Planning to Go Back Home Because of the Economic Crisis? Evidence from the Short-Term Return Intentions of Bolivian Migrants in Spain

On the basis of data collected by the RETTRANS survey (2011–2012) on Bolivian migrants living in Spain, this article aims to identify the influence of the 2008 economic crisis on the determinants marking the migrants’ return intention. In this regard, there is a need to incorporate the concept of double embeddedness based on migrants’ social networks (meso level), the way migration is imbricated in persons’ lives (micro level); and how, at a macro level, economic and social transformations of different contexts affect these decisions. The article highlights that unexpected changes in living conditions trigger a reconfiguration of these mixed factors that condition return intentions.

On the basis of data collected by the RETTRANS survey (2011)(2012) on Bolivian migrants living in Spain, this article aims to identify the influence of the 2008 economic crisis on the determinants marking the migrants' return intention. In this regard, there is a need to incorporate the concept of double embeddedness based on migrants' social networks (meso level), the way migration is imbricated in persons' lives (micro level); and how, at a macro level, economic and social transformations of different contexts affect these decisions. The article highlights that unexpected changes in living conditions trigger a reconfiguration of these mixed factors that condition return intentions.
According to Sayad (2000), "return" is an inseparable element of the immigrant condition, mostly expressed as a myth and difficult to assign to a specific moment in time. Because of social, economic, and political constraints, return may become a desire rather than a reality that involves the mobilization of specific resources (Gmelch 1980). Therefore, the return intention may vary over time or may be postponed while immigrants await improved macroeconomic opportunities. These assumptions require a conceptual distinction between return migration (behavior) and intention to return (motivation). Although decisions and practices connected to return have been included in academic agendas, there is still a lack of research on intention motivations (Cassarino 2004;De Haas and Fokkema 2011).
Moreover, the debate regarding the main determinants of returning is still open. Based on data collected by the RETTRANS survey 1 (2011-2012) on Bolivian migrants living in Spain, this article aims to identify the influence of the economic crisis (macrostructural factors) on the main determinants marking migrants' return intention to Bolivia. From the neoclassical perspective, the return intention is the result of a rational assessment of benefits that migrants expect to achieve in the destination country. In contrast, theories focused on personal characteristics point out variables such as level of education, gender, family situation, and legal status as key factors explaining the probability to return (Curran and Rivero-Fuentes 2003;Agadjanian, Gorina, and Menjívar 2014;Ravuri 2014). Recent approaches highlight the role of transnational ties (Gmelch 1980;Carling and Bivand 2014) and the level of integration in the receiving country.
Although the literature shows several variables that explain the return intentions of migrants, they explain only partially the complexity of the phenomenon. In this regard, we argue for the need to understand return intentions based on the concept of double embeddedness (King 2002). The incorporation of double embeddedness into the study of return intentions allows for an understanding of return intentions that takes into account internal dynamics based on migrants' social networks (meso level) and the way that migration is imbricated in migrants' lives (micro level) (King 2002), and also the way that, at a macro level, economic and social transformations of societies affect these decisions. Regarding the specific case of Bolivian immigrants in Spain, the article highlights that unexpected changes in living conditions trigger a reconfiguration of these mixed factors that condition return intentions (meso-and micro-analytical levels).
The choice of Bolivia as a case study results from several reasons. First, the intensity of the immigrant flow from Bolivia to Spain must be highlighted, mainly in the two years before visa requirements were introduced on April 1, 2007 (Gadea, Benencia, and Quaranta 2009). This intensity of flows translated into high percentages of irregular statuses for migrants, which plunged migrants, especially men working in the construction sector, into an extremely vulnerable situation after the global economic downturn of 2008. This economic crisis also explains the increase in returnees to Bolivia from Spain. Despite the fact that some of these immigrants return to Bolivia through programs run by national and international organizations (e.g., International Organization of Migration, Spanish Labor Ministry, European Union, Bolivian government, nongovernmental organizations), most of them are "silent returns," as official statistics are unable to capture real return patterns and their extent (Parella and Petroff 2014).
With these elements as starting points, the article is structured as follows. The first section reviews the main theoretical perspectives that focus on return intention and concludes by highlighting this article's key hypothesis on the extent to which the economic crisis explains migrants' return intention by triggering a reconfiguration of classical determinants. Second, we present Bolivian migration patterns in the international and Spanish context in recent decades. We then outline the methodological approach, followed by a bivariate analysis and regression model. The article ends with a section of final remarks and conclusions.

An Overall Review of the Return Intention and Its Main Determinants
The neoclassical perspective explains return intention as a result of a rational assessment of benefits (e.g., economy, human capital, social capital) that migrants expect to obtain in the destination country. Van Baalen and Müller (2008) estimate the determinants of (not) revising the intended duration of stay, and they observe the dynamic inconsistency in the evolution of return intention. The proposed explanation for this inconsistency is that migrants tend to be too optimistic about their future saving, which is why they have to revise their plans and to postpone their return decision, to the extent that they are not able to achieve their initial goals (Van Baalen and Müller 2008).
Another set of theories highlights the importance of personal characteristics as key elements explaining return intention. Although there is no consensus in the existing research concerning the type of effect that educational level has on return intention, some authors conclude that a high level of education reduces the probability of manifesting return intention (DaVanzo and Morrison 1981;Massey and Espinosa 1997;Curran and Rivero-Fuentes 2003). Others, however, show a positive relationship between these variables (Constant and Massey 2003). Although migrants' legal status is a key measure of their success and security, little is known about how this variable affects return migration (Agadjanian, Gorina, and Menjívar 2014).
Gender and family also play a significant role in return intentions. The well-being of children operates both as a constraint and as an incentive to return (Parreñas 2005), and women are more likely to return (Ravuri 2014). Care responsibilities toward children and relatives act as strong social pressures for migrant women (Parreñas 2005).
From a transnational perspective, Gmelch (1980) concludes that contact with the home country, particularly transnational economic practices (De Haas and Fokkema 2011), and visits (Carling and Bivand 2014) are positively related to return intention. Nevertheless, the link between return intentions and the geographic location of family members remains unclear in its effect. According to Agadjanian, Gorina, and Menjívar (2014), neither having a marital partner nor having a child in the country of origin has any effect on the likelihood of reporting firm plans to return.
Finally, assimilationist or integrationist theories focus on time-dependent variables and social links with the destination country. In this regard, a longer stay diminishes the likelihood of returning, even controlling for a set of demographic factors (Jensen and Pedersen 2007). Moreover, Jensen and Pedersen (2007) point out the importance of variables measuring labor-market attachment, as unemployment increases the probability of returning (Constant and Massey 2003). In contrast, the establishment of an ethnic community in the destination country and the existence of strong social ties are two elements that explain to a large extent a low intention to return (Agadjanian, Gorina, and Menjívar 2014).
Although there is very little research that simultaneously incorporates integration processes and transnational involvement, some scholars highlight the interplay of individual decisions and broader collective pressure. In this sense, Mortensen (2014) stresses that issues connected to return are partly matters of identity; for Ley and Kobayashi (2005), return is a strategic switching between unified social fields that transcend national borders. Aiming to incorporate both approaches, Carling and Pettersen (2014) employ a theoretical framework to evaluate how integration and transnationalism relate to each other and in turn shape return intentions. The results show that return intentions are conditioned by people's attachment to both their country of residence and their country of origin, and that it is not possible to identify a zero-sum game of attachment among migrants.
The wide-ranging theoretical perspectives described in this section discuss a large set of variables that operate as drivers for return intentions. None of them is able to explain by itself the complex decision to return, and all agree that there is no single element that triggers the decision to return. Instead, we are dealing with a mixture of motivations linked to microstructural elements such as personal features (e.g., gender, life course, level of education, legal situation), mesostructural elements such as integration variables (e.g., cultural or labor attachment, length of stay), and transnational ties (e.g., transnational social networks, transnational practices, contacts).
Despite this assumption, we bring into discussion the concept of double embeddedness (King 2002), assuming that the studies on return intentions should take into account the interplay generated by different analytical levels (integrating human agency with state and other structures). More specifically, we argue that, in the case of Bolivian community in Spain, macrostructural elements (e.g., significant changes in labormarket conditions in the destination country) displace the mixture of micro-and meso-level factors and reconfigure them, by establishing new priorities in terms of return intention. Therefore, we expect that the global economic and employment crisis that especially affected Southern European countries in 2008 not only boosted return intentions but also changed the significance of traditional variables linked to personal features or transnational ties. We make the following five proposals.
Massive unemployment rates in the construction sector in Spain places Bolivian men in a more vulnerable position than women, making them more likely to express a return intention. Irregular migrant status (common among Bolivian migrants in Spain), considered jointly with the negative impact of the economic crisis on employment, leads to a higher probability to admit return intention. In contrast to before the crisis, most of these Bolivian migrants have been left out of the underground economy. Since 2012, although many migrants adopted a wait-and-see approach, hoping for improvement in the labor market, available indicators have shown a significant increase in return migration, particularly given the length and severity of the crisis, and especially among irregular migrants (Parella and Petroff 2014). In this context, undocumented migrants have probably stopped aspiring to regularize their legal situation at some point in the future through a work contract (i.e., the social integration residence permit).
Regarding level of education, we expect an ambivalent effect. Although the literature shows that the likelihood of returning is lower among people with tertiary education, for Bolivian migrants in Spain we expect a different trend. The deterioration of life and working conditions in the Spanish labor market is likely to discourage the idea of remaining in Spain in order to find better job prospects in the short or medium terms. In fact, in the European Union, third-country nationals are often overqualified, and this problem worsened with the economic crisis. This feature is particularly common in countries like Spain because of the characteristics of its local labor market.
As the literature points out, both having children in, and developing economic links with, the country of origin are positively linked to return intention. Nevertheless, in a context of crisis, this transnational scenario becomes a trigger for returning home, giving these elements, always present in migrants' projections, higher priority.
Although according to the literature, return intention declines as a consequence of a longer stay and also social and labor establishment in the destination country, in the case of Bolivian migrants in Spain, this may not be the case. Predictors associated with the negative impact of the economic crisis could exercise a stronger influence in the same direction as transnational ties, and consequently could undo the effect of integration factors. Therefore, we expect that migrants who perceive that they have been more adversely affected by the crisis in economic terms will also tend to express strong intentions to return to Bolivia.

Bolivian Migration in the International and Spanish Contexts
Bolivian migration is an exceptional phenomenon for a variety of reasons (Lafleur and Duchesne 2017). From the beginning of the twentieth century, Bolivia struggled with a deep economic, social, and political crisis. Poverty and corruption are prominent in Bolivia (Hummel 2018). In such a hostile context, internal and cross-border migration have become common practices, leading to a so-called culture of mobility (De la Torre 2006; Hinojosa 2009). According to the Bolivian Census and International Organization for Migration estimates, in 2012, between 1.4 million and 2.5 million Bolivian lived abroad; the country has a total population of only 9 million (Hinojosa and de la Torre 2014).
The chosen destination of Bolivian migrants has changed over time. Traditionally, Argentina and Brazil attracted Bolivian immigrants for their labor opportunities and geographical proximity (Tapia 2014). During the 1950s, Bolivian migrants to Argentina were typically peasant males, with low levels of education, who found employment in the agriculture sector. During the 1960s and 1970s, cities such as Buenos Aires also started to attract Bolivian citizens to employment in precarious, informal jobs (Sassone and Cortes 2014).
Nevertheless, during the 1980s and 1990s there was a shift in the tendency of Bolivian migration: intraregional movements were replaced by extraregional flows, the United States being one of the main destination countries. As in the case of Argentina, the flows toward the United States were also largely composed of low-skilled men who regularly returned to their communities of origin (De la Torre 2006).
Another turning point in migration from Bolivia took place in 2000, around the time of Argentina's economic crisis at the end of the 1990s. This event led to new destinations for migrants: the manufacturing industry in Brazil, as well as Europe, especially Spain and Italy (Gadea, Benencia, and Quaranta 2009;Lafleur and Duchesne 2017). For Spain and Italy, the arrival of these immigrants operate in a context of rising demand for low-skilled labor in certain sectors, mainly domestic services and construction (Gadea, Benencia, and Quaranta 2009). According    (MEYSS, 2009(MEYSS, -2016. The second program was much less successful; 134 out of 918 applicants for the Program on Productive Voluntary Return were Bolivians (MEYSS, 2010(MEYSS, -2016. Finally, through the program APRE (Abono Acumulado y Anticipado para Extranjeros de la Prestación Contributiva) only 280 immigrants returned to Bolivia out of 11,930 applicants (MEYSS, 2011(MEYSS, -2016. 3 The rule, which prohibits returning to Spain for three years after the initial return to Bolivia, as well as the lack of monitoring measures after return largely explain the low number of migrants who have returned using the Spanish government's programs (Parella and Petroff 2014).

Methodology Data and Sample
This article presents analysis of data gathered in the framework of the research RETTRANS (Return from a Transnational Perspective) funded by the Spanish Research and Innovation Council (CSO2010-15924). More specifically, the findings are based on the RETTRANS survey, focused on the return intentions of Bolivian immigrants residing in Madrid, Barcelona, and L'Hospitalet de Llobregat, the main residential enclaves of Bolivian migrants in Spain (INE 2011). The representative sample included four hundred cases corresponding to people born in Bolivia, between the ages of eighteen and sixty-five, and with a minimum of twelve months' residence in Spain. A stratified sample was computed by three criteria based on data provided by the Municipal Register of Inhabitants (Padrón Municipal 2010): sex, age, and district of residence. The survey, conducted during the fourth quarter of 2011 and the first quarter of 2012, included a questionnaire with 101 questions. The questions were organized into thematic blocks as follows: (1) sociodemographic characteristics of the immigrant and his or her family; (2) the main features of the migratory trajectory; (3) economic and legal integration in Spain; (4) perception of the set of opportunities in Spain and Bolivia; (5) involvement in transnational practices, both economic (e.g., remittances, investment, business visits) and social (e.g., family responsibilities, visits during holidays); and (6) degree of settlement in Spain.

Measurement
To measure return intention, the following closed question was included in the questionnaire: "What are your plans for the following twelve months?" The available response options were return to Bolivia permanently; return to Bolivia temporarily; remain in Madrid, Barcelona, or L'Hospitalet de Llobregat; move to another Spanish city; and move to another country. A dichotomous variable was created as a result of this question-short-term return intention to Bolivia-and used as a dependent variable in the analysis. This variable clusters 35.5 percent of the sample and is constituted by all those who reported an intention to return to Bolivia either temporarily or permanently (received a value of 1), distinguishing them from the rest of the sample (value of 0).
For the purpose of testing the hypotheses on the determinants of return intention, we organized the predictors into three blocks that were incorporated into the bivariate and multivariate analyses in a stepwise manner. These blocks were as follows: • Migrants' socioeconomic characteristics: gender, marital status, year of arrival, legal status and educational level (a dichotomous variable measuring whether the migrant has tertiary education or not) • Transnational ties: having children in Bolivia, sending remittances, and the identity of the person who manages these remittances in Bolivia (i.e., "the migrant," "migrant's partner," "parents," "children" or "other"), as well as whether migrants kept themselves informed of the Bolivian economic and politic situation • The impact of the economic crisis on living conditions: migrants' perception of their own situation in Spain; whether the migrant delayed a mortgage or rent payment because of the crisis; and whether he or she is interested in applying for humanitarian voluntary return programs offered by the Spanish government. The last variable is employment status, as we are interested in distinguishing those who reported being unemployed and those who are employed. In this regard, we differentiate between two status-in-employment categories: those who are employed or self-employed and affiliated with social security, and those who are not.

Discussion
The first part of the discussion presents the data on return intention in the short term, based on the coded responses given by the interviewees, and shows the extent to which these plans, according to respondents, are connected with the impact of the Spanish economic crisis. We have included a bivariate analysis showing some characteristics of the sample of those who manifested return intention. For descriptive results, appendix 1 shows the association between the key predictors or correlates and their outcomes in terms of percentage distributions across cross-tabulations as well as the chi-square tests of independence. The second part of the discussion presents the results of the multivariable logistic regression, based on a set of predictors and relevant theoretical variables that allow for empirical contrast of the hypothesis. We enter various blocks of predictors into the multivariate analyses in a stepwise manner, adding the blocks of predictors sequentially across the models. To address the issue of missing data, listwise deletion was performed on all variables. To the closed question "What are your plans for the next twelve months?" (Table 1), the results show that 32.5 percent of the sample planned to return to Bolivia, whether temporarily (20 percent) or permanently (12.5 percent). In contrast, 57.8 percent of the sample reported planning to remain in Spain, and almost 5 percent were planning an onward migratory project (mainly to Argentina or Chile). The economic crisis in Spain accounts for 82.3 percent of those who plan to return temporarily and 90 percent of those whose intention is to return permanently. This suggests the everyday engagement of individuals' decisions with different structural environments in order to explain return intentions (King 2000). Literature shows that the percentage of return intention in other migrant communities is quite similar. An online survey of Romanian health professionals abroad in 2010 shows how, in the context of crisis, 24 percent of the sample was willing to return at some point to Romania (Roman and Goschin 2014). Another study, conducted with Mexican immigrants residing in the United States, found that 29 percent of the sample expressed an intention to return to Mexico (Tezcan 2018).
A bivariate analysis is presented in appendix 1 and gives the main variables used in the multivariate analysis. For the bivariate statistics, we present the association between key predictors and outcomes in terms of percentage distributions, as well as the chi-square tests of significance. In terms of sex distribution, the analysis shows that the Bolivian community in Spain skews toward more women, as six out of ten of respondents are women. Despite this, men show higher probabilities of manifesting return intention. Another sociodemographic feature, level of education, reveals that just over half of the overall sample (56.5 percent) has a secondary education. Nevertheless, this variable is not significant in terms of predicting return intention. The analysis reveals that 56.5 percent of respondents are married or living with a partner. Marital status is a significant variable in relation to return intention. Therefore, being single plays a significant role in migrants' intentions to remain in Spain.
Migrants' employment situation also correlates with return intention. More specifically, those who are unemployed are more likely to express a return intention. It is also relevant that 46 percent of the overall sample is self-employed or employed and affiliated with social security.
Concerning legal status, when focusing on the distribution of the overall sample, the biggest category among Bolivian immigrants is those with temporary residence permits (49 percent). Those who only possess a Bolivian passport (irregular status) represent 13.3 percent. Legal status is a significant predictor of return intention, although to a lesser extent, the probability increases for migrants with an EU card or permanent residence permit. A possible explanation could be the flexibility inherent in developing circular migratory patterns if the return to Bolivia turns out to be unsuccessful. Finally, the likelihood of expressing return intentions in the case of those immigrants with the most vulnerable status (a pending residence permit or passport only) is also significant: nearly four out of ten who reported this legal situation also reported wanting to return. In the case of migrants with temporary residence permits, the expectation of a permanent residence permit may inhibit return intention.
Year of arrival was included in the analyses, as the literature identifies it as a strong predictor for return intention. Nevertheless, as expected, this variable in the case of Bolivians does not have a significant effect, given the recent arrival of this immigrant community in Spain.
When focusing on the second block of variables, transnational ties, although 56 percent of the overall sample declared that they do not have children in Bolivia, the analysis reveals that having children in Bolivia is a strong predictor of return intention. In fact, this finding reinforces the relevance of family reasons given by respondents when asked about their motivations to return. Second, staying informed about the situation in Bolivia is also significant for those who manifest return intention. In contrast, the fact of having sent remittances in the previous year does not appear to affect return intention, although in the case of both studied samples (those who manifest return intention and the overall sample), a large number of respondents report having sent remittances (around 65 percent). Although sending remittances does not affect return intention, the person who manages these remittances reveals a strong correlation with the decision to return. Parents in Bolivia are typically those who manage the remittances of Bolivian immigrants, and at the same time, the parents' role is statistically associated with a higher probability of return intentions. As pointed out by other studies, although migration requires separation, it is part of a strategy for maintaining a "home" and represents a form of caring for both people and property (Fouratt 2017). Table 2 shows the results of regression models for short-term return intentions. This analysis enables us to measure the effect of each independent variable on the likelihood of intending to return to Bolivia by controlling for all other variables in the model (Carling and Erdal 2014). Considering the nature of the dependent variable, the categories used in the multivariate logistic regression analysis are "with return intention in the following twelve months" as opposed to the other categories (see Table 1) The predictors were incorporated progressively into the regression model in three stages (step-by-step). Both the first and second step include predictors for which there is a theoretical justification for causal relation between variables. The third step considers whether economic recession has any influence on the economic situation of migrants after controlling for the variables introduced in the previous steps. The global financial crisis of 2008, given its destructive impact on employment, is considered a key incentive to return, especially for migrants with precarious economic status in Spain.
In the first step, the following objective predictors were considered dummy variables: sex, level of education, marital status, and legal status. For legal status, we grouped together all categories related to nonpermanent status (excluding the permanent residence permit and Spanish citizenship). Number of years migrants resided in Spain was omitted in the analysis, as the data do not reveal a significant association with return intentions. The block referring to transnational ties was included in the second stage as dummy variables. In terms of remittances, we have included two categories: parents (the relatives who most commonly manage them) and others (including those migrants who do not send remittances regularly).
Variables concerning migrants' economic situation and the subjective perceptions of this issue were incorporated in the last step. Two variables refer to migrants' current economic situation: whether they are affiliated with social security (the latter value includes both unemployed people and those with regular or sporadic incomes proceeding from the informal economy) and whether they experienced delay in being able to pay for housing (mortgage or rent). In terms of subjective dimensions, we have included migrants' perception of the economic situation together with their interest in humanitarian return programs. Some variables (e.g., "keeping informed about home country") do not give rise to obvious causal links from a theoretical perspective, as it is impossible to determine whether they predict return intention or represent the consequences of it.
For the first block of predictors, only two of the sociodemographic indicators significantly predict return intention, mostly in the same direction as proposed in our hypotheses: female and single migrants are less likely to return to their native country. According to this finding, it is more likely that men plan to return to their countries of origin, which is consistent with the data on returning migrants from Residential Variation Statistics (Parella and Petroff 2014) and with the set of theories developed by Ravuri (2014). This likelihood of return intentions for men is linked to diminished employment opportunities in key sectors of the labor market, namely the construction sector. In addition, return migration is less prevalent for unmarried people. As pointed out by Parreñas (2005), not having care responsibilities decreases the probability of a return intention, whereas having children in Bolivia increases the probability. This finding reinforces the contributions of others that explain the return through the existence of affective and family dynamics that mark this decision, especially in the case of women (Vega 2014;Martínez-Buján 2015). Finally, in contrast to results from the bivariate analysis, variables related to legal status do not appear to have a significant effect on return intention.
Regarding transnational practices, having children in Bolivia is a significant, very strong predictor of return intention. Those who have children back in Bolivia are themselves almost ten times more likely to return. As pointed out by Vega and Martínez-Buján (2016), the altered labor market as a result of economic crisis explains to a great extent the return decision of migrants. Nevertheless, other factors linked to the life course and family life as well as the tensions in the organization of transnational care must be taken into consideration.
In addition, when migrants keep informed about, and receive news from, their home country (through newspapers, internet, television, or contact with family members and friends), this is also positively associated with return intention. As also pointed out by King (2000), noneconomic factors affect return intentions to the same extent as economic ones.
Surprisingly, the inverse association of sending remittances during the previous year (versus not sending) and return intention is statistically significant. Although the literature emphasizes that migrants with a return intention are more likely to remit regularly, in this case, the negative association may indicate other contexts and circumstances. In times of financial hardship, when Bolivian migrants have less access to income and employment, this vulnerable situation negatively affects the decision to remit and increases the probability of returning home. Equally, if we consider limited access to economic resources, it is more likely that migrants use the small amount of money they do have to prepare for returning rather than remit to the country of origin.
Finally, those who send remittances managed by their parents tend to exhibit a greater probability of returning than those who do not send remittances or those who send remittances managed by other people (migrants themselves, spouses, children, or other family members). This is a very strong correlate, more so than any other indicator related to transnational ties. The statistical significance of this variable indicates that migrants' parents are a better mechanism for ensuring that the money will yield the expected income, especially if these remittances become investments in Bolivia (e.g., small business, building a house). Once variables linked to transnational ties are introduced, level of education appears to have a significant effect; tertiary education correlates negatively with return intention, as also found by DaVanzo and Morrison (1981), Massey and Espinoza (1997), and Curran and Rivero-Fuentes (2003). In this regard, those who manage educational resources probably have access to other kinds of capital that place them in a privileged situation in the Spanish labor market despite economic crisis.
Turning to the final block of predictors, which pertain to how the economic crisis has affected migrants' lives, the results corroborate our hypotheses. In fact, the model's explanatory capacity increases from 15.5 percent to 24.6 percent. Having an unfavorable employment status (e.g., part-time job, being unemployed or employed in the informal economy) or having economic difficulties in paying for housing appear to be unrelated to return intention. In contrast, subjective perceptions have a significant effect on return intention. Our findings show that when migrants perceive their economic situation in Spain as "bad" or "very bad," and when they are interested in applying for humanitarian return programs, they are more likely to express return intentions. Marital status and transnational ties continue to emerge as the main determinants. However, when other economic predictors are introduced, sending remittances, as expected, ceases to be a significant variable.

Conclusion
The current article was written with two aims in mind. First, to identify the short-term return intentions of Bolivians immigrants residing in Spain in light of the 2008-2009 global economic crisis and its effects in Spain. Second, to examine the extent to which these unexpected changes relate to the macro-analytical level and cause a reconfiguration of traditional factors linked to return, such as personal characteristics or transnational links (meso-and microanalytical levels).
As our analysis reveals, the profile of Bolivian immigrants wanting to return during the Spanish economic crisis is as follows: a man, with children in Bolivia, who keeps informed of and receives news from Bolivia, whose parents manage his remittances, who perceives his economic situation in Spain as "bad" or "very bad," and who is interested in applying for humanitarian return programs. Additionally, the model highlights how being female, single, having tertiary education, or not sending remittances all diminish the propensity to return. In this regard, findings confirm the effect of some determinants identified by the literature as positively linked to return intention (e.g., unemployment rate, having children in the home country, forming economic links with the country of origin). Nevertheless, empirical data also show the involvement of factors (e.g., legal status, length of stay, level of education) that capture the specificities of both the Spanish case, in a context marked by the global economic crisis, and the characteristics of the Bolivian community in Spain.
These findings emphasize the complexity inherent in dealing with return intention. As far as structureagency mechanisms are concerned, the implications of our findings deserve particular attention. Based on the assumption of King (2002), we have stressed the need to appeal to the double embeddedness of the return intention process. Double embeddedness refers, on the one hand, to internal dynamics based on migrants' social networks (meso level) and the way that migration is imbricated in migrants' lives (micro level) (King 2002), and, on the other hand, to the way that, at a macro level, economic and social transformations affect these decisions (Castles and Miller 2009, 47). The analysis of return intentions in the case of Bolivian immigrants willing to return after the downturn of the economy in Spain shows this double embeddedness.
On the role played by macrostructural factors, the economic and labor crisis in Spain has triggered the return home for many, especially men, whose unemployment rates were especially affected during the crisis. Nevertheless, factors linked to life cycle (e.g., having family in Bolivia) are also relevant, especially for women, who suffer more pressure regarding their socially constructed role as caregivers. In this regard, the unexpected changes in the set of opportunities caused by the crisis could represent not so much a change of plans concerning return intentions, but the need to anticipate them because of a reconfiguration of priorities.
The inclusion of the meso-analytical level to explain return intentions highlights the interplay of transnational ties (Gmelch 1980;Carling and Bivand 2014) and level of integration in the receiving country. One of the contributions of this article is to integrate return intention into a broader approach, showing how human agency with regard to return intentions evolves in interactive responses to the problems posed by changing situations. This leads to a reconfiguration of factors associated with the micro, meso, and macro levels. Therefore, the severe deterioration of living conditions in Spain as a result of the employment crisis is not necessarily a driver of return, but it does assign new meanings to meso and micro variables. Especially in the case of women, the return decision does not necessarily respond to the economic crisis. The reassessment of maternal responsibilities (embedded into cultural frameworks) and the expectations of the communities in Bolivia explain return intentions to great extent. Once the remittances decrease and the initial reasons to migrate lose weight, the pressure to return also increases.
Studies on return intention should integrate human agency with state and other structures (Massey et al. 1998, 281), taking into account the interplay of different analytical levels. Although the economic crisis and the lack of job opportunities were the main explanations for return intention in the case studied here, they also play a significant role in the reconfiguration of personal characteristics, transnational ties, and degree of integration in the host country, which end up determining return intentions. We stress the relevance of generating new quantitative post-crisis data in similar contexts and with different immigrant communities. Taking into consideration the high number of Bolivians who obtained Spanish citizenship in 2012, we must consider the potential reversibility of these return patterns, as recent research points out an increase of onwards migrations of this immigrant community to Spain and other European countries (Parella et al. 2017). As highlighted by Prunier (2017), the global migratory systems allow understanding the return flows as reversible. Therefore, new opportunities in the European context or traumatic experiences faced during the post-return and reintegration, may explain the increase of those returnees who decide to migrate once again.