Migrating to be closer to family or friends is usually categorized under personal communities (Pahl and Spencer 2004). This semantic analysis of motivations allows us to better differentiate among different clusters of motivations. As mentioned, ‘family reasons’ refers to family and health—a rather focused meaning. ‘Reasons related to friends’ has a much larger scope, encompassing friends but also improvement of language skills, welfare, and housing.

Motivations for return have very small contributions over and above the contributions of the other predictors. The R² difference tests in Table 2 suggest small but statistically significant contributions of these variables to the prediction of motivations to remigrate (the largest R² difference amounted to 0.036), over and above those of the other control variables. Motivations for the first migration retained their significant effects even after the introduction of additional controls for return motivations, with the same exception—motivations related to education (results not presented but are available upon request).

The previous analyses supported H1, suggesting that past and future motivations for migration are intertwined. H2 tries to further clarify these relationships between past and future motivations for migration by examining in which cases the motivation inertia is higher. It stipulates that, compared to economic reasons for potential migration, noneconomic reasons are, to a larger degree, related to the same type of past experiences and motivations for migration. Testing the hypothesis involves a comparison of pairs of regression coefficients that estimate the effects of past motivations on the motivations for future potential migrations. In our interpretations we rely both on a comparison of coefficient sizes and we also examine statistical tests for significant differences between regression coefficients. Purely economic motivations are those that refer to salary and employment. We divided the noneconomic motivations referred to in H2 into several standard subcategories: social motivations (reasons related to family and friends), cultural motivations (reasons related to values and information such as being an exchange student, getting an educational degree, and improving language skills), and a combination of these two. The higher the regression coefficients estimating the influence of past motivations on future motivations for the same content item are, the higher the inertia effect is.

The results provide support for H2. Gaining new language skills as a cultural motivation for future migration is highly associated with the same motivation for the first migration (the partial OLS regression coefficient, b = 0.37, is statistically significant; see Table 2). The intention to migrate for climate reasons is also highly associated with the same motivation for the past migration (b = 0.38, which is statistically significant; the results are not shown but are available upon request). The inertia effects are also statistically significant for social reasons, such as moving to join friends or family (with statistically significant b values of around 0.23–0.26). Employment and salary economic motivations for future migrations have a lower inertia (with partial OLS regression coefficients of 0.12 and 0.15, respectively, both of which are statistically significant; see Table 2).

To test whether these differences between inertia effects were statistically significant, we also examined chi-square tests of the equality of the regression coefficients across models (results presented in Table A3, Supplemental materials). The results of these comparison tests confirm the stated conclusion, with a few exceptions. Cultural motivations (related to education and language skills) had consistently much higher inertia than all the economic motivations in our models (i.e. employment, career, salary, and job skills). Social motivations (related to friends and family) had statistically significantly higher inertia than employment motivations (but not distinguishable from the amounts of inertia in career and job skills motivations). In addition, motivations related to friends had significantly higher inertia than salary motivations.

What lies beyond the links between past motivations for the first migration and motivations for future potential migration? What is the role of returnees’ personal history of migration in shaping the patterns of motivations? These questions will be addressed in the next subsection.

We used the results of the same regression models we discussed in the previous section to explore the role of place variables in future mobility motivations. The urban status of the smallest territorial units of origin—villages, towns, or cities—does not contribute significantly to explaining the intensity of potential motivations. The only exception is motivations related to language skills, which young urban residents consider more important than rural residents do (see Table 3). The variables related to places of residence that have statistically significant effects on several motivations for future migration are predictors at the NUTS 2 regional level (i.e. GDP per capita, life expectancy at birth, and population density) and in the countries of residence (results presented in Table 3).

As expected under hypothesis H3, giving importance to economic motivations, such as employment- and salary-related motivations, for future migration or stay decisions is significantly associated with living in emigration countries, such as Latvia and Romania (see Table 3). The higher poverty levels in these countries are among the factors that increase the probability of potential emigration with the motivation of obtaining nonprecarious jobs and higher salaries. However, H3 is not supported in the case of Slovakia, which is an emigration country, as are Latvia and Romania. This deviation in the pattern could most likely be explained by the differences between the levels of economic development of the three countries. According to Eurostat data, Slovakia had a much higher level of economic development in 2015 than Latvia and Romania—the percentage of Slovakia's per capita GDP to the EU28 average was 77%, whereas that of Latvia was 64% and that of Romania, 56%. These country effects persists even after controlling for the per capita GDP as a percentage of the EU average at the level of the NUTS 2 regions in the OLS regression models.

As a further support for H3, residence in immigration countries (Germany and Sweden) or in emigration-immigration countries (Spain, Italy, Ireland) did not have statistically significant effects on the intensity of economic reasons for potential emigration (the UK was considered the reference category for the residence country variable; see Table 3).

In order to formally test for statistically significant differences between each of the three emigration countries and the remaining countries in the survey in terms of the intensity of economic motivations for potential future migration, we examined a series of alternative models. We re-estimated each regression model by cycling through the reference category among the three emigration countries, in order to test for differences between motivations in these countries (Romania, Latvia, and Slovakia) and motivations in the rest of the countries. By and large, the results (presented in Table A4, Supplemental materials) support H3, with the aforementioned exception in the case of Slovakia and a couple of other exceptions, noted below. Young returnees from Latvia place significantly more importance on each of the four economic motivations, compared to young returnees in immigration and emigration/immigration countries. The young returnees from Romania showed the same pattern for the employment and career motivations, but not for the salary and job skills motivations. On the other hand, as previously mentioned, Slovakia did not fit the expected pattern. In fact, there were no statistically significant differences between the young returnees in Slovakia and those in the immigration and emigration/immigration countries in terms of their regard for economic motivations. The country comparisons reveal one more notable exception from the pattern: the importance attached to motivations regarding job skills is actually no different in emigration countries compared to the rest of the countries.

In Table 2, some additional indirect support for H3 is provided by the finding that potential cultural migration for lifestyle is rare in all three emigration countries (Romania, Latvia, and Slovakia). Only one of the nonemigration countries (Italy) recorded a similar situation, with a very low level of potential lifestyle emigration.

H4 orients our analysis toward the impacts of the economic and social development of the regions on the reasons for potential migration therein. The expectation is that people from areas with higher social development levels are more inclined to emigrate based on noneconomic reasons and that people from areas with higher economic development are less inclined to emigrate based on economic reasons. Support for this hypothesis was tested by examining regression coefficients and their statistical significance for effects of regional per capita GDP (which is relevant for economic development) on economic motivations for migration, and by examining regression coefficients and their statistical significance for effects of life expectancy at birth (which is a measure of social development) on non-economic motivations for migration.

Both the per capita GDP and the life expectancy at birth were measured at the NUTS 2 level (see the results in Table 3). The implied logic is that the development profile of the region influences the motivation profile for future potential remigration. In this case, our results again supported H4. Better social conditions in the country of origin, as measured by a higher life expectancy at birth, were associated with a significantly higher importance given to public services, health conditions, and quality of life in the decision to migrate. All of these are social or sociocultural reasons for emigration. The effects of regional economic development on economic motivations to remigrate were negative, as we expected. However, these effects were not statistically significant. Thus, while our results supported the effects of social development on enhancing social motivations to remigrate, as stated in H4, our results did not support our expectations regarding the effects of economic development on depressing economic motivations for migration. In fact, regional economic development levels had very few statistically significant effects in our models. The only significant relationships were that young people who lived in less economically developed regions tended to consider climate and lifestyle more important when deciding whether to migrate or remain.

This study examined the migration process of young returnees, focusing on explaining their motivation dynamics. To see how this process is structured at the individual level, we investigated to what degree motivations of first migrations influence motivations for potential remigration or stability of young returnees. Even when keeping under control a large number of factors related to communities, regions, and countries of origin, as well as sociodemographic status and life satisfaction, first-migration motivation constitutes a key block of variables that predict the motivation to remigrate abroad or to stay in the same place.

This finding confirms H1, which states that there is a chain effect of motivations, at the individual level, over time; that is, past migration experiences are connected to future migrations, not only through changes in the network capital of migrant communities but also through an inertia of motivations from the first to future migrations of returnees.

There are a number of factors that could account for the continuity or inertia of migration motivations. Returnees feeling that they failed or missed their initial targets in their first migration could be inclined to retry migrating to accomplish their initial goals. The continuity might be sustained on the grounds of accumulated knowledge and networks in the course of the first migration, precisely to attain the initial targets. This could also be an effect of recalling and reconstructing the motivations for the first migration in order to make them more aligned with the current motivations for migration or stay.

It is also likely that the inertia of some motivations in the migration process is related to social mobility reasons (Favell 2008). It is possible that some returnees may wish to remigrate in order to achieve their social mobility goals that they were not able to accomplish when they first emigrated or even when they returned home. The analysis could also be expanded by connecting the dynamics of the migration motivation to the types of personalities of migrants from the points of view of work, achievement, power, and family orientation (Boneva and Frieze 2001). Further research and appropriate data are necessary to allow for such extensions.

The inertia effect of migration motivations differs according to the type of motivation. Noneconomic motivations for migration have a higher inertia over time than economic motivations. With a couple of exceptions, our findings offer support for H2. Our analyses clearly indicate that one of the highest past to future links is related to the motivation to acquire new language skills. This is an exemplary case of motivation inertia in the noneconomic area of motivations. In a similar manner, there is a high inertia of climate-related motivations. Compared to these findings, there is a much weaker link between the motivations related to employment and salary for future migration and similar motivations for the first migration.

The factors that mostly predict the variation of the motivations for potential migration or stability of returnees are first-migration motivations, country of residence, and sociodemographic status. In certain cases, community and regions of residence in countries of origin as well as life satisfaction are statistically significant predictors, but they are of secondary importance.

H3 formulated the expectation that economic motivations are more important in emigration countries. This was clearly confirmed by the youth from Romania and Latvia, who had very high regression coefficients in the regression models that predicted motivations related to employment. In addition, the Latvian youth were the most focused on career and salary motivations. Another study that used different models and the same dataset as ours had the same conclusion of a strong similarity in motivations for future migration between the youth from Latvia and those from Romania (Williams et al. 2018). Among the emigration countries covered by the survey, Slovakia was an exception to this pattern, likely due to its higher level of economic development. Among economic motivations, job skills were another exception, with young returnees in emigration countries attaching similar importance to these motivations as their counterparts in nonemigraton countries.

For H4, people from socially developed regions, as measured by the life expectancy at birth, were more inclined to remigrate for noneconomic reasons. Future studies using more detailed data should test whether this finding is consistent with the idea that anomie (in the Durkheimian sense) is a reason for elites to emigrate from crisis to noncrisis societies, as Bygnes (2017) found. On the other hand, the expectations in H4 related to the effects of regional economic development were not supported by the results of our analyses. Although these effects of economic development on economic motivations were negative, as we expected, they were not statistically significant, suggesting that people living in more economically developed regions do not necessarily discount the importance of economic motivations in their decisions to migrate or stay in the future.

Our analyses largely supported the view that the dynamics of motivation at the individual level are a key component of the larger system of the internal dynamics of migration. Former migration contributes to a self-perpetuation of migration not only through cumulative causation mechanisms at the community level but also through an inertia of motivations at the individual level, starting from motivations in the first or in a previous migration and influencing motivations for remigration among returnees.

Our approach to examining motivations in the migration process through the four lenses—time, place, social space, and internal structures—proved to be an efficient one.

It allowed for the construction and testing of regression models with good predictive power and with plausible results in light of what we know from previous research on motivations in the migration process. The value-expectancy theory for predicting motivations in the migration process was tested by including a whole set of reasons for past and future migration. These values continued to be relevant in explaining past or future migration even when controlling for place, time, and social status variables.

We proposed a dual-framework approach to studying the dynamics of migration motivations, considering external frameworks—time, place, social space, and internal frameworks—variables linked to individual intentions, values, goals, and motivations. This approach intersects and draws from several different theories of migration, such as the new economics of migration, the social capital theory, cumulative causation (Massey et al.1999), youth transition, and lifestyle migration (King 2018). The theoretical model we proposed also implies a multilevel approach, by simultaneously considering the individual, community, regional, and national levels (Fischer–Souan 2019; Williams et al.2018).

One of the key contributions of this study is the consideration of the role of the first migration motivation in structuring the motivations for potential future remigration of returnees. This study proposed a method of examining migration motivation in a cumulative way, within the framework of the migration process, and using a model that specifies both internal and external factors. Multiple comparisons between countries and between effects of regional development have allowed a better understanding of how patterns of motivations for future remigration differ between emigration versus immigration countries and between socially developed and economically developed regions. This study supports the idea that a cumulative nexus of factors influences migration dynamics, not only at the community level but also at the individual level.

Because it was impossible to measure all possible confounders, the estimates in our model are likely not unbiased causal estimates of past migration motivations on motivations for future migration. However, we attempted to reduce potential biases related to omitted variables by using a large array of possible confounders that are relevant predictors of motivations for potential future migration. While these confounders are likely associated with the omitted confounders and, therefore, partly accounted for their effects, this remains a limitation of this study that could not be circumvented with the data that we used.

The specification of the prediction models could have benefited from the inclusion of other measures but unfortunately, they were not present in our data. Other relevant measures that could be included in future studies are, for example, aspects related to the migration experience of family members, household incomes, subjective social classes, and self-perceptions of return as success or failure.

The data that we used in this study did not allow for comparisons between our target group (and the target group of the survey—the youth) and other age groups. Age and life course are well documented as playing key roles in structuring migration motivations (see, e.g. Dommermuth and Klüsener 2019). Starting from such studies and continuing the analyses presented here, future research could test the hypothesis that economic motivations decline with age.

No potential conflict of interest was reported by the author(s).

Dumitru Sandu is an Emeritus Professor of Sociology at the University of Bucharest. His main publications are on transnational migration, transition sociology, community and regional development, measures of social capital, and health sociology. He has published in various academic journals, including Current Sociology, International Sociology, Population Space and Place, International Migration Review, Central and Eastern European Migration Review, Migrationes, Romanian Journal of Population Studies, Sociological Forum, Population Research and Policy Review, Population, Current Research in Vaccines Vaccination, and The Journal of Primary Prevention.

Paula A. Tufiș is an Associate Professor of Sociology at the University of Bucharest. She holds a Ph.D. in Sociology from Pennsylvania State University (2007) and MA degrees in Sociology from Pennsylvania State University (2003) and Central European University (2001). Her main research areas are social stratification, migration, gender beliefs, child-rearing values, and quantitative research methods. Her publications have appeared in Sociological Forum, Sex Roles, The Annals of the American Academy of Political and Social Sciences, Current Sociology and other outlets.