High Selection,Low Success: The Heterogeneous Effect of Migrants’ Access to Employment on Their Remigration

Abstract

This article reconciles contrasting findings on the effect of access to employment on remigration by showing that this effect is actually heterogeneous and depends on migrants’ initial educational selection from the origin country. Combining longitudinal data from England and Wales (1971–2011) with data on educational attainment distributions in migrants’ origin countries, we find that the impact of being out of a job (unemployed or inactive) on the probability to remigrate is larger among migrants who were initially more positively selected in terms of educational attainment. This interaction effect appears stronger for male and recent migrants. Thus, in addition to migrants’ access to employment in the host country, the mismatch between migrants’ initial selection — that partly captures their premigration expectations — and their employment status at destination helps explain remigration behaviors.

Keywords

remigration unemployment educational selection

Introduction

Although migrants’ remigration has traditionally occupied a marginal position in the literature on migration, it has become an issue of growing interest to researchers in recent years (Dustmann and Görlach 2016). Many migrants do not stay permanently in the host country and eventually remigrate, either returning home or migrating to another country (Piore 1979; Warren and Peck 1980; Dustmann and Weiss 2007). In developed OECD countries, in fact, it is estimated that 20 to 50 percent of migrants leave the destination country within five years after arrival (OECD 2008). Who are those migrants who remigrate?

Existing studies agree that sociocultural attachment to the destination country decreases migrants’ probability to remigrate (e.g., Constant and Massey 2003; de Haas and Fokkema 2011). Nevertheless, while migrants’ economic integration at destination constitutes a high-stakes policy issue in a context of prolonged economic crisis and substantial migration flows, its effect on migrants’ remigration behavior is less clear. While most existing research finds that migrants who are unemployed are more likely to leave the host country (Constant and Massey 2003; Bijwaard, Schluter, and Wahba 2014), other studies paint a more complex picture by finding no significant effect of migrants’ labor-market participation in the receiving country on their decision to remigrate (de Haas, Fokkema, and Fassi Fihri 2015).

This article sheds new light on the role of migrants’ access to employment at destination in their potential remigration, defined here as leaving the host country, whether returning to the origin country or migrating to a new destination. We contribute to reconciling contrasting findings on the effect of access to employment on remigration by showing that this effect is actually heterogeneous and depends on migrants’ initial educational selection from the origin country. To do so, we combine linked-census longitudinal data on foreign-born individuals in England and Wales with data on their birth countries to construct an individual-level measure of initial selection that compares migrants’ educational attainment with that of same-sex and same-age peers within their origin country. This measure of migrant selection allows us to highlight the heterogeneous impact of migrants’ employment status at destination on remigration.

We show that the impact of unemployment on migrants’ likelihood to leave England and Wales is significantly larger among those who held higher levels of relative educational attainment within their origin country. We interpret our main finding following the idea that remigration not only is driven by migrants’ absolute level of economic integration at destination but also results from a mismatch between migrants’ expectations prior to migration and their actual economic situation at destination. Because our measure of initial selection represents a comparison with a relevant reference group in the origin society, we argue that it partly captures migrants’ socioeconomic aspirations when emigrating. Migrants who were at the top of the educational distribution in their origin country have higher socioeconomic expectations at destination and are, therefore, more likely to remigrate if these expectations are unfulfilled. The article further discusses the role of duration of stay, gender, and household composition in enhancing or diminishing the interaction effect between initial educational selection and employment status on remigration. We find a larger interaction effect for recently arrived male migrants and show that the “unfulfilled expectations” of migrants’ spouses also play a role in their own decision to remigrate.

This study focuses on England and Wales for several reasons. First, Great Britain is an old country of immigration (Owen 2003) where the share of migrants is of significant importance: in 2011, around 13 percent of residents of England and Wales were born outside the United Kingdom (ONS 2012). Second, remigration is far from negligible in Britain. Dustmann and Weiss (2007) estimated that, among migrants who arrived in Britain between 1992 and 2002 and were still in the country one year after arrival, about 40 percent of all males and 55 percent of all females had left the country five years later. Finally, the great diversity of Britain’s migrant population allows us to investigate a variety of origin countries and hence to understand the role of premigration characteristics in modifying the effect of economic integration on remigration.

The rest of the article is structured as follows. We first provide an overview of existing studies on remigration determinants. We then argue for a more direct consideration of migrants’ selectivity to understand remigration behaviors. After setting out our hypotheses and describing our data and methods, we present our findings, which show that the effect of migrants’ employment status on their probability to remigrate varies according to their initial educational selectivity. The conclusion discusses the significance of our results and their limitations.

Previous Literature: The Determinants of Remigration

Research in the United States has underlined the importance of networks in migrants’ initial migration (Massey et al. 1993). Similarly, sociocultural attachment to the origin country and the host society plays key roles in migrants’ decisions to remigrate, especially in the case of return migrations. For example, in their seminal work on Mexico–US migration, Massey and Espinosa (1997) showed that having a wife in the United States, as well as having US-born children, strongly lowered the likelihood of return migration. Symmetrically, maintaining strong social and economic ties with the origin country, usually referred to as “transnationalism” (Schiller, Basch, and Blanc 1995; Portes 1999), tends to increase return migration rates (van Hook and Zhang 2011). Despite some exceptions (de Haas et al. 2015), there is broad agreement that return migration intentions or behaviors are largely shaped by the relative strength of this “integration-transnationalism matrix” (Carling and Pettersen 2014).

However, research on remigration tends to provide more conflicting theories and findings on the effect of migrants’ economic integration at destination on their probability to remigrate. Most research on this issue focuses on return migration (Dustmann and Görlach 2016). Until the 1980s, the then-dominant neoclassical model explained labor migrations as a result of an individual’s income maximization strategy based on supply and demand imbalances in international labor markets (Lewis 1954; Harris and Todaro 1970). As Constant and Massey (2002) clearly put it, this theoretical framework assumed that migrants permanently remained at destination, unless their project of economic integration “failed” (i.e., migrants’ pre-migration expectations for a better socioeconomic position were not met at destination). According to this perspective, migrants experiencing economic hardship, such as unemployment or low-paid jobs, are more likely to return. The New Economics of Labor Migration (NELM) offers an alternative conception of remigration by modelling it as a household-level strategy of risk diversification (Lucas and Stark 1985; Stark 1991). The key idea here is that one household member’s migration can constitute a financing strategy for the family left behind in the origin country through remittances (Lucas and Stark 1985; Stark 1991) or for projects planned by migrants themselves once they return (Dustmann and Kirchkamp 2002). In this framework, migrants return if and when they have successfully reached a certain level of savings. Therefore, return migrants are more likely to hold an income-generating job, although unemployment is still expected to increase the probability to remigrate as migrants’ initial “raison d’être for migration has disappeared” (Constant and Massey 2002, 11).

These different theoretical predictions partly mirror the heterogeneity of results produced by empirical research. Migrants who are unemployed are often found to be more likely to leave the receiving country, such as Germany (Constant and Massey 2003) or the Netherlands (Bijwaard et al. 2014). Massey (1987b) showed that Mexican return migrants from the United States were negatively selected on wages, which is consistent with findings in Sweden (Edin, LaLonde, and Åslund 2000) and Canada (Picot and Piraino 2013). Abramitzky, Boustan, and Eriksson (2014) also identified that among migrants who arrived in the United States during the age of mass migration (1850–1913), those who eventually left were more likely to hold lower-paying occupations.

Nevertheless, other studies show that the effect of migrants’ employment status or earnings at destination on remigration is not always clear cut. Bijwaard and Wahba (2014), for example, report higher probabilities for migrants to leave the Netherlands in both low- and high-income groups. In Germany, Kırdar (2009) shows that unemployment’s effect on remigration depends on the length of time the migrant is unemployed: migrants whose unemployment spell is shorter than three years are more likely to return, but older migrants experiencing longer unemployment periods are more likely to stay. Emigration rates and the role of migrants’ economic position in the destination labor market are also likely to vary over the life course and duration of stay. In their literature review, Dustmann and Gorläch (2016) underline that out-migration rates are often highest during the first decade after migrants’ arrival and then level out. Duleep (1994) highlighted a bimodal pattern of remigration among legal migrants who eventually left the United States. “Mistaken migrants” returned soon after arrival because of the integration “failure” described earlier and contrasted with “retirement migrants” who were more economically successful and remigrated when they no longer participated in the labor market.

As de Haas and his colleagues (2015) underline, these mixed empirical results illustrate that there is no uniform process of remigration and call for more attention to migrants’ heterogeneity to understand such complexity. This article aims at contributing to this discussion by investigating the heterogeneous effect of migrants’ access to employment in the British labor market on their probability to remigrate. More precisely, we examine how migrants’ educational attainment relative to their counterparts at origin differently shapes their decision to remigrate as a response to unemployment in the receiving society.

Measuring migrants’ selection and contextual attainment to understand remigration patterns

A possible differentiating role of initial educational selection on remigration echoes Borjas and Bratsberg’s (1996) core idea that the determinants of remigration depend on migrants’ skill selection into initial migration from their origin country. In their model based on the relative returns to skills in the origin and destination countries, they argued that return migration accentuated the direction of initial selection. This seminal work and its predictions have led to a relatively large body of literature on outmigration (e.g., Reagan and Olsen 2000; Jensen and Pedersen 2007; Cohen and Haberfeld 2007; Bönisch, Gaffert, and Wilde 2013). However, studies which have empirically tested this idea remain scarce, as migrant selection into initial emigration is difficult to measure (e.g., Rooth and Saarela 2007; Bönisch et al. 2013). The notion of migrant selection means that individuals who migrate are different from those who do not. As a consequence, measuring selection requires a combination of information comparing two subpopulations (Feliciano 2005; Ichou 2014): a group of interest (“those who left” the origin country) and a reference population (“those who stayed”).

To link migrants’ initial selection to their trajectories at destination, information from both the origin and destination countries is required. Yet, with a few exceptions, such as the Mexican Migration, MAFE (Migrations between Africa and Europe), and 2,000-Families projects (Massey 1987a; Beauchemin 2014; Güveli et al. 2017), multisited longitudinal datasets providing such information remain extremely rare. Abramitzky and his colleagues (2016) developed an ingenious matching method between censuses to examine migration and return migration of Norwegian migrants to the United States from 1865 to 1910. A number of other studies have also attempted to measure selectivity for Mexican migrants to the United States (Massey 1987b; Chiquiar and Hanson 2005; McKenzie and Rapoport 2010). Although these methods have undeniable merits, their two main drawbacks are the small number of countries they can capture (typically one origin country and one destination country) and their relatively low matching rates.¹

These few studies aside, past research has overwhelmingly relied on remote aggregate proxies, such as macro-characteristics of origin and destination countries, that are assumed to influence migration’s costs and benefits. These country-level characteristics range from geographical distance to differences in Gross Domestic Product (GDP) per capita, income inequality, migration policies, and returns to education between the origin and destination countries (Jasso and Rosenzweig 1986; Cobb-Clark 1993; Levels and Dronkers 2008; Pedersen, Pytlikova, and Smith 2008). Such measures are problematic because they do not directly compare “those who left” to “those who stayed” and because they potentially hide large differences in the level of selection of individual migrants from the same origin country.

In this article, we measure migrants’ initial selection through their relative education level compared to the population of the same age and sex in their birth country, restricting the analysis to migrants who did not study or obtain a degree in England or Wales. We match individual-level data on migrants’ educational attainment in the destination country (England and Wales) with aggregate data on educational attainment in migrants’ origin countries, as described in the next section. This approach, introduced by Ichou (2014) and expanded by Feliciano and Lanuza (2017), offers several advantages. First, it provides an individual-level measure of migrants’ educational selection into initial emigration by comparing migrants with similar counterparts within their origin country. Second, this relative educational attainment measure is available for a wide range of origin countries. Third, it is likely to be a good indicator of migrants’ (un)observed skills, as it provides insights into what Feliciano and Lanuza (2017) call “contextual attainment.” Relative educational attainment partly captures migrants’ perception of the social position they held in their origin country’s social class hierarchy (Ichou 2018). Since this perception influences their social aspirations, “placing educational attainments within the contexts they were attained captures broader dimensions of education and social class” (Feliciano and Lanuza 2017, 214). Contextual attainment is, thus, likely to affect the formation of migrants’ premigration expectations and their hope in the socioeconomic returns to migration (Santelli 2001; Sayad 2004).

We argue that migrants may use their relative educational attainment as a benchmark to assess the success of their economic integration in the receiving country. The key idea in our approach is to position migrants within the educational attainment distribution of their peers at origin, and not in comparison to natives or other migrants in the host society, as “people do not move to compete with other groups in the destination society but to improve their life chances — and their children’s — relative to what they would have been in the origin society” (Zuccotti, Ganzeboom, and Güveli 2017, 98). Hence, migrants with high relative educational attainment will tend to place higher hopes in their migration undertaking, to be more disappointed if they are not socioeconomically successful at destination, and to be more likely to remigrate.

This argument is compatible with the ‘Unfulfilled Expectations Hypothesis’ (Beenstock 1996), which also echoes a strand of literature pointing to the role of education–occupation (mis)matches in migration decisions. In Mexico, Quinn and Rubb (2005) and Villarreal (2016) show that individuals who are overeducated for their occupations are more dissatisfied with their jobs and, consequently, more likely to emigrate from their native country in search of opportunities more appropriate to their skills. We adopt a similar perspective to understand remigration behaviors by focusing on the gap between migrants’ contextual attainment and the economic reality they face in their access to employment in the destination country.

Hypotheses

We expect that the effect of unemployment in the receiving country on remigration varies, depending on migrants’ initial educational selection. Specifically, we suggest that individuals who were initially positively selected (i.e., who have a higher education level than peers in their origin country) but experience unemployment at destination will have a higher likelihood to remigrate than other unemployed migrants.

Furthermore, we suggest that migrants’ duration of stay at destination affects this interaction effect. On the one hand, this interaction effect could be larger for migrants who have recently arrived in England and Wales, as remigration shortly after arrival is more often related to a difficult socioeconomic integration (DaVanzo 1983; Duleep 1994). Premigration expectations are known to be related to the information that migrants possess about the destination country before leaving (Massey et al. 1993), and migrants could adjust their initial socioeconomic expectations to the reality of integration processes after some time spent at destination (Gmelch 1980). In Borjas and Bratsberg’s model (1996), imperfect information regarding returns to skills at destination can lead to unmet expectations for migrants, who will be constrained to reassess their migration plan in light of better information on the host country (Sabates-Wheeler, Taylor, and Natali 2009). Therefore, highly selected migrants who find themselves unemployed might be especially prone to remigration shortly after arriving in the destination country. On the other hand, we can imagine that migrants who were initially positively selected in terms of education had better information before emigrating and formulated long-term plans for their integration in the host society. From this perspective, they may more easily tolerate economic difficulties at the beginning of their stay, bearing in mind that their economic integration might come with time. Under these assumptions, highly selected migrants who are unemployed would only be more likely to remigrate after some time in the destination country.

We also expect that this initial selection–unemployment interaction effect is more pronounced for male migrants than for female migrants. If, on average, men have mainly migrated to the United Kingdom for economic reasons and female migrants subsequently followed them (Owen 2003), the latter are less likely to have developed economic aspirations and, thus, to be disappointed by difficult working conditions in the receiving country. In the United Kingdom, in the period covered by our data, female migrants are characterized by lower labor-market participation rates (Dustmann and Fabbri 2005). Therefore, male migrants who find themselves unemployed may experience more social stigma and disappointment than female migrants, for whom this situation is more common. Job satisfaction has also been shown to be gendered (Phelan 1994; Sloane and Williams 2000): If female migrants tend to have lower occupational expectations (Clark 1997), they may be less disappointed if they are unemployed or hold a part-time job.

As described earlier, remigration is also likely to be driven by migrants’ social and familial ties to the receiving country. As spouses constitute strong intimate ties, we expect their characteristics to have an observable impact on migrants’ remigration decisions. Among migrants who are married, we investigate the effect of a mismatch between their spouses’ relative educational attainment and economic position in England and Wales on migrants’ decision to remigrate. We hypothesize that, in addition to their own characteristics, migrants whose spouses experience difficult economic integration relative to their initial contextual attainment will also tend to remigrate more.

Data

The Office for National Statistics Longitudinal Study (LS)

Our main data source is the Office for National Statistics Longitudinal Study (LS),² which is a representative 1-percent sample of the population of England and Wales. Since 1971, the LS has longitudinally linked census and life event information on people born on one of four undisclosed birth dates. The sample is regularly updated to include newborn and migrants born on any of those four dates, as well as information on all deaths and exits notified to the National Health Service (NHS). Currently, five censuses (1971, 1981, 1991, 2001, and 2011) compose the main information source of the LS. We use this long-term prospective longitudinal dataset to capture remigrants indirectly. In panel data that follow individuals at the national level, attrition can be attributed to death, outmigration, or the inability to reach respondents. When respondents’ death information is systematically collected and respondents’ tracking is of very high quality, attrition can be an appropriate way to measure outmigration. This approach, employed in our analysis, has been repeatedly used in the literature (e.g., Dustmann 2003; van Hook et al. 2006; Picot and Piraino 2013; Abramitzky et al. 2014).

The Barro–Lee Educational Attainment Dataset

We also use the Barro–Lee Educational Attainment Dataset (Barro and Lee 2013) to create our measure of initial educational selectivity. This dataset compiles international data on educational attainment from 1950 to 2010 in 146 countries, using harmonized data from UNESCO, Eurostat, and other sources. It contains the distribution of educational attainment in the adult population by sex, five-year period, and five-year age group. This dataset has been used in a wide range of research, including in macroeconomics (e.g., Hanushek and Woessmann 2012; Aisen and Veiga 2013), demography (e.g., Ichou and Wallace 2019), and sociology (e.g., Ichou 2014; Feliciano and Lanuza 2017; Engzell and Ichou 2019).

Methods

Variables

Remigration from England and Wales is our dependent variable. We define “leavers” as migrants present in England or Wales at t, absent at t₊₁ and t₊₂ , and for whom we do not have a death certificate in England or Wales between t and t₊₁ . By considering the absence at two consecutive censuses, we restrict the potential bias due to miscounts of migrants at any one census. Since three census years are needed to construct our dependent variable (i.e., information at t, t₊₁ , and t₊₂ ), t corresponds to the 1971, 1981, and 1991 censuses.³ Figure 1 presents the evolution over the study period of the share of migrants who left England and Wales at each census, by sex and employment status. Overall, the proportion of leavers is similar for male and female migrants and remains rather stable, despite a small increase over the period (thick black lines). The average proportion of leavers ranges between 23 and 33 percent, which is consistent with figures often reported in the empirical literature (OECD 2008; Dustmann and Görlach 2016).

Figure 1.

Proportion of Remigrants among Migrants by Gender, Employment Status and Inter-census Period.

To explain remigration, our analysis uses two key independent variables of interest. First, migrants’ educational selection — what Feliciano and Lanuza (2017) named “contextual attainment” — captures migrants’ initial level of selection into emigration by comparing their educational attainment to that of the population from the same origin country. We combine the LS and Barro–Lee datasets to construct this measure, following the method proposed by Ichou (2014). In the LS data, we know migrants’ educational attainment (e.g., a male Indian migrant born in 1950 and living in England).⁴ We match each migrant with the distribution of educational attainment of individuals of the same sex and birth year available in the Barro–Lee dataset (e.g., the educational attainment distribution of Indian males born in 1950). Our measure of migrants’ relative level of educational attainment locates the educational attainment of each migrant from the LS data within the educational distribution of his/her origin country for same-sex and same-age peers drawn from the Barro–Lee dataset. Formally, migrants’ relative educational attainment $π_{i}$ is calculated as:

π_{i} = 100 \frac{\sum_{o r i g} X = X_{i}, L < L_{i} + \frac{1}{2} \sum_{o r i g} X = X_{i}, L = L_{i}}{\sum_{o r i g} X = X_{i}}

where X_i denotes a vector containing the sex and birth cohort of migrant i, L_i for her educational level, and $\sum_{o r i g} (.)$ is the population in the origin country with characteristics X_i and L_i . The resulting variable of the relative educational position, measured in percentile, shows the percentage of individuals from the birth country with identical characteristics X_i (sex, birth cohort) who have a lower education level, plus half those with the same education level.

We restrict the analysis to migrants whose educational attainment level did not increase after their arrival in England or Wales and who were never recorded as students at any census, ensuring that the measure of migrants’ initial selectivity captures only premigration characteristics. This restriction is justified by the fact that remigration decisions are likely to be specific for migrations initially undertaken for the purpose of studying abroad (Dustmann and Glitz 2011). This restriction slightly reduces the proportion of highly educated migrants in our analytical sample; however, distributions of other characteristics, including sex, age, and origin country, remain basically unchanged.

Figure 2 shows the dispersion of this measure of relative educational attainment for permanent migrants (stayers) and remigrants (leavers) by sex. Among both females and males, remigrants are more positively selected than stayers. Among remigrants, relative educational attainment is on average (black circle) 61.6 for males and 60.7 for females, which means that they respectively hold a higher level of educational attainment than 61.6 percent and 60.7 percent of the population of the same sex and birth cohort in their origin country. Among migrants who remained in England or Wales over the period, this proportion is only 53.7 percent for males and 54.0 for females. Although the distributions appear similar across genders, the difference in median relative educational attainment between remigrants and permanent migrants proves larger among men (63.8 percent for leavers vs 44.2 percent for stayers) than women (49.5 percent vs 46.0 percent).

Figure 2.

Distribution of Migrants’ Relative Educational Attainment by Type of Migration and Gender.

Our second explanatory variable of interest measures migrants’ employment status (employed full-time, employed part-time, unemployed, inactive). Since it is census based, the LS data do not provide information on income. However, employment status is a key indicator of the degree of migrants’ economic integration in the destination country’s labor market (England and Wales), and access to employment is known to be an important economic driver of remigration decisions. Constant and Massey (2003), for example, show that guest workers’ decisions to leave Germany are more strongly driven by their access to employment than by their wage level.

Figure 3 displays the evolution of the proportion of migrants in each employment status, respectively for men and women. While almost 50 percent of female migrants are economically inactive over the period, only a small share of their male counterparts are in this situation. Similarly, part-time employment is more common among female workers. Further descriptive statistics (not displayed) show that these part-time jobs are overrepresented among low-skilled routine occupations. We also observe a significant increase in the unemployment rate between 1971 and 1991, especially for male migrants, going from 5.2 to 16.2 percent.

Figure 3.

Distribution of Employment Status by Gender and Census.

A further observation of Figure 1 (above) suggests that, with the exception of part-time employment among women and inactivity among men, migrants’ remigration behavior does not seem to vary substantially by employment status on average. However, as we hypothesize, these small average effects might hide significant differences in the impact of employment status according to migrants’ initial educational selection.

Finally, Figure 4 presents the distribution of our measure of relative educational attainment by employment status. On average, migrants exhibit levels of initial educational selection that are surprisingly similar across employment status, with a mean between 52.9 percent for the economically inactive and 57.4 percent for migrants employed full-time. Yet, the variance of this indicator is larger for migrants who are employed, especially part-time, in comparison to those who are not, especially the inactive.

Figure 4.

Distribution of Migrants’ Relative Educational Attainment by Employment Status.

Remigration Equations

Our main analysis is based on the estimation of two equations that examine the determinants of the probability for migrants to leave England or Wales:

Y_{i t} = β_{0} + β_{1} π_{i} + β_{2} δ_{i t} + β_{4} y e a r_{t} + β^{'} X_{i t} + ∊_{i t}

Y_{i t} = β_{0} + β_{1} π_{i} + β_{2} δ_{i t} + β_{3} (π_{i} * δ_{i t}) + β_{4} y e a r_{t} + β^{'} X_{i t} + ∊_{i t}

where the binary outcome $Y_{i t}$ is equal to one if individual i left England or Wales between t and t₊₁ and did not reappear in the census at t₊₂ .

Equation (1) includes the two variables of interest: migrant i’s initial educational selection $π_{i}$ and his/her employment status in England and Wales at t $δ_{i t}$ . We use $π_{i}$ as a linear continuous variable ranging from 0 to 100. We do not introduce a squared term in the models presented, as it never proved statistically significant. We control for the census year and for a vector of sociodemographic characteristics X_it that are expected to affect remigration behaviors: age group, sex, absolute level of educational attainment (less than middle school, middle school, high school, and higher education), marital status (married to a migrant, married to a native, or not married), having children, and birth country. According to existing research presented in the previous section, we can expect migrants who are married to a native and who have children living in the destination country to be less likely to remigrate. Beyond information on marital status and the presence of children, the administrative nature of the data we use prevents us from undertaking a broader investigation of social capital’s effect on migrants’ remigration.

Equation (1)’s aim is to examine the overall determinants of the probability of leaving the host country, with a particular focus on the roles played by migrants’ initial educational selectivity and employment status. Equation (2) additionally includes an interaction effect between migrants’ initial educational selection $π_{i}$ and employment status $δ_{i t}$ , to test our main hypothesis that the effect of migrants’ employment status on their remigration depends on their initial educational selection. Examining this interaction effect’s statistical significance, direction, and magnitude will be the focus of this article’s empirical section. We estimate these equations with random-effects logistic regressions to ensure that predicted values fall between zero and one, yet robustness checks using linear probability models provide similar results.

The second part of the empirical analysis explores the extent to which these effects vary between male and female migrants and by duration of stay. We estimate equation (2) separately by sex and for “recent” and “settled” migrants. “Recent” migrants are those who are present at only one census and then leave the country within the first 10 years of their arrival, which corresponds to the timespan between two UK censuses. By contrast, “settled” migrants refer to migrants who are present in at least two consecutive censuses. For estimations using this distinction by duration of stay in England and Wales, we cannot use panel regressions since recent migrants who remigrate are by definition present only at one observation date. Instead, we use cross-sectional logit regressions with clustered standard errors for each individual.

Since we study economic integration in the destination country’s labor market, all analyses presented hereafter are restricted to individuals who are aged between 18 and 50 at the time of the census and not students. We intentionally do not study the specific remigration decisions related to either the end of studies or retirement, since these decisions have been shown to be of a different nature from those related to the economically active population (Güngör and Tansel 2008; Cobb-Clark and Stillman 2013) and since our main theoretical model focuses on migrants’ situation in the labor market. Finally, we restrict the sample to married migrants to investigate the role of the mismatch between spouses’ initial educational selection and economic integration on the migrant respondent’s decision to remigrate.

Although the prospective longitudinal LS data allow us to track migrants’ outmigration from England and Wales over a long period of time, these census-derived data do not provide a full picture of the complex processes underlying remigration decisions. In particular, we cannot control for the nature and strength of migrants’ cross-border network connections with their origin country. Additionally, the LS data do not include information on some key variables of migrants’ sociocultural integration at destination, such as linguistic abilities. Finally, the direction of causality between employment status and remigration behaviors remains uncertain since our models do not control for potential issues of endogeneity. It is always possible that an existing propensity to remigrate affects migrants’ decisions to participate in the labor market. Migrants who have already decided to remigrate may, for example, be more likely to invest in their origin country and less in the host society, adapting their current economic behaviors to their future departure (Chabé-Ferret, Machado, and Wahba 2018). Nevertheless, our core aim in this article is not to assess the average causal effect of migrants’ employment status at destination on their decision to remigrate but to provide insights on how this relationship varies according to migrants’ initial selection, as the next section lays out.

Results

Table 1 compares the average marginal effects of migrants’ relative educational attainment and employment status in England or Wales. Models 1 and 2 provide the estimations of equations (1) and (2), respectively.

Table 1.

Average Marginal Effects of Migrants’ Relative Educational Attainment, Employment Status and Their Interaction on Their Probability to Remigrate.

		Model 1		Model 2
		Coeff	SE	Coeff	SE
Initial Educational Selectivity		–.0007*	.0003	–.0008**	.0003
Employment Status	Employed Full-Time
	Employed Part-Time	–.0205**	.0072	–.0459**	.0158
	Unemployed	.0109	.0067	–.0329*	.0148
	Inactive	.0266***	.0058	.0217^†	.0123
Employment Status*Initial Educational Selectivity
	Employed Part-Time			.0004^†	.0002
	Unemployed			.0008**	.0002
	Inactive			.0001	.0002
Person Years Observations		35,495		35,495
Number of Individuals		25,205		25,205

Control variables: Absolute level of education, age group, sex, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

^† p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed z-tests).

Source: Longitudinal Study (ONS 2011), and Barro and Lee (2013).

The first model shows that the two variables of interest have a statistically significant independent effect on remigration. Controlling for a number of individual characteristics including absolute educational level, we observe a small negative impact of migrants’ initial educational selection on their probability to leave England and Wales. A 10-point increase on the 0-to-100 scale of relative educational attainment corresponds to a 0.7-percentage point decrease in remigration probability: migrants who were initially more positively selected than peers within their origin country are slightly less likely to eventually remigrate. We also find that, compared to those who are employed full-time, migrants who are inactive (i.e., those who are out of a job and not immediately seeking employment) are 2.66 percentage points more likely to leave the host country, all things being equal. Unemployed migrants also seem more likely to remigrate, but the average marginal effect fails to reach conventional thresholds of statistical significance. The boundaries between unemployment and inactivity are in any case not always clear cut. Individuals who have been unemployed for months or years may not declare themselves actively looking for a job at the time of the census and would, thus, be identified as inactive rather than as unemployed.

This first model also shows that migrants employed part-time are more likely to settle in England and Wales in comparison to those in full-time employment. This finding is compatible with the NELM theory: because part-time jobs more likely reflect precarious labor-market situations, it would be more difficult for migrants to accumulate sufficient savings to return to the home country. These migrants may also hold part-time jobs due to a lack of documentation that would prevent them from participating in the formal labor market, something for which we cannot control in our data. Undocumented migrants may also be less likely to leave, as they would not be able to reenter British territory easily.

The central goal of our research is, however, not to document the average effect of migrants’ employment status at destination on their propensity to remigrate but rather to shed light on how this effect may differ depending on migrants’ premigration attainment. To examine this question, Model 2 introduces an interaction effect between migrants’ educational selection and employment status (equation 2). By allowing the role of employment status to vary depending on migrants’ relative educational attainment, this model reveals more nuanced patterns than those displayed in Model 1. Results specifically show that the effect of being unemployed rather than employed full-time varies significantly across levels of initial educational selection, as the interaction effect between the two variables of interest is significant and positive. The small magnitude of the marginal effect of this interaction (0.0008) should be understood with reference to the scale of our measure of initial educational selection ranging from 0 to 100. Thus, the impact of being unemployed, rather than employed full-time, on the probability to remigrate increases by 0.08 percentage point for each increase of 1 percentage point in initial selection. Put differently, when a migrant has an initial educational selection level that is 10 points higher than an otherwise-similar migrant, his/her probability to remigrate when unemployed is 0.8 percentage point higher. We also find a positive and significant interaction effect involving the contrast between part-time and full-time employment, yet both the significance level and magnitude of this coefficient (0.0004) appear smaller than for unemployment.

To have a better grasp of these interaction effects, Figure 5 displays the average predicted probabilities to leave England and Wales by employment status for different levels of initial educational selection. The effect of being unemployed rather than employed full-time on the probability to remigrate varies substantially as a function of migrants’ initial level of educational selection. Among unemployed migrants, those who have low levels of relative educational attainment are less likely to remigrate than migrants with similar relative education levels who are employed full-time. The opposite is true for migrants with high levels of relative educational attainment: the unemployed are more likely to leave than the full-time employed. This finding resonates with our hypothesis on premigration expectations. The probability to leave the host country is highest for migrants who hold higher degrees than their counterparts at origin and find themselves unemployed at destination, as they expected better economic integration at destination.⁵

Figure 5.

Predicted Probabilities to Remigrate by Migrants’ Relative Educational Attainment and Employment Status. Control variables: Absolute level of education, age group, sex, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

Average marginal effects associated with control variables of Models 1 and 2 are available in Supplemental Table 1A in Supplemental Appendix. Educated migrants are consistently more likely to remigrate than those who did not go to school or who dropped out before completing middle school. Model 2 exhibits significant sex differences, with men being more likely to remigrate, ceteris paribus. The probability to leave also depends on the origin country. We take as a reference category individuals from India, which is the largest group of migrants in England and Wales over the period of our analyses (Dustmann et al. 2003). Compared to Indians, migrants from sub-Saharan African and other South Asian countries (Pakistan and Bangladesh) are consistently more likely to stay at destination, whereas the reverse is observed for migrants from the United States, Australia, or European countries and other advanced economies.⁶ Assuming that these remigrations are returns, these findings may illustrate that unstable sociopolitical and economic situations, including contexts of violence, at origin are likely to discourage migrants’ remigration.⁷ They are also consistent with recent studies that underlined restrictive migration policies’ role in remigration decisions, especially for sub-Saharan African migrants, who are all the less likely to return as the possibility of moving back to the destination country is uncertain (Flahaux 2017).

In line with the existing literature (e.g., Massey and Espinosa 1997; Constant and Massey 2003), our results also confirm the role of migrants’ sociodemographic characteristics in remigration decisions (see Table 1A in the Appendix). Migrants are more likely to leave England or Wales when they are not married and have no children or when they are in the first half of their working lives. Interestingly, marriage’s effect varies, depending on whether the spouse is British or foreign born: intermarriage with a native reduces the probability to remigrate in comparison to being married to another migrant.

Heterogeneity across Sex and Duration of Stay

We now test the hypothesis that heterogeneity in the effect of migrants’ employment status on remigration is stronger among men, estimating equation (2) separately by sex. We report the average marginal effects for the two variables of interest and their interaction in Table 2 for male (Model 3) and female (Model 4) migrants. Once again, we represent these effects visually in Figures 6 and 7, which show the probabilities to remigrate by migrants’ relative educational attainment and employment status, respectively, for male and female migrants.

Table 2.

Average Marginal Effects of Migrants’ Relative Educational Attainment, Employment Status and Their Interaction on Their Probability to Remigrate, by Sex.

		Men (Model 3)		Women (Model 4)
		Coeff	SE	Coeff	SE
Initial Educational Selectivity		–.0008^†	.0004	–.0011**	.0004
Employment Status	Employed Full-Time
	Employed Part-Time	–.0979*	.0381	–.0217	.0181
	Unemployed	–.0369*	.0186	–.0373	.0251
	Inactive	–.1014***	.0302	.0428**	.0152
Employment Status*Initial Educational Selectivity
	Employed Part-Time	.0016**	.0006	.0000	.0003
	Unemployed	.0009**	.0003	.0007^†	.0004
	Inactive	.0021***	.0006	–.0004^†	.0002
Person Years Observations		16,799		18,696
Number of Individuals		11,820		13,385

Control variables: Absolute level of education, age group, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

^† p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed z-tests).

Source: Longitudinal Study (ONS 2011), and Barro and Lee (2013).

Figure 6.

Predicted Probabilities to Remigrate by Migrants’ Relative Educational Attainment and Employment Status among Male Migrants. Control variables: Absolute level of education, age group, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

Figure 7.

Predicted Probabilities to Remigrate by Migrants’ Relative Educational Attainment and Employment Status among Female Migrants. Control variables: Absolute level of education, age group, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

Interestingly, these models uncover a different pattern of remigration by sex. For both sexes, the effect of being unemployed, rather than employed full-time, on remigration becomes stronger as their initial educational selection level increases, but this interaction’s significance and magnitude appear larger for male migrants. Higher initial educational selection also makes it more likely for male migrants who are employed part-time or inactive, rather than employed full-time, to remigrate, as is visible in Figures 6. The slope of the curve representing the predicted probabilities to remigrate for migrants who are employed full-time (black line) differs strongly from the others: it is the only one that decreases as the level of relative educational attainment increases. The effect’s magnitude is especially large for the economically inactive, as these migrants’ probability to remigrate increases by 0.2 percentage points for each percentage-point increase in their initial educational selection. By contrast, we find no significant interaction effect associated with part-time employment for female migrants and a barely significant, very small, and negative interaction effect for the inactive. This result corroborates one of our hypotheses: given that men experience part-time employment or economic inactivity less frequently (see Figure 3), they might be more likely to be disappointed by the mismatch between a high initial educational selection and these often unwanted and potentially stigmatizing employment statuses, and, hence, more likely to remigrate.

Next, we explore whether heterogeneity in the effect of migrants’ access to employment on remigration varies depending on migrants’ length of stay in the host country. We refine the estimation of equation (2) by using separate models by sex and length of stay (for “recent” migrants, present at only one census, and “settled” migrants, present at more than one census). Average marginal effects for the variables of interest are reported in Table 3. The positive interaction effect between initial educational selection and employment status is statistically significant among recent male migrants (Model 5, unemployment or inactivity compared to full-time employment) and recent female migrants (Model 7, unemployment compared to full-time employment). By contrast, no significant interaction effects appear among settled male migrants (Model 6), and a very small and barely significant negative interaction associated with inactivity is observed among settled female migrants (Model 8). Thus, the interaction effect between initial educational selection and employment status at destination on remigration behavior seems more pronounced in the first years after arrival. This finding is consistent with the idea that migrants readjust their initial socioeconomic expectations over time spent in the destination country. This readjustment tends to decrease the impact on remigration decisions of any initial mismatch between high educational selection and poor economic integration.

Table 3.

Average Marginal Effects of Migrants’ Relative Educational Attainment, Employment Status and Their Interaction on Their Probability to Remigrate, by Sex and Duration of Stay.

		Men		Women
		Recent(Model 5)	Settled(Model 6)	Recent(Model 7)	Settled(Model 8)
Initial Educational Selectivity		–.0010^† (.0002)	.0005 (.0005)	.0001 (.0005)	.0007 (.0005)
Employment Status	Employed Full-Time
	Employed Part-Time	–.0658 (.0826)	–.0080 (.0435)	.01563 (.0456)	–.0166 (.0159)
	Unemployed	–.1058** (.0358)	.0352 (.0244)	–.1159** (.0410)	.0265 (.0320)
	Inactive	–.1596** (.0526)	–.0367 (.0305)	.0167 (.0268)	.0463** (.0177)
Employment Status*Initial Educational Selectivity
	Employed Part-Time	.0011 (.0011)	.0006 (.0008)	–.0007 (.0006)	.0001 (.0003)
	Unemployed	.0013* (.0006)	.0003 (.0005)	.0012^† (.0007)	.0002 (.0003)
	Inactive	.0031** (.0009)	.0012 (.0009)	.0002 (.0004)	–.0005^† (.0003)
Number of Observations		7,720	9,079	8,905	9,791

Control variables: Absolute level of education, age group, marital status (married to a migrant, to a native, or not married), having children, country of birth, and census year.

^† p < .10. *p < .05. **p < .01. ***p < .001 (two-tailed z-tests).

Source: Longitudinal Study (ONS 2011), and Barro and Lee (2013).

The role of migrants’ spouses’ characteristics

For married migrants, the decision to remigrate is likely to depend on their spouses’ access to employment. To explore this issue further, we ran the last model, this time including information on the employment status and relative educational attainment of respondents’ spouses, as well as an interaction term between these two variables. Estimations are displayed in Model 9 (Table 4). As in Model 2, we observe that the effect of respondents’ unemployment status significantly varies, depending on his/her level of initial educational selectivity. We observe a very similar pattern of interaction for respondents’ spouses’ characteristics: when migrants’ spouses’ relative educational attainments are higher but they experience unemployment, migrant respondents are more likely to remigrate. Thus, in addition to migrants’ own characteristics, the mismatch between their spouses’ initial educational selection and employment status affects the probability for migrants themselves to remigrate.

Table 4.

Average Marginal Effects of Migrants’ and Their Spouses’ Relative Educational Attainment, Employment Status and Their Interaction on Migrants’ Probability to Remigrate.

		Model 9
		Coeff	SE
Initial Educational Selectivity		–.0000	.0003
Employment Status	Employed Full-Time
	Employed Part-Time	–.0259^†	.0149
	Unemployed	–.0289^†	.0150
	Inactive	.0252*	.0117
Employment Status*Initial Educational Selectivity
	Employed Part-Time	.0002	.0002
	Unemployed	.0007**	.0002
	Inactive	–.0001	.0002
Spouse’s Initial Educational Selectivity		–.0060***	.0005
Spouse’s Employment Status	Employed Full-Time
	Employed Part-Time	–.0162	.0210
	Unemployed	–.0628*	.0273
	Inactive	–.0028	.0167
Spouse’s Employment Status*Spouse’s Initial Educational Selectivity
	Employed Part-Time	.0003	.0004
	Unemployed	.0012*	.0005
	Inactive	.0002	.0003
Person Years Observations		24,339
Number of Individuals		18,174

Control variables: Absolute level of education, age group, sex, spouse’s migration status (migrant or native), having children, country of birth and census year, spouse’s absolute level of education. Model 9 is run only on married migrants.

^† p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed z-tests).

Source: Longitudinal Study (ONS 2011), and Barro and Lee (2013).

Conclusion

Existing research disagrees on the effect of migrants’ unemployment on their decision to remigrate. While many studies find that migrants who are unemployed are more likely to remigrate from the host country (e.g., Constant and Massey 2003; Bijwaard et al. 2014), others fail to find any such effect (e.g., de Haas et al. 2015). This article contributes to reconciling these contrasting findings by showing that the effect of migrants’ employment status on their remigration depends on their initial level of educational selection.

To do so, we constructed a measure of migrants’ initial educational selection from their origin country by combining data on migrants in England and Wales with internationally harmonized data on educational attainment distribution in the population of migrants’ birth countries. Results underline that migrants who were initially highly selected in terms of education relative to their same-sex and same-age peers in the origin country and who find themselves out of a job, rather than employed, in the host country are more likely to remigrate than initially less-selected migrants in a similar labor-market situation. We find this interaction even when controlling for migrants’ absolute level of education. This finding suggests that, beyond their nominal educational degree, migrants’ relative position within the origin society specifically matters when they assess their economic integration at destination. The heterogeneity in unemployment’s effect is such that it actually reverses, depending on initial educational selection. Among negatively selected migrants, those who are unemployed rather than employed full-time are more likely to stay in the destination country, whereas positively selected migrants who are unemployed are more likely to leave. This interaction effect between initial educational selection and employment status proved stronger among male and recent migrants. Finally, while our data prevented us from analyzing the known influence of social networks in detail, results highlight the importance of spouses’ characteristics in remigration behaviors. Indeed, migrants’ remigration decision are affected not only by the mismatch between their own educational selection and employment status but also by the same mismatch in their spouses. Spouses whose higher relative educational attainment has not protected them against unemployment weaken the household’s attachment to the destination country. Hence, migrants’ in these households are more likely to remigrate.

Empirically, our article improves the existing literature in two main ways. First, it capitalizes on more than 40 years of a large-scale prospective longitudinal and representative data to track migrants’ presence in England and Wales, allowing us to construct a quality measure of remigration. Second, instead of ad-hoc aggregate proxies, we measure migrant selection by creating a direct individual-level measure of migrants’ relative educational attainment compared to the population of their origin country. Rather than using an overall indicator of “migrant quality” to capture selection, we focus on migrants’ initial selection in terms of premigration relative educational attainment. This measure of “contextual attainment” (Feliciano and Lanuza 2017) both provides information on migrants’ skills selection and partly captures migrants’ own perception of their social position within their origin country, which is likely to have influenced their socioeconomic expectations before their initial migration to England and Wales.

Theoretically, this article supports the idea that the discrepancy between migrants’ initial educational selection and their actual economic integration plays a role in the decision to remigrate. Our findings suggest that the decision to remigrate is not only driven by migrants’ level of economic integration at destination but also likely to be affected by a broader expectations–achievement mismatch among initially highly selected migrants who find themselves in unfavorable labor-market situations in the host country, while their relative position at origin made them expect better socioeconomic outcomes. The effect of this expectations–achievement mismatch is gendered (higher for men) and larger for recent migrants. Our findings corroborate the more general idea that migrants’ “social position before migration provides an important reference point by which [they] judge their success in the new country” (Engzell and Ichou 2019, 1).

While our results are compatible with an interpretation that places unfulfilled initial expectations as a core social–psychological mechanism shaping remigration decisions, other mechanisms could also play a role. Highly selected migrants may be more likely to remigrate because they believe that they will have better job opportunities or higher economic returns to their skills elsewhere, regardless of whether their premigration position was matched in terms of economic integration at destination. We can also imagine that unemployment benefits represent a larger proportion of the expected income of migrants who were initially negatively selected and that the incentives to stay in England and Wales are, therefore, higher for this migrant group, which is something that we cannot measure with our data.

To help adjudicate between these alternative, yet nonexclusive, interpretations of our empirical results, future research could provide more direct tests of the unfulfilled expectations hypothesis by including direct measures of migrants’ initial expectations and by capturing the situation of unemployed migrants more precisely (potential income source, odd jobs, training, adaptation to the host country’s language, etc.). In addition, distinguishing return and onward migration would be of interest, as research has shown that motivations for these two types of remigration can differ (Nekby 2006; Larramona 2013; Toma and Castagnone 2015). The two types of migration could correspond to two different responses to economic challenges at destination and the unfulfilled expectations they create. Disappointed migrants might return home to end their migration endeavor — at least temporarily, while those who remigrate to a new destination country might be those who hope for better economic success elsewhere.

Supplemental Material

Appendix - High Selection, Low Success: The Heterogeneous Effect of Migrants’ Access to Employment on Their Remigration

Appendix for High Selection, Low Success: The Heterogeneous Effect of Migrants’ Access to Employment on Their Remigration by Louise Caron and Mathieu Ichou in International Migration Review

Footnotes

Acknowledgments

The permission of the Office for National Statistics (ONS) to use the Longitudinal Study is gratefully acknowledged (Project 0301766), as is the help provided by staff of the Centre for Longitudinal Study Information & User Support (CeLSIUS). CeLSIUS is supported by the ESRC Census of Population Programme (Award Ref: ES/K000823X/1). The authors alone are responsible for the interpretation of the data.

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

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is supported by a public grant overseen by the French National Research Agency (ANR) as part of the “Investissements d’Avenir” program LIEPP (reference: ANR-11-LABX0091, ANR-11-IDEX-0005-02).

ORCID iD

Louise Caron

Supplemental Material

The Supplemental Appendix is available in the online version of this article at .

Notes

References

Abramitzky

Boustan

Eriksson

. 2019. “To the New World and Back Again: Return Migrants in the Age of Mass Migration.” ILR Review 72(2):300–22.

Abramitzky

Boustan

Eriksson

. 2016. “To the New World and Back Again: Return Migrants in the Age of Mass Migration.” NBER Working Paper 22659:1–37.

Abramitzky

Boustan

Eriksson

. 2014. “A Nation of Immigrants: Assimilation and Economic Outcomes in the Age of Mass Migration.” Journal of Political Economy 122(3):467–506.

Ahrens

Kelly

van Liempt

. 2016. “Free Movement? The Onward Migration of EU Citizens born in Somalia, Iran, and Nigeria.” Population, Space and Place 22(1):84–98.

Aisen

Veiga

F. J.

. 2013. “How Does Political instability affect economic growth?” European Journal of Political Economy 29:151–167.

Barro

R. J.

Lee

J. W.

. 2013. “A New Data Set of Educational Attainment in the World, 1950–2010.” Journal of Development Economics 104(C):184–198.

Beauchemin

2014. “A Manifesto for Quantitative Multi-sited Approaches to International Migration.” International Migration Review 48(4):921–938.

Beenstock

1996. “Failure to Absorb: Remigration by Immigrants into Israel.” International Migration Review 30(4):950–978.

Bijwaard

G. E.

Schluter

Wahba

. 2014. “The Impact of Labor Market Dynamics on the Return Migration of Immigrants.” Review of Economics and Statistics 96(3):483–94.

10.

Bijwaard

G. E.

Schluter

Wahba

. 2014. “Do High-Income or Low-Income Immigrants Leave Faster?” Journal of Development Economics 108:54–68.

11.

Bönisch

Gaffert

Wilde

. 2013. “The Impact of Skills on Remigration Flows.” Applied Economics 45(4):511–24.

12.

Borjas

G. J.

Bratsberg

. 1996. “Who Leaves? The Outmigration of the Foreign-Born.” The Review of Economics and Statistics 78(1):165–176.

13.

Carling

Pettersen

S. V.

. 2014. “Return Migration Intentions in the Integration–Transnationalism Matrix.” International Migration 52(6):13–30.

14.

Chabé-Ferret

Machado

Wahba

. 2018. “Remigration Intentions and Migrants’ Behavior.” Regional Science and Urban Economics 68:56–72.

15.

Chiquiar

Hanson

G. H.

. 2005. “International Migration, Self-Selection, and the Distribution of Wages: Evidence from Mexico and the United States.” Journal of Political Economy 113(2):239–281.

16.

Clark

A. E.

1997. “Job Satisfaction and Gender: Why Are Women So Happy at Work?” Labour Economics 4(4):341–72.

17.

Cobb-Clark

D.A.

1993. “Immigrant Selectivity and Wages: The Evidence for Women.” The American Economic Review 83(4):986–93.

18.

Cobb-Clark

D.A.

Stillman

. 2013. “Return Migration and the Age Profile of Retirement among Immigrants.” IZA Journal of Migration 2(1):1–20.

19.

Cohen

Haberfeld

. 2007. “Self-Selection and Earnings Assimilation: Immigrants from the Former Soviet Union in Israel and the United States.” Demography 44(3):649–68.

20.

Constant

Massey

D. S.

. 2002. “Return Migration by German Guestworkers: Neoclassical versus New Economic Theories.” International Migration 40(4):5–38.

21.

Constant

Massey

D. S.

. 2003. “Self-Selection, Earnings, and Out-Migration: A Longitudinal Study of Immigrants to Germany.” Journal of Population Economics 16(4):631–53.

22.

DaVanzo

J. S.

1983. “Repeat Migration in the United States: Who Moves Back and Who Moves On?” The Review of Economics and Statistics 65(4):552–59.

23.

de Haas

Fokkema

. 2011. “The Effects of Integration and Transnational Ties on International Return Migration Intentions.” Demographic Research 25:755–82.

24.

de Haas

Fokkema

Fassi Fihri

. 2015. “Return Migration as Failure or Success? The Determinants of Return Migration Intentions among Moroccan Migrants in Europe.” Journal of International Migration and Integration 16(2):415–29.

25.

Duleep

H. O.

1994. “Social Security and the Emigration of Immigrants.” Social Security Bulletin 57(1):37–52.

26.

Dustmann

2003. “Children and Return Migration.” Journal of Population Economics 16(4):815–30.

27.

Dustmann

Fabbri

. 2005. “Immigrants in the British Labour Market.” Fiscal Studies 26(4):423–70.

28.

Dustmann

Fabbri

Preston

Wadsworth

. 2003. “Labour Market Performance of Immigrants in the UK Labour Market.” Home Office Online Report 05 /03:1–81.

29.

Dustmann

Glitz

. 2011. “Migration and Education.” Handbook of the Economics of Education 4:327–439.

30.

Dustmann

Görlach

J.-S.

. 2016. “The Economics of Temporary Migrations.” Journal of Economic Literature 54(1):98–136.

31.

Dustmann

Kirchkamp

. 2002. “The Optimal Migration Duration and Activity Choice after Re-migration.” Journal of Development Economics 67(2):351–72.

32.

Dustmann

Weiss

. 2007. “Return Migration: Theory and Empirical Evidence from the UK.” British Journal of Industrial Relations 45(2):236–56.

33.

Edin

P.-A.

LaLonde

R.J.

Åslund

. 2000. “Emigration of Immigrants and Measures of Immigrant Assimilation: Evidence from Sweden.” Working Paper. Department of Economics, Uppsala University 13:1–38.

34.

Engzell

Ichou

. 2019, June 18. “Status Loss: The Burden of Positively Selected Immigrants.” International Migration Review (Online First):1–25.

35.

Feliciano

2005. “Educational Selectivity in U.S. Immigration: How Do Immigrants Compare to Those Left Behind?” Demography 42(1):131–52.

36.

Feliciano

Lanuza

Y. R.

. 2017. “An Immigrant Paradox? Contextual Attainment and Intergenerational Educational Mobility.” American Sociological Review 82(1):211–41.

37.

Flahaux

M.-L.

2017. “The Role of Migration Policy Changes in Europe for Return Migration to Senegal.” International Migration Review 51(4):868–92.

38.

Gmelch

1980. “Return Migration.” Annual Review of Anthropology 9(1):135–59.

39.

Güngör

N. D.

Tansel

. 2008. “Brain Drain from Turkey: An Investigation of Students’ Return Intentions.” Applied Economics 40(23):3069–87.

40.

Güveli

Ganzeboom

Baykara-Krumme

Platt

Eroğlu

Ş.

Spierings

Bayrakdar

Nauck

Sozeri

E. K.

. 2017. “2,000 Families: Identifying the Research Potential of an Origins-of-Migration Study.” Ethnic and Racial Studies 40(14):2558–76.

41.

Hanushek

E. A.

Woessmann

. 2012. “Do Better Schools Lead to More Growth? Cognitive Skills, Economic Outcomes, and Causation.” Journal of Economic Growth 17(4):267–321.

42.

Harris

J. R.

Todaro

M. P.

. 1970. “Migration, Unemployment and Development: A Two-Sector Analysis.” The American Economic Review 60(1):126–42.

43.

Ichou

2014. “Who They Were There: Immigrants’ Educational Selectivity and Their Children’s Educational Attainment.” European Sociological Review 30(6):750–65.

44.

Ichou

. 2018. Les enfants d’immigrés à l’école. Inégalités scolaires du primaire à l’enseignement supérieur. Paris: Presses universitaires de France.

45.

Ichou

Wallace

. 2019. “The Healthy Immigrant Effect: The Role of Educational Selectivity in the Good Health of Migrants.” Demographic Research 40(4):61–94.

46.

Jasso

Rosenzweig

M. R.

. 1986. “What’s in a Name? Country-of-Origin Influences on the Earnings of Immigrants in the United States.” Research in Human Capital and Development 4:75–106.

47.

Jensen

Pedersen

P. J.

. 2007. “To Stay or Not to Stay? Out-Migration of Immigrants from Denmark.” International Migration 45(5):87–113.

48.

Kırdar

M. G.

2009. “Labor Market Outcomes, Savings Accumulation, and Return Migration.” Labour Economics 16(4):418–28.

49.

Larramona

2013. “Espagne: l’émigration des immigrés.” Population 68(2):249–71.

50.

Levels

Dronkers

. 2008. “Educational Performance of Native and Immigrant Children from Various Countries of Origin.” Ethnic and Racial Studies 31(8):1404–25.

51.

Lewis

W. A.

1954. “Economic Development with Unlimited Supplies of Labour.” The Manchester School 22(2):139–91.

52.

Lucas

R. E. B.

Stark

. 1985. “Motivations to Remit: Evidence from Botswana.” Journal of Political Economy 93(5):901–18.

53.

Massey

D. S.

1987a. “The Ethnosurvey in Theory and Practice.” International Migration Review 21(4):1498–522.

54.

Massey

D. S.

. 1987b. “Understanding Mexican Migration to the United States.” American Journal of Sociology 92(6):1372–403.

55.

Massey

D. S.

Arango

Hugo

Kouaouci

Pellegrino

Taylor

J. E.

. 1993. “Theories of International Migration: A Review and Appraisal.” Population and Development Review 19(3):431–66.

56.

Massey

D. S.

Espinosa

K. E.

. 1997. “What’s Driving Mexico–US Migration? A Theoretical, Empirical, and Policy Analysis.” American Journal of Sociology 102(4):939–99.

57.

McKenzie

Rapoport

. 2010. “Self-Selection Patterns in Mexico–U.S. Migration: The Role of Migration Networks.” The Review of Economics and Statistics 92(4):811–21.

58.

Nekby

2006. “The Emigration of Immigrants, Return vs Onward Migration: Evidence from Sweden.” Journal of Population Economics 19(2):197–226.

59.

OECD (Organisation for Economic Co-operation and Development). 2008. International Migration Outlook 2008. Washington, DC and Paris: Organisation for Economic Co-operation and Development.

60.

ONS (Office for National Statistics). 2012. “International Migrants in England and Wales: 2011.” London: Office for National Statistics.

61.

Owen

2003. “The Demographic Characteristics of People from Minority Ethnic Groups in Britain.” In Explaining Ethnic Differences. Changing Patterns of Disadvantage in Britain, edited by Mason

, 21–52. Bristol: The Policy Press.

62.

Pedersen

P. J.

Pytlikova

Smith

. 2008. “Selection and Network Effects—Migration Flows into OECD Countries 1990–2000.” European Economic Review 52(7):1160–86.

63.

Phelan

1994. “The Paradox of the Contented Female Worker: An Assessment of Alternative Explanations.” Social Psychology Quarterly 57(2):95–107.

64.

Picot

Piraino

. 2013. “Immigrant Earnings Growth: Selection Bias or Real Progress?” Canadian Journal of Economics/Revue canadienne d’économique 46(4):1510–36.

65.

Piore

M. J.

1979. Birds of Passage: Migrant Labor and Industrial Societies. Cambridge: Cambridge University Press.

66.

Portes

1999. “Conclusion: Towards a New World-the Origins and Effects of Transnational Activities.” Ethnic and Racial Studies 22(2):463–77.

67.

Quinn

M. A.

Rubb

. 2005. “The Importance of Education-Occupation Matching in Migration Decisions.” Demography 42(1):153–67.

68.

Reagan

P. B.

Olsen

R. J.

. 2000. “You Can Go Home Again: Evidence from Longitudinal Data.” Demography 37(3):339–50.

69.

Rooth

D.-O.

Saarela

. 2007. “Selection in Migration and Return Migration: Evidence from Micro Data.” Economics Letters 94(1):90–95.

70.

Sabates-Wheeler

Taylor

Natali

. 2009. “Great Expectations and Reality Checks: The Role of Information in Mediating Migrants’ Experience of Return.” The European Journal of Development Research 21(5):752–71.

71.

Santelli

2001. La mobilité sociale dans l’immigration: itinéraires de réussite des enfants d’origine algérienne. Toulouse: Presses Universitaires du Mirail.

72.

Sayad

2004. The Suffering of the Immigrant. Cambridge, UK: Polity Press.

73.

Schiller

N. G.

Basch

Blanc

C. S.

. 1995. “From Immigrant to Transmigrant: Theorizing Transnational Migration.” Anthropological Quarterly 68(1):48–63.

74.

Sloane

P. J.

Williams

. 2000. “Job Satisfaction, Comparison Earnings, and Gender.” Labour 14(3):473–502.

75.

Stark

1991. The Migration of Labor. Oxford: Basil Blackwell.

76.

Toma

Castagnone

. 2015. “What Drives Onward Mobility within Europe? The Case of Senegalese Migrations between France, Italy and Spain.” Population 70(1):65–95.

77.

van Hook

Zhang

. 2011. “Who Stays? Who Goes? Selective Emigration among the Foreign-Born.” Population Research and Policy Review 30(1):1–24.

78.

van Hook

Zhang

Bean

F. D.

Passel

J. S.

. 2006. “Foreign-Born Emigration: A New Approach and Estimates Based on Matched CPS Files.” Demography 43(2):361–82.

79.

Villarreal

2016. “The Education-Occupation Mismatch of International and Internal Migrants in Mexico, 2005–2012.” Demography 53(3):865–83.

80.

Warren

Peck

J. M.

. 1980. “Foreign-Born Emigration from the United States: 1960 to 1970.” Demography 17(1):71–84.

81.

Zuccotti

C. V.

Ganzeboom

H. B.

Güveli

. 2017. “Has Migration Been Beneficial for Migrants and Their Children? Comparing Social Mobility of Turks in Western Europe, Turks in Turkey, and Western European Natives.” International Migration Review 51(1):97–126.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB