The role of returned migrants in the perpetuation of the Brazil- US - - PDF document

The role of returned migrants in the perpetuation of the Brazil- US migration system during a period of economic stress

Leonardo Sousa1 Gilvan Ramalho Guedes2 Alisson Barbieri3 Dimitri Fazito4 Abstract During the 2000’s Brazil and the United States have undergone economic shocks (positive and negative). Historical migration flows from the Brazilian municipality of Governador Valadares (GV) towards the US, however, remained unaffected. A key factor explaining the stability of this migration system is the role played by the return migrant through social and informational support. Brazil now faces a deep recession, promoting incentives for the increase in out-migration flows, but the US election outcome and its prospect for the US migration policies work as a counteracting force for potential international migrants. Using data on ego-centered networks of migrants, return migrants, and non-migrants for the main sender of international migrants to the US in Brazil, collected in 2006 and 2016, this paper analyzes the role of return migrants in the perpetuation of the Brazil-US migration system during a period of economic and institutional shocks. Our preliminary analysis reveal the important role of the return migrant in the perpetuation of this flow by providing or facilitating the exchange of strategic material and informational resources in the sending community. The 2016 data suggests that migratory social networks remain active and provide intensive material and immaterial exchange among their peers, acting as a maximizer for new, voluntary displacements in the face of the current financial crisis in Brazil. Keywords: International Migration, Returned Migrants, Social Network Analysis, Brazil-US Migration System

1 Phd in Demography, Center for Development and Regional Planning – Cedeplar/ Federal University of

Minas Gerais – UFMG, and Professor at Pitágoras College. leogosousa@gmail.com

2 Core Faculty at Cedeplar and Associate Professor at the Demography Department, UFMG, Brazil. 3 Core Faculty at Cedeplar and Associate Professor at the Demography Department, UFMG, Brazil. 4 Associate Professor at the Sociology Department – Federal University of Minas Gerais.

Introduction

The 2000’s was a decade marked by an increasing surveillance for illegal immigrants coupled with the mortgage crisis in the United States. The result was an increase in the number of return migrants and a significant decline in remittances, with non-negligible impact for many sending communities around the world. The scenario in Governador Valadares (GV), the main Brazilian municipality in terms of international out-migration to the United States (US), was not different (Sousa and Fazito, 2016; Fazito, 2005). According to the institutional migration framework, international migration flows in this environment should be strongly reduced, with likely consequences for the stability of the established migration system (de Haas, 2010). Despite the increased difficulty for illegal immigrants and the economic crisis in the US on the one hand, and the advances in the Brazilian economy and employment opportunities during the 2000’s on the other hand, migration from GV to the US remained virtually unaltered (Sousa and Fazito, 2016). We argue that this apparent paradox is partially explained by the unique role played by the international return migrant who, through his/her personal network, provided social, material, and informational support to the prospect migrants, contributing to the stability of the historically established migration system. Since 2015, Brazil has undergone the deepest economic and political crisis in the last

• century. The institutional and economic instability has promoted social unrest, civil conflicts,

and unemployment throughout the country, increasing the stimulus for both legal and illegal international out-migration. Using data collected in 2006 and 2016, this paper analyzes the likely differences in the network structure and activity of GV over time. More specifically, we hypothesize that the propensity to out-migrate have increased recently due to the institutional and economic crisis in Brazil, but international destinations might have become more diverse as migration laws in the US after the election outcome will likely impose

additional barriers to illegal immigration. The role of return migrants, however, might have intensified from 2006 to 2016, working as a counteracting force to the consequences of the election outcome for those prospect migrants who have the US as their main destination of

• interest. In this circumstance, we might expect not only an intensification in the social

support operated by the return migrant, but also a likely change in the type of resource channeled through his/her ego-centered network.

Migration as a System

Theoretically, migration can be understood as a social (Massey et al., 1987) and collective process (Sayad, 2000), sustained and operationalized by an integrated system of structures such as social networks, market institutions, psychosocial attributes (representations, beliefs, norms, and values), and strategic material and informational resources. These structures and their links define the territorial, economic, and motivational space for the migration system (Fawcett, 1989; Kritz and Zlotnik, 1992; Fazito, 2005). The modeling of migration networks represents the core issue in the operationalization of a migration system, defined and expressed by the social and institutional interactions among the many actors in the migration process. These networks of migrants are made of a set of interpersonal ties connecting individuals with migration experience (both current and past) to themselves and to other actors in the sending and destination communities (Fazito, 2010). These ties are mostly established by kinship, friendship, or shared interests of many sorts (Massey et al.,1993), and hold a rich amount of diverse flows that contribute to a privileged visualization

• f the migration system (Wasserman and Faust, 1994; Fazito; 2005; Fazito, 2010).

Potential migrants are expected to have different ties with qualified actors of migration through direct and indirect contacts embedded in the social structure. Migration is also a

consequence of cumulative causation, where repeated migration experiences reinforce and expand migrants’ social capital in both places – origin and destination communities. This expanded social capital would reflect a wider pool of direct and indirect contacts, as well as a larger number of strong and weak ties, crossing the limited space of communities and linking individuals through social bonding (Massey, 1990; Massey et al., 1993). Although the concept of social capital is very encompassing, its concept for migration networks translates into a set of strategic resources (contacts) provided by the migrant’s personal network. These resources facilitate the concretization of the migration project by the provision of material (money, documents, information), instrumental (transportation, communication, falsification), or expressive (physical and psychological security) support (Lin, 2001). Thus, the structural positions available and effectively occupied by migrants and non-migrants in a network with local (here or there) and global (here and there) dimensions provide the strategic resources that may facilitate or constrain decisions regarding the migration project. In this scenario, the migration develops upon the daily relations between the many actors strategically positioned in the social structure (Fazito and Soares, 2010). Little explored in the social network literature of migration, the return migration holds all the positional structures to facilitate the flow of resources in the community. Different from other social actors, the return migrant occupies a privileged position in the migration network as he/she is able to mediate resources in the local (here) and global (here and there) network, having a privileged access to the intervening mechanisms of the legal or illegal migration

• process. This actor may be of different bonding nature, from a formal broker to a family

member reinforcing the transnational family link (Lin, 2001). Studies on the perpetuation of migration systems must then pay special attention to how return migrants operate their social capital within the system because of their unique ability to mediate resources across space at different levels of aggregation (Massey et al., 1993; Kritz and Zlotnik, 1992).

Method and Materials

In order to collect the relational data of individuals and members of their network in the GV area, we conducted fieldwork data collection in 2006 and 2016 using the personal networks approach (McCarty, 2002). Personal networks can be understood as an analytical model capable of revealing the pillars of community interactions, where people develop sociability and create niches of shared interests through subcommunities of specific interests (Wellman, 1988). In 2006, 50 interviews were conducted with migrants who had lived in the United States and returned to the GV region and did not move out again for at least six months. In 2016, 100 interviews were conducted. It is important to note that the respondents of the two surveys may be the same in some cases. However, this study is not essentially longitudinal and rather focuses on the analysis of the migratory network activity in the region. All interviews were face-to-face, conducted by the authors of this study and graduate students after intensive fieldwork training. The returned migrants answered individual sociodemographic questions (ego), questions about the members of their personal network (alteri), and the closeness and intensity of the links between the migrant and his/her network members (ego- alteri), as well as among the members themselves (alteri-alteri). The interviewees living in GV area at the time of the interview were taken as an observation unit and their personal networks as the unit of analysis. The dataframe was designed with three basic linked structures. The first contains the attribute data of the return migrants and some measurements of their personal network. The second has the attribute and relational data on all alteri that each ego informed. Each ego was invited to report the 40 closest alteri, totaling 2000 alteri in the full 50-interviews database for the 2006 data and 4000 alteri in the full 100-interviews database for the 2016 data. Finally, adjacency matrices (that is, the matrix

• f the dyads) were created containing the weighted relations between the alteri of each

personal network (a 40x40 adjacency matrix for each ego). The data were cataloged using the software Egonet (2006 data) and EgoWeb (2016 data), after which the attribute and relationship data were extracted. The attribute data were processed in the R software to generate the descriptive statistics at the ego level. The relational data, containing the ego- alteri and the alteri-alteri valued (weighted) matrices, were modeled in the Gephi software for metrics measurement and visualization. Respondents were asked to sign an informed consent with a copy handed to them and another filed for the research team. All protocols established by the Ethics in Research of the home institution of the research team were strictly followed, based on the instrument versions approved by the Ethics in Research Committee of the Universidade Federal de Minas Gerais (CAAE 55007116.7.1001.5149). The research project encompassing the data collected for this study was funded by the Brazilian Research Council (CNPq Process 431872/2016-3), the Minas Gerais Research Foundation (FAPEMIG Process CSA - APQ- 01553-16) and the Brazilian Research Network on Climate Change (FINEP / CNPq).

Results and Discussion

The descriptive analysis of the 2006 and 2016 sample data are shown in Table 1. Most respondents in the 2006 dataset were male (56%) and over 30 years old (88%). About 57% had up the elementary school completed, while the remainder were divided among those who completed high school (32%) and Higher education (10%). Most respondents (66%) stated that they had legally out-migrated with some type of visa (tourism, work, or student), while 17 reported having entered the US illegally, either by crossing the border (26%) or using false documentation (8%).

TABLE 1 – Proportions of attribute variables among the returned migrants in the GV Study, 2006 and 2016. Variables 2006 2016 N % N % Sex Male 28 56 52 52 Female 22 44 48 48 Education Low 21 42 30 30 Medium 19 38 56 56 High 10 20 14 14 Marital Status Single 26 51 26 27 Married 17 34 60 59 Other (widowed/divorced) 7 14 14 13 Race/Ethnicity White 26 51 24 24 Non-White 24 47 76 76 Residence Time in the USA Up to 1 year 10 20 10 10 From 1 to 4 years 17 34 40 40 From 5 to 10 years 8 16 30 30 Above 10 years 15 30 20 20 Migration Strategy Legal Visa 33 66 46 46 Falsified Visa 4 8 6 6 Irregular Border Crossing 13 26 48 48

Source: Primary data from fieldwork in Governador Valadares, Brazil (2006, 2016)

Both samples show very similar sociodemographic composition of respondents, except that the 2016 sample has a larger share of married and non-white. The migration strategy is also different; while in 2006 only 34% migrated illegally to the United States (falsified visa + irregular border crossing), in 2016 the proportion was as high as 54%. This very result is suggestive of likely pull factors in Brazil (the 2010-2014 boom period) and push factors in the US, as rules became stricter for the remain of undocumented migrants in the latter. Table 2 shows descriptive statistics derived from the ego-centered networks of the returned migrants in the GV area. The table is broken into four blocks: (1) network measures, describing its average topological configuration; (2) ego-alteri proximity, summarizing the relative importance of weak and strong ties; (3) alteri composition, characterizing the likely homophily of attributes between egos and alteri; and, (4) social support, describing the

direction of material and immaterial resource flows among social actors in the returned migrant networks. Table 2 – Descriptive Statistics of the Returned Migrants’ Personal Networks, 2006 and 2016. Variables Mean 2006 2016 Network Measures Density 0.38 0.35 Degree (Centrality) 44.18 41.74 Betweenness 14.58 18.04 Ego-alteri proximity Strong Ties 26.93 34.25 Weak Ties 13.07 5.74 Alteri Composition Men 22.59 19.84 Women 17.41 20.16 Returned Migrants 2.35 4.18 Migrants in USA 7.01 3.60 Migrants in other countries 0.63 0.97 Non-Migrants 30.02 31.25 Brokers 0.12 0.10 Social Support Ego helped alter 1.54 1.75 Alter helped ego 0.98 0.44

Source: Primary data from fieldwork in Governador Valadares, Brazil (2006, 2016)

As part of the network topology block, the density measure describes the general level of connectivity between the points in a reticular configuration (Scott, 2000). Ranging from 0 (isolates) to 1 (complete graph), it is given by the ratio of the number of effective connections in the network by the total number of possible connections between all pairs of nodes. The average density in 2006 and 2016 was 0.38 and 0.35, respectively5. The relative constancy in the network density measure over time suggests that the links between returned migrants and their personal networks

5 It is important to note that the density varies with the size of the networks. In general, networks between 50 and 150

individuals constitute medium-sized social networks and densities between 0.20 and 0.40 are considered relatively stable and well-connected networks, and networks over 60 and density greater than 0.50 are considered unlikely in the real world (Prell, 2012; Kadushin, 2012).

are fairly stable, reinforcing their positional role in a society marked by intensive migration activity from and to the United States. The average degree summarizes the network activity and corresponds to the number of edges a node establishes with other nodes in the network. Ranging from 0 to 100, its normalized version determines the direct connection capacity of a node (Scott, 2000). Here we use a summary measure of global centrality for the full network, that is, a weighted average of the centrality of all alteri in the ego network as an indicator of global connectivity (Prell, 2012). The distribution of the global centralization measure is not linear but parabolic, so the proximity of the indicator by 50% indicates a marked centralization. The figures for 2006 (44.2) and 2016 (41.7) complements the density measure, suggesting that most connections established in the network are of short distance. In addition, the proximity to the 50% threshold is indicative of a privileged position of the returned migrant in providing stability to the network structure over time6. In order to understand the ability of a social actor to intermediate the exchange of information in the network, the node betweenness is always used as the best candidate (Scott, 2000). It refers to the existence of actors who occupy the position of greater convergence of "alternative routes" defined by short paths between any two people in the network. The larger its value, the greater the fragmentation and the likelihood of structural holes in the

• network. It also suggests the existence of brokers in the network; that is, a large

betweenness coefficient for a particular node (above 20%) indicates that the node is a social actor able of managing structural holes due to his/her greater relative social capital (Wasserman and Faust, 1994; Prell, 2012). Here we use a global weighted measure of betweenness to evaluate the average behavior of each personal (ego-centered) network.

6 A topological analysis of the correlation between the egos’ degree and the average degree of their

topological neighbors suggest a strong negative correlation in both years. That is to say that the returned migrants centralize the links in the network. Results available upon authors’ request.

The value of 14.6 for 2006 and 18.0 for 2016 suggests an intermediation activity with few breaks (structural holes) in the network, and emphasizes how the compact networks contribute to the perpetuation of both the illegal migration market and the spread of the culture of migration in the region (Sousa and Fazito, 2016). Actually, the increase in the average betweenness from 2006 to 2016 reinforces the increasing social capital of the returned migrants and the vibrant exchange of vital information on when and how to migrate among residents in our study area. The next block category in Table 2 refers to the proximity of the alteri to the egos. The number of strong and weak ties was determined by the degree of self-reported proximity, indicating the type of relationship and its intensity7. In 2006, the personal networks of the returned migrants had an average of 27 strong ties and 13 weak ties. In 2016, strong ties increase to 34. These values highlight the existence of dense social groups in the social structure of the returned migrants, and suggest the diffusion of their values and ideas in the social network of a community where we find an established migration culture. Although it is difficult to go much beyond this, it is likely that the increase in the number of strong ties are reflective of a more strategic rule of the returned migrant in reducing the costs and risks

• f out-migration as prospects of migration surveillance in the US become more salient.

The composition of the ego networks reveals a fairly stable sex ratio over time. Differences in the composition of the migration status of the alteri, however, are revealing. In 2006, each ego was linked to 30.0 individuals with without international migratory experience, to 7.6 international migrants, and to 2.4 international returned migrants from the US. In 2016, his/her average ego-alteri link by migration status was 31.3, 4.6, and 4.2, respectively.

7 In the original questionnaires, relation intensity between ego and alteri was measured in a 4-points scale,

as follows: (1) distant, (2) acquaint, (3) close, (4) very close. For the alteri-alteri relation intensity, an additional category was available for those who did not know each other. The weak ties were defined as the answers (1) + (2). The strong ties were classified as the answers (3) + (4). The total number of ego-alteri ties is restricted to 40 in each survey.

Fazito and Soares (2014) emphasize that the presence of actors with migration experience in a personal network is a key factor to understand the ability of networks to propagate key information regarding the costs and benefits of the migration project. In this sense, the presence of alteri migrants and returned migrants found in the composition of ego personal networks suggests the preponderance of returned migrants in the dissemination of information and qualified resources to other potential migrants in these personal networks (Fazito, 2010). The increase in the number of returned migrants in the composition of the networks between 2006 and 2016 suggests a possible relation with the effective increase in the volume of returned migrants in Brazil observed at the end of 2010 (IBGE, 2011). The variable broker considers the average number of alteri in the personal network characterized as that individual that works directly with adulteration of documents or some

• ther agency related to the illegal border crossing. Considering all the ego-centered

networks, only 11 alteri (6 in 2006 and 5 in 2016) were characterized as brokers, producing a very low average of brokers for each personal network (0.10 in 2006 and 0.12 in 2016). Fazito and Soares (2014) broadly discuss the issue of these agents in the so-called 'Industry

• f Illegal Migration', pointing out that, due to the nature of their activity, such actor is

expected to have weak ties with potential migrants, maintaining only strategic connections and occupying positions where there are structural holes in the migratory system. In other words, in order to access a broker who conducts an illegal border crossing or produces illegal documentation, the potential migrant will rarely have a direct and strong tie, accessing these brokers indirectly through their qualified ties. This indirect link (potential migrant accessing a person who is linked to a broker) is the basis for the formation of triads between them. The human and social capital acquired by the returned migration due to his/her international migration experience is considered to be positively associated with directed support to

potential migrants (Soares, 2009). This support may be of many sorts, from information, to falsification of documents, to financial resources, to contacts at the origin and destination. Because of this capital, returned migrants would act strategically in the diffusion and perpetuation of a migration system. One way to look at this strategic action is to measure the direction of support between returned migrants and their alteri. Supporting our working hypothesis, the returned migrant has a much larger role in providing social support for their alteri in the realization of the migration project than being helped by them. The level of social support from returned migrants is also higher in 2016 than in 2006. In 2006, for instance, an ego has helped on average 1.54 alter in his/her personal network with information or financial resources for the realization of the migration project, compared to being helped by

• nly 1 individual by the time of his/her migration to the US. In 2016, the data showed a slight

expansion in the social support offered by the ego to his/her alteri (1.75) and a reduction in the help of the alteri to the ego (0.44). These evidence reinforce the theoretical perspective of migration that recognizes social networks and other institutions in the perpetuation of the migratory system represented by actors and their structural positions. Above all, the data from the two moments in time highlight the activity of the migratory social networks and how the consolidation of the “Industry of Illegal Migration” in the GV area can contribute to new displacements in the future. The analysis of some specific profiles of social support can contribute to a better understanding of the structural configuration of the migration system linking GV to the US. Figure 1 shows three sociograms picturing different roles performed by a returned migrant in the origin community. The first sociogram (Vanessa) was labeled as the one where the returned migrant performs an active social support role, contributing directly to his/her alteri’s strategies to out-migrate. The second sociogram (Poliana) emphasizes Fazito’s (2010) argument that not all human and social capital acquired by returned migrants is applied to

mediating actors between origin and destination communities. Finally, the third sociogram (Silvano) pictures the personal network of an ego that has become a specialized agent in the process of international migration as a direct mediator of illegal border crossing (broker). Figure 1 – Comparison of profiles of social support from the personal networks of Returned Migrants.

Personal Network Vanessa (Supporter) Personal Network Poliana (Inactive) Personal Network Silvano (Broker) Non-Migrants Returned Migrant International Out Migrant Migrants who received support to migrate Brokers

Source: Primary data from fieldwork in Governador Valadares, Brazil (2006, 2016) using Gephi software.

The analysis of these three examples indicates that returned migrants may or may not

• ccupy structural positions that carry the necessary social support for the legal or illegal

migration, as well as the reduction of risks related to housing, financing, and employability in the destination community (Fazito, 2010). When put together, however, these stacked and linked ego-centered networks of returnees produce a dense web where key information and more broad forms of social support are channeled, contributing to the strengthening and amplification of the migration system in the GV area and its surroundings, facilitating migration in good times and reducing the costs and risks of migration in periods of crisis. This kind of ‘positive contagion’ occurs through the connections existing in the social network

• f local residents. Figure 2 shows the connectivity and flow of the exchanges between

several actors through the representation of a total network containing all the interviewees and their alteri8. This total network helps visualizing how and with who interactions occur, strengthening both the migration market and the ‘culture of migration' in the region (Sousa, 2016; Sousa e Fazito, 2016). Figure 2 – Total network with Egos (respondents) and their Alteri (out-migrants in the United States, returned migrants, and non-migrants)

Source: Primary data from fieldwork in Governador Valadares, Brazil (2016) using software Gephi.

8 The identification of the same actor in different personal networks (disambiguation) obeyed the following

criteria: returned migrants possessing equal names of non-migrant individuals were considered as of different individuals, had they returned under the same name, their family connections and the structure of snowball data collection were checked to keep separate or unite two actors as one; to link migrants, we followed the same criterion, separating non-migrant homonyms and verifying their family composition; Non- migrants were only united when their names were the same and the family structure reported by the egos was compatible Imagine, for instance, that Ego A stated that X1 is his daughter, and Ego B said that X2 is his wife, but also that he is the son-in-law of Ego A. Then, if X1 and X2 had the same name they were assumed to be the same person. The details of this composition can be found in Table 3 in the appendix.

Figure 2 reveals a very dense network, with links between non-migrants and individuals with different levels of migration experience (from out-migrants to returned migrants). This mixed link by migration status across the network helps explain why individuals from GV area continue to come and go, keeping the migratory social network active even in apparently adverse conditions. The routes remain open and activated, constantly adjusting to the local and global contexts of economic and demographic dynamics, as in a moment of economic crisis observed in 2016. Official channels are sometimes replaced by more fluid, proximate mechanisms of migration represented by the figure of the returned migrant. The GV-US migration system, structured by personal and institutional social networks, has thus emerged as the result of three different but related forces: the historical social representations of the relations established between the United States and the GV area, the migrants’ remittance flows, and the returned migrants’ human and social capital. These factors have apparently served as important mechanisms for the preservation and reproduction of flows of international emigration in this decade. The latter, in particular, has been used strategically as a more responsive mechanism in periods of crisis, where access to support needs to be quick, cheap, and readily (locally) available. The very returned migrants’ lifestyle has been considered an emulation factor providing the necessary motivation for prospect migrants to initiate the search for a better life elsewhere.

Concluding Remarks

As one of the most studied regions in Brazil for international migration, the GV area has a solid history of out-migration to the US, and more recently to Portugal (Soares, 2009; Pereira and Siqueira, 2013). The region was marked by a steady decline in economic activity due to the failure of the extensive cattle ranching/mining system in providing a sustainable flow

• f production over time.

Many studies have looked extensively at the importance of migrants’ remittances from the US to the region (Soares, 2009), the psychological and health effects of the fragmented families of migrants (Siqueira and Santos, 2013), and the perpetuation of the migration system looking at the transnational family links (Fonseca et al., 2016). Very little is known about the role played by the returned migrants for the stability of this migration system and for the weathering of internal/external shocks that represent increased risks and costs related to the migration process (Fazito, 2010). This study is an effort to shed some light on this issue. While the 1990’s and 2000’s in Brazil were decades of difficult opportunities in the job market for the less qualified, motivating the search for employment in other countries, the 2008- 2014 economic boom in Brazil was highlighted intensively in the news worldwide, creating an image of a land of endless prosperity. The image of a booming economy contrasted sharply with the hardship faced by the developed world in this period (mainly the US). In 2015, Brazil entered a strong period of economic recession, while the US slowly recovered from the 2008 financial crisis (Nassif et al., 2015). Then Trump was elected under a political platform of strong restrictions to unqualified and undocumented migration. Against all odds,

• utmigration from GV to the US remained virtually unaltered (Sousa and Fazito, 2016).

This study paid special attention to returned migrants in the GV area, and to how they positioned themselves in the local social network. The returned migrant is a fundamental element for the perpetuation of the migratory system due to his/her ability to mediate and directly provide strategic resources in the community. Upon returning, this individual is characterized as an actor of unique social capital in the migration system, capable of providing financial and psychological support, managing contacts between source and destination, and disseminating key values for the achievement of the migration project. His/her positional role in the social network and the type of flows of support exchanged

between him/her and his/her alteri are elucidative of the stability of a migration system and its resilience to economic shocks. Our analysis has shown that this returned migrant has been increasingly involved in the migration system by facilitating illegal border crossing (through indirect links), reinforcing transnational family links, and fostering the image of international migration as a highly profitable project through his/her lifestyle. The very presence and active connections of these migrants in the local community (including connections with many residents without any migration experience) help explain why flows of migrants from Brazil to the US remained stable in the last decades, and suggest that this trend is likely to continue in the future. Different from traditional institutional arrangements, the returned migrant is local (readily available), approachable, and involuntarily triggers an emulation effect to prospect migrants even when they are not directly involved in the exchange of strategic flows of social and material support. We believe that those are enough reasons to justify more in-depth studies about the role of returned migrations in the consolidation and perpetuation of migration

• systems. This may be the missing link in the transnational families theory that help explain

the stability of migration flows even when these families are broken (Jorgensen, 2017). While advancing the literature of migration systems, this study has several limitations. First, the data collected are based on a non-probabilistic sample design. While highly recommended for the study of networks (especially for the study of undocumented migration), it may overestimate the density of a social network (Fawcett, 1989; Fazito, 2010). Second, it is mostly descriptive; all inferences are based on topological and relational metrics instead of an inferential graph model. This is a smaller problem, since the inferential modeling of ego-centered networks with more realistic (markovian) assumptions is still very

• limited. Most models available impose unrealistic independence assumptions for the

existence of connections that fit very poorly studies on social network analysis (Hunter,

2007). Finally, the time comparison used in this study is not indicative of a probabilistic transition of characteristics, attributes or positional roles. The data were created as independent samples for each year, missing the true longitudinal aspect of follow-up studies. We hope that, despite these limitations, the findings presented in this study are robust enough to motivate and to foster new studies on migration systems and their resilience to shocks.

References

DE HAAS, H. Migration and development: a theoretical perspective. In: International Migration Review, 44(1):227-264. 2010. FAWCETT, J. T. Networks, Linkages, and Migration Systems. International Migration Review 23:671-80. 1989. FAZITO, Dimitri. Reflexões sobre os sistemas de migração internacional: proposta para uma análise estrutural dos mecanismos intermediários. Tese (Doutorado em Demografia), Centro de Desenvolvimento e Planejamento Regional, Universidade Federal de Minas Gerais. 2005. FAZITO, Dimitri. Análise de Redes Sociais e Migração: Dois aspectos fundamentais do “retorno”. Revista Brasileira de Ciências Sociais. Vol. 25. N. 72. p.89-100. 2010. FAZITO, Dimitri; SOARES, Weber. Capital social, análise de redes e os mecanismos intermediários do Sistema migratório Brasil/EUA. Revista Geografias, Vol. 6, N. 1, p. 27-

• 41. 2010.

FAZITO, Dimitri; SOARES, Weber. The Industry of Illegal Migration: Social Network Analysis

• f the Brazil-US Migration System. International Migration / OIM. doi: 10.1111/imig.12034.

2014. FONSECA, M.C.; ESCALDA, P.M.F.; SIQUEIRA, S.; ASSIS, G.O.; THIO, C.B. Divisão Sexual do Trabalho, Rearranjos Familiares e Relações de Gênero em Comunidade Rural de Emigração Masculina. Annals of the 2016 Meeting of the Brazilian Association of Population Studies. HUNTER, D. R. Curved exponential family models for social networks. Social Networks, v. 29, n. 2, p. 216-230, 2007. IBGE – Instituto Brasileiro de Geografia e Estatística. Censo Demográfico Brasileiro 2010. Rio de Janeiro: IBGE, 2011. JORGENSEN, N. Migração Internacional, Remessas, Família e Gênero: o caso de Governador Valadares. Dissertação (Mestrado em Demografia), Centro de Desenvolvimento e Planejamento Regional, Universidade Federal de Minas Gerais. 2017. KADUSHIN, Charles. Understanding Social Networks: theories, concepts and findings, Oxford University Press, UK. 2012. KRITZ, M.; ZLOTNIK, H. Global interactions: migration systems, processes and policies. In: Kritz Mary et al. (Ed.) International migration systems, a global approach. Oxford: Clarendon, p.1-16. 1992.

LIN, Nan. Social Capital: a theory of social structure and action, Cambridge University Press, UK. 2001. MCCARTY, Christopher. Structure in personal networks. Journal of Social Structure, 23(1). 2002. Disponível em: <https://www.cmu.edu/joss/content/ articles/volume3/McCarty.html> Acesso em: 01 de Abril de 2015. MASSEY, D. S.; ARANGO, J.; ALARCÓN, R.; DURAND, J.; GONZÁLEZ, H. Return to Aztlan: the social process of international migration from western Mexico. Berkeley, Los Angeles, London, University of California Press, 335 p. 1987. MASSEY, D. S. Social Structure, Household Strategies, and the Cumulative Causation of

• Migration. Population Index, v. 56, n. 1, spring/1996, p. 3-26. 1990.

MASSEY, D. S.; ARANGO, J.; HUGO, G.; KOUAOUCI, A.; PELLEGRINO, A.; TAYLOR, J.

• E. Theories of International Migration: a Review and Appraisal. Population and
• Development. Review, v. 19, n. 3, set/1993, p. 431-466. 1993.

NASSIF, A.; FEIJO, C.; ARAUJO, E. Structural change and economic development: is Brazil catching up or falling behind?, Cambridge Journal of Economics, v. 39, n. 5, p. 1307– 1332, 2015. PEREIRA, Sónia; SIQUEIRA, Sueli. Migração, retorno e circularidade: evidência da Europa e Estados Unidos. Revista Interdisciplinar da Mobilidade Humana, [S.l.], v. 21, n. 41,

• dez. 2013.

PRELL, Christina. Social Network Analysis: history, theory and methodology, SAGE,

• UK. 2012.

SAYAD, Abdelmalek. A imigração ou os paradoxos da alteridade. São Paulo: EDUSP,

• 2000. 299p.

SCOTT, John. Social Network Analysis, A Handbook, SAGE, London. 2000 SIQUEIRA, Sueli; SANTOS, Marcelo Henrique. Condições de saúde do emigrante no retorno para sua terra natal. Revista Interdisciplinar da Mobilidade Humana, [S.l.], v. 21,

• n. 40, jun. 2013. ISSN 2237-9843.
• SOARES. Weber. Da associação entre os retornados internacionais e os intermediários da

rede migratória valadarense. Revista Interdisciplinar da Mobilidade Humana – REMHu. Ano XVII, N. 32, 2009. p.47-59. SOUSA, L. Redes Sociais, Mercado e Cultura Migratória – Um estudo sobre fatores associados à mobilidade populacional na Microrregião de Governador Valadares no Século

• XXI. 2016. 154p. Tese (Doutorado em Demografia) – Centro de Desenvolvimento Regional

e Urbano da Universidade Federal de Minas Gerais. SOUSA, L.; FAZITO, D. Um estudo sobre as oscilações e determinantes da emigração internacional para os EUA a partir da Microrregião de Governador Valadares entre 2000 e

• 2010. Revista Bras. de Estudos Populacionais. v. 33, n. 3, 2016.

WASSERMAN, Stanley, FAUST, Katherine. Social network analysis: methods and

• applications. Cambridge: Cambridge University,1994. 825p.

WELLMAN, Barry. Structural analysis: from method and metaphor to theory and substance”, in Wellman, B.; Berkowitz, S.D. (Eds), Social Structures: A Network Approach, Cambridge University Press, Cambridge. 1988.

Appendix

TABLE 3 – Total Network composition, 2016

Variáveis Freq % Network composition with loops Interviewees 'Ego' 2006 and 2016 100 Alteri 2006 and 2016 4000 Network composition joined Interviewees Returned Migrants USA 100 3,30 International Out-Migrants from USA 359 11,85 International Out-Migrants from Portugal 37 1,22 International Out-Migrants from Other Countries 7 0,23 Returned Migrants from USA 188 6,16 Returned Migrants from Portugal 48 1,61 Returned Migrants from Other Countries 6 0,22 Non-Migrants 2284 75,41 Total 100,0 International Migratory Experience International Out-Migrants and Returned 745 24,6 Non-Migrants 2284 75,4 Alteri helped Ego* Money 170 2,8 Documents 46 0,8 Crossborder Contacts 42 0,7 Destination Contacts 86 1,4 Broker 8 0,1 Ego helped Alteri** Money 90 5,1 Documents 22 1,2 Crossborder Contacts 16 0,9 Destination Contacts 66 3,7

* An individual can receive and give various kinds of help to another person ** Percentage based on number of alteri migrants and returned Source: Gv área Case Study (2006; 2016)