Migration responses to the Spanish economic crisis among Latin - - PDF document

1

Migration responses to the Spanish economic crisis among Latin American immigrants*

Victoria Prieto-Rosas† Joaquín Recaño-Valverde‡ Abstract This paper assesses the changes in migration for internal and international outmigration of foreign immigrants in Spain, paying special attention to the Latin American origins. First, based on register data from Residential Variation Statistics published by the Spanish National Statistics Office, we describe the magnitude and intensity of inter-regional migration, return, and outmigration to a third country for the period 2006-2013. Second, we analyze the demographic characteristics of migrants and the incidence of time and context

• f origin on the probability of moving abroad with respect to migrating internally. Results

discuss which individual and territorial characteristics orient migrants in preferring different forms of out-migration in reference to internal migration and contribute to understanding which of these strategies was dominant at the various stages of the recent economic recession. KEYWORDS: return, internal migration, remigration, European Union, Spain, Latin America

* This paper has been carried out in the framework of the research project: Movilidad geográfica y acceso a la

vivienda: España en perspectiva internacional (Ref. CSO2013-45358-R) funded by the Ministry of Economy and Competitivity, National R+D+I.

† Population Program, School of Social Sciences, University of the Republic (Uruguay). E-

MailVictoria.prieto@cienciassociales.edu.uy.

‡ Department of Geography and Centre of Demographic Studies, Autonomous University of Barcelona

(Spain). E-Mail: jrecano@ced.uab.es.

2 Introduction The migration responses to the 2008 Spanish financial crisis captured most of the attention from migration scholars, but the greatest consideration was given to return migration (Ramos & Matano 2013; Cerrutti & Maguid 2014; Recaño & Jáuregui 2014), followed by remigration (Larramona 2013; Mateos 2015; Pereira 2012; Mas-Giralt 2016). However, the changes in internal migration were relatively overlooked with a few exceptions (Quintero- Lesmes 2015; Recaño & de Miguel Luken 2012). Joining the literature that bridges the gap between internal and international migration (King et al. 2008; Czaika & de Haas 2012; Czaika 2012), this paper argues return and remigration to be competing responses to internal migration. We argue that the simultaneous assessing of these three movements is as important as the individual study of them, at least aiming to understand the diversity of responses adopted by immigrants to face the increase in unemployment. Based on the analysis of Spanish Residential Variation Statistics, we discuss the changes in migration occurred between 2006 and 2013 for internal and international movements of foreign immigrants from Spain, paying particular attention the Latin American origins. Latin Americans were responsible for most of the growth of foreign population in Spain for almost a decade and led together with Africans the return migration outflow since 2009. They also contributed to the recovery of internal migration before the recession, and still played a significant role after the advent of the crisis; between 1998 and 2014 the total number of movements of this origin was three times the observed for the native population (Recaño & de Miguel Luken 2012). During the crisis, return and remigration flows of the native and foreign-born population increased while the intensity of internal migration first declined and remained stable after 2008. Another consequence of the economic collapse was expressed changing the geography of immigrants’ internal migration: before the crisis, Madrid, Barcelona, the Mediterranean Coast, together with the Canary and Balears Islands, were the main immigrant-receiving regions. After 2008 only Madrid and Barcelona have remained in this group (Quintero-Lesmes 2015). Despite all the underwent transformations in the inner and international migration pattern of Latin Americans in Spain, the research agenda has predominantly focused on return migration (Koolhaas 2015; Recaño & Jáuregui 2014), and returnees´ socio-economic integration (Cerrutti & Maguid 2014; Prieto & Koolhaas 2014; Prieto 2016; Vela Peón & Cabezas 2015). Exceptionally, Larramona (2013) assesses the determinants of both return and remigration from Spain. Nevertheless, her work does not account for internal migration, nor distinguishes between remigration and outmigrations with ignored destination, and focuses on the period before the economic recession (2002-2009). The migration responses to the recession adopted by male Mexican migrants in the United States did introduce a major precedent for this paper by considering both internal and international mobility responses together. Torre-Cantalapiedra & Giorguli (2015) accounted for return as an alternative response to internal migration within the United

• States. Based on individual longitudinal data for the period 1942-2011, they assessed the

preference for these reactions pointing to return as the preferred strategy to face the

3 American economic crisis. However, their findings acknowledge significant transformations in the intensity and direction of internal migration of Mexicans in the United States too1. Our perspective shares with this last article the interest on including inner and international mobility responses to the economic recession, but differs in the approach adopted. We were not able of estimating the hazard of both types of events since we worked with the Residential Variation Statistics published by the Spanish National Statistics Office, which does not include information on those that do not move. Instead, we can compare the probability of opting for any form of international outmigration respect of internal migration, similarly to Larramona´s study of outmigration from Spain for 2002-2009 (Larramona 2013). Along these lines, we discuss which individual and territorial characteristics orient migrants in preferring return, remigration or outmigration with an unknown destination, in respect to moving internally. The reason for including this last category responds to the significant share of outmigrations captured as a result of the statistical operation of clearing the Population Register of foreign-born individuals who do not renew their inscription after two years (Ortega et al. 2013; Gil 2010). First, we describe the intensity of inter-regional migration, return and other forms of

• utmigration. Second, using multinomial logit modeling we analyze the demographic

characteristics of migrants and the incidence of context variables at origin on the probability of moving abroad over moving internally, first for all origin, and then for Latin Americans only. The inclusion of dummy variables for three periods of time (2006-2008 pre-crisis, 2009-2010 economic collapse, and 2011-2013 recession) contributed to explaining if the prevalence of each migration responses varies according to the different contexts of origin and stages of the economic downturn. The paper structures in six sections, five?? following this introduction. In the second section, we discuss the literature centered on the evidence for return, remigration and internal migration of foreign-born population in Spain with a focus on Latin American

• immigrants. In the third section, we present the data and methods used. In the fourth

section, we discuss the key findings for the intensity of internal and international migration

• f foreign-born population. In the fifth section, we introduce the results of multivariate

analysis of the determinants of different forms of international outmigration. Finally, in the sixth section, we draw the main conclusions and provides a general discussion of results.

1 Regarding the individual determinants for both kinds of migration, they point to the duration of the stay, the

legal status in the receiving country and the occupation as the main factors explaining mobility trends in 1946-2011.

4 Background During the crisis international migration flows have followed two different trends: on one side a reduction of the inflow as a consequence of the deterioration of the economic and social situation in Spain; on the other side, foreign-born population outmigration followed a sharp increment. Looking at the population stock, in January 2013, for the first time in Spanish recent history, the registered population decreased 3.3% due to the loss of 190,020 foreign residents (Quintero-Lesmes 2015). The workforce of Latin Americans drop about 16.3% after the crisis, and the “de-Latin Americanization” of immigration in Spain emerged as one of the demographic consequences of the economic crisis. This decline in the share of this origin in the total foreign-born population was particularly pronounced in Madrid, Barcelona and other cities located along the Mediterranean coast (Vidal-coso et al. 2012; Quintero-Lesmes 2015; Torres-Pérez, Francisco Moncusí & Esteban 2013). The growth in the outflow of foreign-born coincides not only with the decline of employment in the sectors where most Latin Americans were working -manufacturing, construction, and tourism- but with the onset of public aid supporting return for immigrants2. Previous work has shown that migration responses of immigrants to economic shocks could be very diverse, varying over time and among different individuals (Pandit 1997; Mas- Giralt 2016; Zimmermann & Zaiceva 2012; Czaika 2012; Czaika & de Haas 2012; Pereira 2012). Our hypothesis suggests that internal migration might have been preferred at early stages of the Spanish recession, but once the unemployment becomes endemic return and remigration could have increased significantly becoming the preferred responses for

• immigrants. Therefore, in the following lines, we discuss the trends of immigrants internal

and international out-migration, as well as the impact of the Spanish economic cycle on

• migration. Although we focus on Latin American migrants, we discuss the impact on other

foreign origins too. Internal migration vs. outmigration of immigrants According to the literature, internal or outmigration of foreign-born population in Spain and United States -two of the most relevant destination for Latin American international migrants- has been associated with variations in human and social capital, citizenship or regular status of foreign residents, but also on the economic conditions in the region of

• rigin (Torre-Cantalapiedra & Giorguli 2015; Gurak & Kritz 2000; Larramona 2013;

Recaño & de Miguel Luken 2012; Quintero-Lesmes 2015).

2 At this time three programs were developed targeting potential returnees from Spain: a) the voluntary return

program for humanitarian reasons, created in 2003, supported by IOM and several NGOs; b) the Social Security program, in force since 2008, that granted advance payment of accumulated unemployment subsidy for non-EU28 workers; and, c) the voluntary return program focused on foreign-born population with solid business plan, launched in 2010. Furthermore, different Latin American countries promoted return with specific return programs, though its impact has been very limited (Gonzalez-Ferrer 2014; Czaika et al. 2013; Koser & Kuschminder 2015).

5 The intensity of internal migration of immigrants in Spain declined since 2007 when the economy showed the first symptoms of destabilization. In this respect, the Latin Americans were the foreign origin that experienced the smallest impact of the crisis. The summary measures of migration intensity for this origin declined at a lower rate than for the rest of foreign-born population, being Bolivians, Argentineans, and Dominicans the three most affected origins. The only exceptions have been Colombians and Ecuadorians, whose intensity rates were even above Spaniards. Furthermore, the length of the movements has enlarged in the period 2009-2011, and most of the movements went from the Mediterranean area and the Canary Islands, the most affected regions due to the increase in unemployment rates, to the Basque Country, Catalonia, and Madrid (Quintero-Lesmes 2015). When the internal migration is analyzed by comparison to other outmigration responses, the role of individual determinants varies significantly. For example, regular residence status or naturalization reduce the odds of return and increases the probability of experiencing an internal migration within the United States (Cantalapiedra & Giorguli 2015). Although the simultaneous study of determinants of outmigration and internal migration has no precedent in Spain, the individual study of these behaviours showed that being a Spanish citizen increases the odds of experiencing non-return outmigration3 among foreign-born population (Larramona 2013; Recaño et al. 2015), and reduces the odds for internal migration (Quintero-Lesmes 2015; Recaño & de Miguel Luken 2012). Regarding the determinants of mobility in the region of origin, the sectoral structure of the local economy has shown to play a differential role for internal or outmigration too. The internal migration rates between agricultural regions of the United States or Spain is particularly high among the foreign-born population (Durand & Massey 2003; Torre-Cantalapiedra & Giorguli 2015; Recaño & de Miguel Luken 2012). Conversely, in both contexts return migration has shown to be higher from regions where services or manufacturing sectors correspond to larger shares of GDP (Torre-Cantalapiedra & Giorguli 2015; Recaño et al. 2015) or higher unemployment rates (Larramona 2013). Despite the fact that remigration has been growing after the onset of economic crisis, the return has been preferred over onwards international migration (Recaño et al. 2015). Remigration rates proved to be larger for dual citizens (Spanish and other non-EU28 citizens) than for non-EU284 citizens. The outflow heading the United Kingdom, followed by Germany, France, Switzerland and The Netherlands, showed the largest migration rates for Latin Americans. United Sates and Italy remain as important destinations for this origin

• utmigration from Spain, but the flow to these countries slightly declined after the crisis.

According to Recaño et al. (2015), the increase observed in male onwards migration rate was concentrated in the Mediterranean provinces and inner rural areas, while the growth at the same rates did not show a specific geography and dispersed all over the country among women.

3 This is the name used by Larramona (2013) to recall the outflow of remigration to a third country different

from birthplace and the outmigration with ignored destination captured by the Spanish Residential Variation Statistics.

4 The 28 member states of the European Union are Belgium, Bulgaria, Czech Republic, Denmark, Germany,

Estonia, Ireland, Greece, Spain, France, Croatia, Italy, Cyprus, Latvia, Lithuania, Luxembourg, Hungary, Netherlands, Austria, Poland, Portugal, Romania, Slovenia, Slovakia, Finland, Sweden and United Kingdom. Here we will refer to non-EU28 when speaking about any country different from EU28.

6 Recently, Larramona (2013) has estimated the probability of return migration in respect to non-return outmigration for the period 2002-2009. She modeled together with the odds of experiencing remigration or departures from Spain for which the migrant did not inform the

• destination. Results showed that the age has a positive effect on the probability of return

migration, while non-return outmigration (remigration and outmigration with unknown destination) is rare at advanced ages. Regarding sex, her findings point to women as more prone to experience remigration than men -the group with the larger probability of returns. Similarly, citizenship distinguishes the profile of returnees from other outmigrants since EU28 citizens have larger odds for return compared to non-EU28 citizens. Finally, individuals born in Asia, Africa or Latin-American are less likely to experience a return to their birth country during 2002-2009, and instead, they preferred to leave the country moving to a third country or without informing their departure at all. At a more aggregated level, Larramona (2013) controls either for the context of origin (locality) and destination (country). The findings point that individuals leaving smaller localities had fewer chances

• f experiencing a non-return outmigration and preferred returning to their country of birth.

The impact of economic crisis on the Spanish labor market The financial crisis has destabilized the immigrants´ process of social integration through its impact on unemployment, quality of employment and poverty. Unemployment rates were particularly severe among the foreign-born who dealt with a 36.6% unemployment respect to 24.3% faced by natives. This gap responds to the overrepresentation of immigrant workers in the most affected economic sectors –construction and manufacturing- , which concentrate the occupations with highest temporary and unskilled jobs too (Aysa- Lastra & Cachón 2012; Pajares 2010; IOM 2012). In the first phase of the crisis, 2008-2010, the most extended strategy was the adjustment to temporary employment, which facilitated the entry into the labor market for many immigrants, but also triggered the departure of workers already employed. Since then, the loss of employment was higher for permanent jobs. On the other hand, there are differences between different groups of immigrants. In 2011, the unemployment rate of Latin Americans was greater than for natives, 28.5%, and 19.5%, respectively, although it remains the lowest compared to the rest of non-EU28 immigrants5. This and other aspects may vary depending on the region of Spain, its economic structure and specialization of each group. We should mention that this economic crisis hit in an already segmented labor market. The recent Latin American immigration to Spain responded to the dynamism of particular spots

• f the Spanish labor market and from the very beginning, their integration in Spain was
• segmented. Most Latin Americans took low-skilled jobs, while the newly educated Spanish

cohorts enrolled in upwards social mobility (Bernardi et al. 2011). It was the dynamism of low-skilled jobs, especially in construction and domestic service, which ensured a rapid insertion for the foreign-born arriving previous to 2008 (Rodríguez-Planas & Nollenberger 2014). Before the beginning of the crisis, the participation and employment rates for this

• rigin were similar to those of native population and higher than for other foreign-born

groups (Muñoz Comet 2013).

5 The unemployment rate of African was reached 49.3% (Colectivo Ioé, 2012: 72).

7 With the advent of the recession Latin Americans were no exception to the decline of employment, and even the most educated suffered the depreciation of their educational credentials (Cebolla-Boado et al. 2015). Moreover, the crisis plunged the bargaining power

• f foreign workers, who accepted worse conditions of employment (Aysa-Lastra & Cachón

2012) or showed more flexibility to geographical and sectoral mobility in the labor market (Vidal-coso et al. 2012; Quintero-Lesmes 2015). The segmentation of the Spanish labor market also responds to other structural factors, which preceded the arrival of Latin Americans. This labor market is split in a primary sector with skilled jobs, high wages, and permanent contracts; and a secondary sector that concentrates less skilled workers worse-paid and exposed to the high rotation (Muñoz, 2013). In any case, the recession worsened the labor segmentation already described, widening the gap between Spanish and foreign-born, fundamentally affecting the men employed in low-skilled sectors, such as construction (Aysa Lastra and Cachón, 2012). Foreign women were less affected by the decline in employment due to their over- representation in economic sectors which are less sensitive to recession, such as domestic and care work (Vidal, Gil, and Domingo 2012). Data and Methods The Statistics of Residencial Variation shows that between 2007 and 2011, 1,226,000 people left Spain, of which 97% were foreign-born. Even if the destination is ignored for more than 80% of the outflow (Recaño & Jáuregui 2014; Prieto Rosas & López Gay 2015), in the cases where the destination is informed -either before departure from Spain or at the arrival at any Spanish Consulate- 85% of the outflow corresponds to returns and 15% to remigrations.The answers to the question on plans for the following five years, which was asked to immigrants surveyed at the National Immigrants Survey 2007, showed a similar distribution to the composition of outmigration with a known destination at the Statistics of Residential Variation (Larramona 2013). For measuring internal migration and outmigration of Latin Americans, we also used the Residential Variation Statistics. The characteristics of this data source –with such a significant share of missing destination for more than 80% of the outflow- demanded a broader classification considering the variety of destination of the outmigration flow. Along this article we use the following classification for migration: i) return outmigration, referring to the outflow heading the country of birth; ii) remigration to EU28 countries; referring to the migration to another EU28 member state which is different from birthplace; iii) remigration to non-EU28 countries, relating to the outflow heading another country different from the country of birth and the rest of the EU28; and iv) outmigration with unknown destination. The study of this fourth behavior challenges the modeling of the migration outcomes since we are modeling a behavior that is a melt of returns and remigrations, and the rationalitites applying for each of its components may differ. However, given that this last type of movement constitutes around 8 out of 10 cases the dataset, we decided to include it as an outcome favoring an accurate estimation of the probabilities of migration for which we do know the destination (return and remigration). To study internal migration and outmigration intensities we estimated migration rates by sex, age, country of birth, and citizenship, with the purpose of describing changes in the

8 demographic and geographical pattern of migration before and during the crisis. The period

• f study goes from 2006 to 2013. Though data on outmigration of foreign-born population

is available since 2004, estimations for 2004-2005 are underestimated, and we have preferred focusing on 2006-2013. The estimation of rates required a combined use of the Residential Variation Statistics, for flows, and the Population Register, for exposures. As stated before the different types of migration responses in the study are 0) inter-regional migration; 1) return outmigration; 2) re-emigration to an EU28 country; 3) re-emigration to a non-EU28 country; 4) outmigration with an unknown destination. We estimated a multinomial logit regression to model the probability of adopting a particular type of

• utmigration (1 to 4) in respect to changing the region of residence within Spain (0). The

controls included individual characteristics, such as sex, age, place of birth, citizenship, and period of the movement (2006-2008 pre-crisis, 2009-2010 economic collapse, 2011-2013 recession), together with context characteristics for regions of origin, such as: unemployment and activity rates by sex and birthplace (EU28 and extra-EU28), total population participation rate by economic sector and sex (services, constructions, industry, agriculture), hourly wage, and Latin American population. In this case, the information on movements and personal characteristics of migrants correspond to the Residential Variation

• Statistics. The information for the 19 Spanish regions in use6 corresponds to the estimations

carried out by the National Statistics Office based on Spanish Labor Force. Regarding the definition adopted here for internal migration, the reader should bear in mind that we are discussing the changes between major administrative geographical level, i.e. the Autonomous Community (onwards referred to as region), excluding from the analysis the changes that occurred within regions7. This decision responds to the availability of published estimations on unemployment and activity rates, which were included as controls for the context of origin in the multivariate analysis.

6 Andalusia, Catalonia, Community of Madrid, Valencian Community, Galicia, Castile and Leon, Basque

Country, Castile-La Mancha, Canary Islands, Region of Murcia, Aragon, Extremadura, Balearic Islands, Asturias, Navarre, Cantabria, La Rioja, CEU28ta, and Melilla.

7 Inter and intra provinces migration of foreign-born was discussed by (Quintero-Lesmes (2015) and Recaño

& de Miguel Luken (2012).

9 Transformations in the intensity of internal and international migration of immigrants The response to the crisis in all groups is the contraction of long-distance internal migration

• propensities. Despite this affected both Spanish and foreign-born population, the magnitude
• f the decline has been greater among the latter, especially for Latin Americans (Table 1)

There are at least two interpretations of these findings. On one hand, the Spanish-born population, who might have major social networks in different regions of the country, could have sought for employment opportunities in other areas, feeling more confident to migrate than the foreign immigrants. In fact, one of the results of the crisis is the return of the natives from the southern regions of Spain to seasonal work in agriculture: collecting olives

• r enrolling the vintage niche that was occupied in previous periods by population born
• abroad. On the other hand, the reduction in foreign-born internal migration could obey to a

conservative strategy of staying within a known area to avoid competition with natives who are also seeking for jobs. Table 1. Inter-regional migration rates, outmigration rates, and Outmigration growth index (ref. 2006-2007) by place of birth. Spain, 2006-2013

Latin Americans Spanish Rest of foreign born Total foreign born Period

I O Index O I O Index O I O Index O I O Index O 2006-2007

39.4 33.3 100% 9.1 0.6 100% 35.5 27.8 100% 37.0 29.9 100%

2008-2010

34.0 51.3 154% 8.5 0.9 142% 27.3 40.5 146% 29.8 44.6 149%

2011-2013

26.5 54.1 162% 8.6 1.4 220% 23.2 50.9 183% 24.4 52.1 174%

Note: I = inter-regions migration rate; O = outmigration rate; Index O = Index for outmigration growth in respect to values of outmigration rate for 2006-2007. Source: Spanish Population Register and Statistics of Residential Variation (National Statistics Office, 2006- 2013).

Table 2. Migration flows by type of movement and share of outmigration in total migration, by place of birth. Spain, 2006-2013

Latin Americans Spanish Rest of foreign born Total foreign born Period I O % O I O % O I O % O I O % O

2006-2007 81347 68789 45.8% 364450 24799 6.4% 116580 91093 43.9% 197927 159882 44.7% 2008-2010 82178 123977 60.1% 343482 35364 9.3% 109237 162010 59.7% 191414 285987 59.9% 2011-2013 64335 131302 67.1% 349596 55223 13.6% 97425 213852 68.7% 161760 345154 68.1%

Note: I = inter-regions migration flow; O = outmigration flow; % O = Share of total migration (inter-region +

• utmigration) that corresponds to outmigration.

Source: own elaboration based on Statistics of Residential Variation (National Statistics Office, 2006-2013).

In contrast to what happened to internal migration, the outmigration from Spain has grown steadily in all groups, including Spanish, Latin American and other foreign-born. In the last two, it becomes the dominant form of mobility since the beginning of the crisis and does not stop its increase in 2011-2013. Outmigration represented more than 67% of the total migration among foreign-born and accounted for only 13% of total mobility of Spaniards in 2011-2013 (Table 2). Among the natives, the increase of outmigration rate has been very significant –doubling the figures before the crisis- but still is a residual form of mobility. Regarding the foreign-born, the Latin Americans rates of emigration are significantly higher than for the rest of the foreign-born population, whereas the speed of the growth has been below the rest of the foreign-born population in recent times (Table 1).

10 In short, we found at least four remarkable transformations. First, a contraction of interregional migration in all origins but with different intensities for native and foreign- born individuals. Second, a general expansion of outmigration. Third, a lower propensity of

• utmigration of Latin Americans among the group of foreign-born for the last period

(2011-2013). Fourth, the growth in external mobility of Spanish people, although it remains as a residual phenomenon in absolute figures. In Table 3 we decompose outmigration in its different forms: return, remigration, and

• utmigration with ignored destination. In a general context of the growth of outmigration,

the migrations from Spain with unknown destination are the strongest form of international mobility among foreign-born individuals. However, the evolution is clearly different between Latin-Americans and the rest of immigrants. While this kind of outflow decreases since 2010 among Latin Americans, it continues growing for the rest of foreign origins. Table 3. Migration flows and rates type of movement and place of birth. Spain, 2006-2013

Latin-Americans Flow Return migration Remigration to EU28 Remigration to non-EU28 Unknown destination Total 2006-2007 10410 981 934 56463 68788 2008-2010 17087 1613 1377 103900 123977 2011-2013 25958 2955 2432 99955 131300 Rates 2006-2007 5.0 0.5 0.5 27.3 33.3 2008-2010 7.1 0.7 0.6 43.0 51.3 2011-2013 10.7 1.2 1.0 41.2 54.1 Rest of foreign born Flow Return migration Remigration to EU28 Remigration to non-EU28 Unknown destination Total 2006-2007 15505 1809 877 72882 91073 2008-2010 22540 2781 1151 135523 161994 2011-2013 23739 4498 1488 184106 213831 Rates 2006-2007 4.7 0.6 0.3 22.2 27.8 2008-2010 5.6 0.7 0.3 33.9 40.5 2011-2013 5.7 1.1 0.4 43.8 50.9

Source: Spanish Population Register and Statistics of Residential Variation (National Statistics Office, 2006- 2013).

The second dominant form of mobility among foreigners has been international return. In 2011-2013 the return migration rate of Latin Americans doubled the one from 2006-2007, and this growth was significantly lower for the rest of foreign origins. The third form of mobility corresponds to re-emigration. This is a growing phenomenon among the different groups considered. Latin-Americans and the rest of foreigners show similar propensities to re-emigration to the EU28 but differ when the outflow directs to the countries outside the EU28. In this case, Latin Americans migrate at similar intensities to both the EU28 and other states, while the latter flow is significantly lower for the rest of the foreign-born population.

11 Determinants of inter-region and international migration of immigrants We estimated four different models. The first was specified for all foreign-born population and includes controls for individual characteristics, such as sex, age, citizenship, birthplace and time of migration. The second model is similar to the previous but is restricted to the Latin American-born individuals. The third model, estimated for all foreign-born origins, differs from the first one in incorporating context variables in addition to individual

• characteristics. Finally, the fourth model replicates the prior only for Latin American
• immigrants. We display the results in four tables, one for every outcome in reference to

internal migration (reference category of dependent variable). Each table reports the odds ratios and standard errors by outcome for the all estimated models. Men are more prone to return to their country of birth in respect of women, as shown by the similar odds ratio obtained for the models for all foreign-born and for Latin-Americans (Table 4). Regarding age, those aged 45 and over have greater probabilities of experiencing return in respect to internal migration, compared to the younger age groups. Among Latin- Americans, the effect of age was steeper at advanced age compared to the same effect for all foreign-born origins ( in Spain (Aysa-Lastra & Cachón 2012). Activity rates by sector show a little adverse effect, reducing the odds for a return around 2% in construction, 3% and 4% for industry and agro-economic sector, respectively. A less expected result concerns the positive effect of dependent workers´ hourly wage, which increases the probability of returning about 2.8% but reduces the chance of a return in a 4% when only Latin-Americans are considered. Finally, the effect of social networks, captured by the log

• f the Latin American population shows that the larger the concentration of Latin

Americans the lower the probability of returning since this network might be able to provide information and support to resist the economic recession within Spain.

12 ). The probability of return increased in 2009-2010, at the onset of the crisis, but its largest effect corresponds to the recession in 2011-2013 (Table 4 in Spain (Aysa-Lastra & Cachón 2012). Activity rates by sector show a little adverse effect, reducing the odds for a return around 2% in construction, 3% and 4% for industry and agro-economic sector, respectively. A less expected result concerns the positive effect of dependent workers´ hourly wage, which increases the probability of returning about 2.8% but reduces the chance of a return in a 4% when only Latin-Americans are considered. Finally, the effect of social networks, captured by the log of the Latin American population shows that the larger the concentration of Latin Americans the lower the probability of returning since this network might be able to provide information and support to resist the economic recession within Spain.

13 ). We found some differences in the size of the effect between the models estimated for Latin-Americans and for the total foreign-born population when the effect of time is

• considered. All immigrants experienced larger odds for return after the crisis, but this was

particularly outstanding among Latin-Americans in 2011-2013 when the odds of return more than doubled the values observed for the pre-crisis period (models 2 and 4 in Table 4). All non-European origins have shown a lower probability of returning in respect to experience inner migration when compared to Europeans (reference group). However, Latin-Americans were the group with the lowest distance from Europeans (only 20% less) respect to Africans, Asians, and Oceanians (more than 70%) (models 1 and 3 in Table 4). Among Latin-Americans, the only national origins with a higher risk of return respect to Venezuelans (reference group) are Brazilians and Paraguayans (models 2 and 4 in Table 4). Regarding citizenship, Latin Americans with Spanish citizenship have more chances to return than those with other passports (Extra-EU28 or EU28 citizenship). Therefore, Spanish citizenship stands as a predictor of Latin Americans´ return, which could have at least two possible explanations. On one hand, it might be that only those foreigners that have already naturalized feel as having completed their migration plan and able to leave the country or EU. On the other hand, and this might be the most likely hypothesis, the foreign- born people with Spanish citizenship have greater incentives to report their return because, for example, they may receive a Spanish Social Security subside, and/or because they are meant to hold their right to vote abroad. We should recall that this outcome, where the destination is reported, concerns only 17% of all the international movements of foreign- born individuals in the Residential Variation Statistics. The variables regarding the region of departure have a minor effect compared to the individual characteristics (models 3 and 4 in Table 4). Even though, the effect of unemployment on return reveals a different behavior whether considering men or women, all foreign-born or only Latin-Americans. Spanish male unemployment rate increases by 8% the probability of return migration respect to internal migration. Conversely, increases in female unemployment rates –for any origin- slightly reduce the likelihood of return for

• women. This evidence supports the literature pointing to females as the most tolerant group

to the lack of employment in Spain (Aysa-Lastra & Cachón 2012). Activity rates by sector show a little adverse effect, reducing the odds for a return around 2% in construction, 3% and 4% for industry and agro-economic sector, respectively. A less expected result concerns the positive effect of dependent workers´ hourly wage, which increases the probability of returning about 2.8% but reduces the chance of a return in a 4% when only Latin-Americans are considered. Finally, the effect of social networks, captured by the log

• f the Latin American population shows that the larger the concentration of Latin

Americans the lower the probability of returning since this network might be able to provide information and support to resist the economic recession within Spain.

14 Table 4. Multinomial regression odds ratios and standard errors for the estimation of return in reference to inter-regional migration. Spain, 2006-2013

All foreign-born // Latin- Americans All foreign-born // Latin- Americans (1) (2) (3) (4) Sex (ref. Man) Woman 0.884*** 0.807*** 0.894*** 0.815*** (0.00435) (0.00617) (0.00461) (0.00649) Age (ref. 16-29) 30-44 0.934*** 1.046*** 0.926*** 1.032*** (0.00536) (0.00943) (0.00556) (0.00969) 45-59 1.418*** 1.580*** 1.410*** 1.571*** (0.0100) (0.0174) (0.0104) (0.0180) 60+ 3.615*** 4.329*** 3.603*** 4.317*** (0.0337) (0.0657) (0.0355) (0.0680) Period (ref. 2006-2008) 2009-2010 1.345*** 1.474*** 1.068*** 1.239*** (0.00877) (0.0155) (0.0195) (0.0354) 2011-2013 1.738*** 2.295*** 1.297*** 1.758*** (0.0100) (0.0213) (0.0317) (0.0676) Citizenship (ref. Spanish cit.) Extra-EU28 0.852*** 0.742*** 0.855*** 0.741*** (0.00592) (0.00677) (0.00625) (0.00706) EU28 1.088*** 0.461*** 1.105*** 0.453*** (0.00972) (0.0111) (0.0104) (0.0113) Birthplace (ref. EU28) Latin America 0.802*** 0.793*** (0.00631) (0.00653) Africa 0.284*** 0.291*** (0.00301) (0.00327) Asian and Oceania 0.262*** 0.257*** (0.00369) (0.00375) Country of birth (ref. Venezuela) Argentina 1.010 1.007 (0.0168) (0.0175) Bolivia 1.014 1.041* (0.0180) (0.0194) Brazil 1.170*** 1.189*** (0.0214) (0.0227) Colombia 0.458*** 0.465*** (0.00769) (0.00810) Cuba 0.439*** 0.440*** (0.00988) (0.0102) Dominican Rep. 0.399*** 0.409*** (0.00893) (0.00941) Ecuador 0.864*** 0.870*** (0.0140) (0.0148) Paraguay 1.063** 1.095*** (0.0231) (0.0251) Peru 0.703*** 0.709*** (0.0140) (0.0147) Uruguay 1.025 0.996 (0.0223) (0.0223)

15

Extra-EU28 female Unemployment rate

0.984*** 0.991*** (0.000852) (0.00143)

Spanish female unemployment rate 0.925*** 0.932*** (0.00224) (0.00350) Extra-EU28 male 1.003** 1.000 unemployment rate (0.000795) (0.00127) Spanish male unemployment rate 1.086*** 1.083*** (0.00277) (0.00430) Construction 0.980*** 0.971*** male Activity rate (0.00200) (0.00318) Agro 0.962*** 0.960*** male Activity rate (0.00229) (0.00355) Services 0.975*** 0.966*** male Activity rate (0.00134) (0.00212) Industry 0.970*** 0.956*** female Activity rate (0.00369) (0.00580) Services 0.989*** 0.998 female Activity rate (0.00262) (0.00419) Log (Latin American 0.955*** Population) (0.00639) Hourly wage for dependent 1.028*** 0.964** Workers (0.00723) (0.0107) Gross Domestic Product 0.999 1.039*** per capita (0.00310) (0.00501) AIC 4165652.5 1530882.8 3760235.7 1405054.7 BIC 4166260.4 1531771.0 3761442.7 1406494.9 R2_p 0.0995 0.130 0.111 0.139 N 2335816 879522 2133680 817585

Exponentiated coefficients; Standard errors in parentheses

* p < 0.05, ** p < 0.01, *** p < 0.001

Attending to the findings for the likelihood for onwards mobility within the European Union, we observe that sex has no effect, whereas age rises from less than 20% to about 30% the chances of Latin Americans aged 30 to 44 to experience this sort of mobility (Table 5). This effect declines right after 45 years old, which suggest that -in contrast to return- this a risk worth taking for the people in young adult ages. In the models controlling for individual characteristics (1 and 2 in Table 5), the odds of emigration to the EU28 have increased about 33% during the crisis eruption and were around 90% greater during the recession (2011-2013). In opposition, when we controlled for hourly wage, sector participation rates, and unemployment rates, the effect of period disappears for 2011-2013 and becomes negative for 2009-2010. This suggests that remigration was not preferred to internal migration at the onset of the crisis. Given that the region of origin variables are a sort of interaction between time and territory8, it could be interpreted that the greater its impact, the lower the effect of time by itself. Then the probability to migrate to a third EU28 country is more related to the region of origin and its levels of employment, activity, or wage, than to the single effect of the crisis eruption captured by the period variables.

8 These variables account for the value of the indicator at regional level and vary along the three periods in

study (2006-2008, 2009-2010, and 2011-2013). This means that these controls are time and region dependent.

16 Table 5. Multinomial regression odds ratios and standard errors for the estimation of remigration to EU28 countries in reference to inter-regional migration. Spain, 2006-2013

All foreign-born // Latin-Americans All foreign-born // Latin-Americans (1) (2) (3) (4) Sex (ref. Man) Woman 0.994 1.003 0.985 1.008 (0.0111) (0.0184) (0.0115) (0.0191) Age (ref. 16-29) 30-44 1.193*** 1.319*** 1.175*** 1.302*** (0.0153) (0.0274) (0.0157) (0.0280) 45-59 1.049** 0.989 1.026 0.955 (0.0176) (0.0278) (0.0180) (0.0279) 60+ 1.060* 0.693*** 1.009 0.624*** (0.0285) (0.0422) (0.0288) (0.0402) Period (ref. 2006-2008) 2009-2010 1.333*** 1.255*** 0.829*** 0.842* (0.0213) (0.0344) (0.0376) (0.0659) 2011-2013 1.960*** 1.936*** 1.060 1.081 (0.0267) (0.0452) (0.0627) (0.116) Citizenship (ref. Spanish cit.) Extra-EU28 0.0797*** 0.110*** 0.0841*** 0.114*** (0.00113) (0.00252) (0.00125) (0.00269) EU28 0.699*** 1.237*** 0.637*** 1.033 (0.0128) (0.0373) (0.0121) (0.0328) Birthplace (ref. EU28) Latin Americas 2.333*** 1.968*** (0.0452) (0.0395) Africans 6.435*** 6.077*** (0.135) (0.132) Asia and Oceania 3.786*** 3.036*** (0.103) (0.0856) Country of birth (ref. Venezuela) Argentina 0.764*** 0.778*** (0.0290) (0.0310) Bolivia 0.562*** 0.578*** (0.0359) (0.0383) Brazil 1.639*** 1.691*** (0.0684) (0.0734) Colombia 1.175*** 1.169*** (0.0402) (0.0418) Cuba 0.625*** 0.617*** (0.0325) (0.0331) Dominican Rep. 0.789*** 0.744*** (0.0350) (0.0342) Ecuador 1.173*** 1.196*** (0.0421) (0.0451) Paraguay 0.422*** 0.407*** (0.0448) (0.0459) Peru 1.633*** 1.500*** (0.0652) (0.0625) Uruguay 0.549*** 0.554*** (0.0338) (0.0351) Extra-EU28 female 1.008*** 0.991*

17

unemployment rate (0.00191) (0.00419) Spanish female 0.899*** 0.952*** unemployment rate (0.00507) (0.00957) Extra-EU28 male 0.997 1.000 unemployment rate (0.00204) (0.00361) Spanish male 1.140*** 1.114*** unemployment rate (0.00711) (0.0120) Construction 1.025*** 0.997 male Activity rate (0.00520) (0.00886) Agro 0.996 1.019 male Activity rate (0.00608) (0.0110) Services 1.017*** 1.014* male Activity rate (0.00307) (0.00584) Industry 1.053*** 1.104*** female Activity rate (0.00967) (0.0180) Services 1.011 1.073*** female Activity rate (0.00689) (0.0133) Log (Latin American 1.116*** Population) (0.0214) Hourly wage for dependent 1.129*** 1.102** Workers (0.0195) (0.0346) Gross Domestic Product 0.993 0.982 per capita (0.00773) (0.0134) AIC 4165652.5 1530882.8 3760235.7 1405054.7 BIC 4166260.4 1531771.0 3761442.7 1406494.9 R2_p 0.0995 0.130 0.111 0.139 N 2335816 879522 2133680 817585

Exponentiated coefficients; Standard errors in parentheses

* p < 0.05, ** p < 0.01, *** p < 0.001

The effect of EU28 citizenship differs compared to the findings for return migration. A Latin-American holding an EU28 passport has about 24% more chances to emigrate to the EU28 than one having Spanish citizenship. However, for the total of the foreign-born population, the EU28 passport reduces about 30% the odds of experiencing this kind of mobility in respect to Spanish citizens. For all origins, an extra-EU28 passport reduces the likelihood inhibits this type of migration response in respect to Spanish citizens. The evidence from models 2 and 4 in Table 5 show that remigration to another EU28 country is a preferred form of migration, compared to return and internal migration for all non-EU28 foreign origins. This is particularly the case for Africans, followed by Asians and Oceanians at second place, and –at last- for Latin-Americans. Among them, the origins with a larger likelihood of experiencing this movement are Brazil, Peru, Colombia, and Ecuador respect to Venezuela. These are some of the origins that massively acquired the Spanish citizenship being granted free mobility in the Schengen area too (Quintero-Lesmes 2015). The probability of remigration to another EU28 country increases if the female unemployment rate for extra-EU28 citizens, the male unemployment rates, the male activity rate in construction and industry, and the female industry participation rate, increases too. In general, the dynamism of the economy, observed in the participation rates

• f sectors such as manufacturing or construction, as well as at the wages, configures a

setting were remigration is more likely than internal migration. Conversely, internal migration might have been preferred by foreign-born individuals residing in areas where agriculture was more important.

18 Finally, we should add that for Latin-Americans the determinants that played a vital role in the region of origin level excluded extra-EU28 female unemployment rate, which in this case reduces the probability of remigration, and included the services activity rate for women. When we discuss emigration to non-EU28 countries or to an unknown destination, women have a lower probability of outmigration and age shows an increasing effect (Tables 6 and 7). Its positive effect is larger for outmigration with unknown destination among people aged 60 and over (Table 7), which coincides with the findings for return migration. The probability of migrating to non-EU28 countries increases during the crisis outbreak and even more during the recession phase among the foreign-born population (Table 6). However, this result only holds when individual characteristics were considered (models 1 and 2 in Table 6), and the effect of time works in the opposite direction when the region of

• rigin variables are considered (models 3 and 4 in Table 6). This is similar to results found

for onwards migration to the rest of EU28. Regarding outmigration with an unknown destination, and similarly to what was observed for return migration, the positive effect of time holds for all models (Table 7). The EU28 or the extra-EU28 citizenship reduces the probability of emigration to a non- EU28 country when compared to Spanish citizenship of all origins (Table 6). In other words, all non-Spanish citizens have greater chances to leave Spain without communicating their destination respect to experiencing internal migration. Despite the size difference in the effect for EU28 citizens and non-EU28 citizens, the general result coincides with the hypothesis stating that individuals who move out of the EU28 –either as a return or

• nwards international migration- are highly likely to join this fourth flow. Bear in mind that

people with Spanish or EU28 citizenship are motivated to communicate their arrival at Spanish Consulate abroad while those heading to the United States or Latin America might not. Latin Americans are the only origin with a larger probability of experiencing outmigration

• ut of the EU28 with respect to EU28. The odds of them is 20% greater than the chances

for EU28 when the individual characteristics are considered, and 10% when other variables for the region of origin are included. Regarding outmigration with an unknown destination, all foreign-born have a lower probability of experiencing this flow in respect of EU28, but Latin Americans have only 30% fewer chances than this reference group, while the likelihood for other origins is about 50%.

19 Table 6. Multinomial regression odds ratios and standard errors for the estimation of remigration to non-EU28 countries in reference to inter-regional migration. Spain, 2006- 2013

All foreign-born // Latin- Americans All foreign-born // Latin- Americans (1) (2) (3) (4) Sex (ref. Man) Woman 0.947*** 0.945** 0.951*** 0.957* (0.014) (0.019) (0.014) (0.019) Age (ref. 16-29) 30-44 1.416*** 1.426*** 1.391*** 1.403*** (0.026) (0.035) (0.026) (0.035) 45-59 1.763*** 1.504*** 1.687*** 1.459*** (0.038) (0.045) (0.038) (0.044) 60+ 2.925*** 1.745*** 2.733*** 1.641*** (0.083) (0.076) (0.081) (0.074) Period (ref. 2006-2008) 2009-2010 1.156*** 1.171*** 0.669*** 0.628*** (0.024) (0.034) (0.039) (0.052) 2011-2013 1.630*** 1.780*** 0.866 0.838 (0.028) (0.045) (0.067) (0.096) Citizenship (ref. Spanish cit.) Extra-EU28 0.128*** 0.147*** 0.133*** 0.149*** (0.002) (0.003) (0.002) (0.004) EU28 0.229*** 0.215*** 0.232*** 0.195*** (0.006) (0.014) (0.006) (0.013) Birthplace (ref. EU28) Latin Americas 1.234*** 1.103*** (0.026) (0.024) Africans 0.433*** 0.425*** (0.015) (0.015) Asia and Oceania 0.895** 0.779*** (0.032) (0.029) Country of birth (ref. Venezuela) Argentina 0.575*** 0.604*** (0.023) (0.025) Bolivia 0.504*** 0.530*** (0.029) (0.032) Brazil 0.547*** 0.590*** (0.031) (0.034) Colombia 0.474*** 0.501*** (0.018) (0.019) Cuba 1.627*** 1.665*** (0.059) (0.063) Dominican Rep. 0.580*** 0.534*** (0.026) (0.025) Ecuador 0.545*** 0.573*** (0.021) (0.023) Paraguay 0.712*** 0.735*** (0.053) (0.057) Peru 0.811*** 0.752*** (0.036) (0.035) Uruguay 0.682*** 0.688*** (0.038) (0.040)

20

Extra-EU28 female 0.984*** 0.991 unemployment rate (0.003) (0.004) Spanish female 0.895*** 0.889*** unemployment rate (0.007) (0.009) Extra-EU28 male 1.004 1.003 unemployment rate (0.003) (0.004) Spanish male 1.126*** 1.117*** unemployment rate (0.009) (0.013) Construction 1.012 0.998 male Activity rate (0.007) (0.009) Agro 0.988 1.021 male Activity rate (0.008) (0.012) Services 1.033*** 1.023*** male Activity rate (0.004) (0.006) Industry 0.990 0.964 female Activity rate (0.013) (0.018) Services 0.964*** 0.976 female Activity rate (0.009) (0.013) Log (Latin American 1.193*** Population) (0.027) Hourly wage for dependent 1.130*** 1.190*** Workers (0.027) (0.041) Gross Domestic Product 0.981 0.962* per capita (0.010) (0.015) AIC 4165652.5 1530882.8 3760235.7 1405054.7 BIC 4166260.4 1531771.0 3761442.7 1406494.9 R2_p 0.0995 0.130 0.111 0.139 N 2335816 879522 2133680 817585

Exponentiated coefficients; Standard errors in parentheses

* p < 0.05, ** p < 0.01, *** p < 0.001

In most of the cases, the odds of migrating to a non-EU28 country are rare except for Cubans (Table 6). This origin has around 62% more chances to experience this type of migration respect to Venezuelans (reference) in model 2, and 66% when the variables for the region of origin are considered too (model 4 in Table 6). A hypothesis could be that the main country of destination, in this case, is the United States. For the rest of origins, the probability reduces about 50% except for Peruvians (-20%) and Paraguayans (-30%) who have not such small chances of heading a non-EU28 state. The outmigration with an unknown destination is not so rare for Paraguayans, Bolivians, Peruvians or Argentineans, and it is even likely among Brazilians, for whom the odds surpass the 20% chances of leaving Spain without registering their departure from the Spanish Population Register (Table 7). Uruguayans also have a minuscule probability of experiencing this migration, but the effect turns negative at the fully controlled model (model 4 in Table 7).

21 Table 7. Multinomial regression odds ratios and standard errors for the estimation of

• utmigration with an unknown destination in reference to inter-regional migration.

Spain, 2006-2013

All foreign-born // Latin- Americans All foreign-born // Latin- Americans (1) (2) (3) (4) Sex (ref. Man) Woman 0.936*** 0.843*** 0.942*** 0.844*** (0.003) (0.005) (0.003) (0.005) Age (ref. 16-29) 30-44 1.102*** 1.106*** 1.097*** 1.105*** (0.004) (0.007) (0.004) (0.007) 45-59 1.912*** 2.018*** 1.906*** 2.042*** (0.009) (0.015) (0.009) (0.016) 60+ 4.792*** 6.238*** 4.849*** 6.470*** (0.034) (0.075) (0.037) (0.081) Period (ref. 2006-2008) 2009-2010 1.756*** 1.645*** 1.489*** 1.530*** (0.00716) (0.0112) (0.0178) (0.0308) 2011-2013 2.218*** 2.267*** 1.904*** 2.110*** (0.00824) (0.0144) (0.0305) (0.0574) Citizenship (ref. Spanish cit.) Extra-EU28 13272.9*** 42265.9*** 14305.2*** 38673.9*** (3831.8) (29885.9) (4524.0) (27338.5) EU28 9644.3*** 45543.7*** 10993.2*** 43626.3*** (2784.6) (32207.3) (3477.0) (30843.2) Birthplace (ref. EU28) Latin Americas 0.771*** 0.787*** (0.005) (0.005) Africans 0.525*** 0.568*** (0.003) (0.004 Asia and Oceania 0.565*** 0.560*** (0.004) (0.004) Country of birth (ref. Venezuela) Argentina 0.921*** 0.883*** (0.012) (0.012) Bolivia 0.697*** 0.669*** (0.009) (0.009) Brazil 1.275*** 1.220*** (0.017) (0.017) Colombia 0.427*** 0.416*** (0.005) (0.006) Cuba 0.575*** 0.581*** (0.009) (0.009) Dominican Rep. 0.478*** 0.488*** (0.007) (0.008) Ecuador 0.549*** 0.536*** (0.007) (0.007) Paraguay 0.939*** 0.933*** (0.014) (0.015) Peru 0.648*** 0.638*** (0.009) (0.009) Uruguay 1.034* 0.967* (0.017) (0.016)

22

Extra-EU28 female 1.002*** 1.007*** unemployment rate (0.000) (0.001) Spanish female 0.938*** 0.952*** unemployment rate (0.001) (0.003) Extra-EU28 male 0.980*** 0.978*** unemployment rate (0.000) (0.001) Spanish male 1.113*** 1.099*** unemployment rate (0.002) (0.003) Construction 1.040*** 1.037*** male Activity rate (0.001) (0.002) Agro 0.975*** 0.970*** male Activity rate (0.001) (0.003) Services 0.988*** 0.977*** male Activity rate (0.001) (0.001) Industry 1.059*** 1.058*** female Activity rate (0.002) (0.004) Services 1.017*** 1.031*** female Activity rate (0.002) (0.003) Log (Latin American 0.947*** Population) (0.004) Hourly wage for dependent 1.240*** 1.206*** Workers (0.006) (0.009) Gross Domestic Product 0.958*** 0.956*** per capita (0.002) (0.003) AIC 4165652.5 1530882.8 3760235.7 1405054.7 BIC 4166260.4 1531771.0 3761442.7 1406494.9 R2_p 0.0995 0.130 0.111 0.139 N 2335816 879522 2133680 817585

Exponentiated coefficients; Standard errors in parentheses

* p < 0.05, ** p < 0.01, *** p < 0.001

Regarding the variables that account for the determinants at the region of origin, once again, the size of the effects is minuscule either for outmigration with the unknown destination or remigration to non-EU28 countries. The promoters of outmigration to non- EU28 countries were male unemployment rates –especially those for for Spanish men, which increases about 12.6% the odds for all foreign origins, and 11.7% for Latin- Americans -, and male construction or services activity rates, which particularly contributed to the departure of Latin Americans from Spain to non-EU28 destinations (Table 6). For

• utmigration with ignored destination, the largest effect corresponds to the hourly wage for

dependent workers, which increases the odds about 24% for all immigrants and 20% for Latin-Americans (Table 7). The second largest effect is Spanish male unemployment rate, increasing about 11% the odds for all foreign origins and 9% for Latin-Americans. The rest

• f controls, including gross domestic product per capita and the Latin American population,

show a negative contribution to the probability of leaving Spain without informing the destination (Table 7).

23 Conclusions To this point, we have described the migration intensities and the determinants of the different responses adopted by immigrants to face the recent economic crisis in Spain, between 2006 and 2013, before the crisis, during its onset, and in the following years. The main purpose of this analysis has been to assess when and who preferred the different kinds

• f outmigration from Spain over internal migration.

Our findings suggest that all origins, either Spanish-born and foreign born, have contracted inter-regional migration right after the crisis hit. The magnitude of the decline in internal migration rates has been particularly sharp among Latin Americans, but their fall was processed in two stages: the first, a decline in 2009-2010, and later, a sharper decline in the recession (2011-2013). This could mean that Latin Americans, above all origins, reduced the internal migration to face economic hardship but after the recession had installed the resentment of internal mobility became a permanent feature. While internal migration has declined, outmigration has increased after the crisis. However, the speed of this decline varies by type of movement and origin. Return migration kept growing since 2008 to 2013 among Latin Americans, whereas it increased soon after the beginning of the crisis and stagnated during the recession the rest of foreign-born. Onwards international mobility was discussed here too, assessing its two forms: remigration to EU28 countries, and remigration to non-EU28 countries. The first of these two kinds of

• utmigration grew significantly for Latin Americans, who also engaged in the increase of

remigration to non-EU28 –in contrast to the rest of foreign-born, who were more oriented to remigration within the EU28-. The second form of outmigration discussed here was the

• utflow with an unknown destination. This type of international migration introduces the

largest differences by origin in migration responses adopted during the crisis. Outmigration with ignored destination first grew among Latin Americans, from 2006-2007 to 2008-2010, but declined later, during the recession. In contrast, its growth has been constant over all the period for the rest of the foreign-born population. In the multivariate analysis of the characteristics that orient migrants in preferring different forms of out-migration over internal migration, we showed that the various stages of the economic cycle had transformed the preferences. Return migration and outmigration with unknown destination increased significantly with respect to internal mobility in 2011-2013, while remigration (to EU28 and to non-EU28) has gradually lost its status as preferred response with respect to internal migration. In fact, remigration was a more likely behavior before the crisis when compared to internal migration, return or other forms of

• utmigration.

With reference to the individual characteristics that are positively associated with the different forms of outmigration, we found sex, age, birthplace, and citizenship, to have a different effect depending on the type of outmigration considered. First, return, remigration to a non-EU28, and outmigration with unknown destination are typically male migration responses, whereas remigration to EU28 and internal migration have a more balanced composition by sex.

24 Second, return, remigration to non-EU28 countries, and outmigration with an unknown destination, share a similar age profile, concentrated in the advanced age groups. The odds

• f return and outmigration with unknown destination is especially high among people aged

60 or older. Contrary, remigration to EU28 is more likely among younger adults (30-44 years old) than among elders.. Third, the origin -defined in reference to the place of birth-, showed that Europeans are the foreign origin with more likelihood of returning or leaving Spain without reporting their

• departure. Latin Americans, in spite of the fewer probabilities of experiencing these types
• f migration if compared to Europeans, have more chances than Africans, Asians, and
• Oceanians. Among Latin Americans, Brazilians stand out as the origin with greater odds of

preferring this type of migration to internal migration. In contrast, Africans, followed by the

• ther non-EU origins, among which Brazilians, Peruvians, Colombians, and Ecuadorians

stand out, are more likely to migrate within the EU. Last, Latin Americans are the only

• rigin with larger chances of migration to a non-EU country after Europeans. Among them,

the odds is especially high for Cubans. Fourth, non-Spanish citizens (both, EU28 and extra- EU28) have preferred internal migration to return or remigration beyond the borders of the EU28, in respect to Spanish

• citizens. One should bear in mind that this result is more an outcome of an observation

effect than an actual finding. It is not that non-Spanish citizenship inhibits return or remigration, instead, as it was already pointed out, Spanish citizens (foreign-born or Spanish-born) may find more incentives to report their departure from Spain. This explains why non-Spanish citizens have greater chances to experience outmigration without reporting their departure from Spanish municipality nor their arrival to a Spanish consulate abroad. In reference to the determinants for the region of departure, findings suggest that their effect on the type of migration response adopted was minor compared to the size and significance of the individual characteristics. However, male unemployment –either of Spanish or extra-EU citizens- contributed to the increase of all international migration responses respect to internal migration. The sector participation rates, especially for construction and services, were another predictor of remigration. The impact of hourly wage or same origin population concentration did not show any significant systematic

• effect. The null effect of social support networks to inhibit outmigration, could be

interpreted in two forms: 1) it could point to their inability to stop the emigration when it comes to a context of widespread economic deterioration; or 2) its effect is hardly perceptible to a level of analysis at a region level and might be better captured at a smaller scale (province or municipality). In short, we find that outmigration with an unknown destination, which represents more than 80% of the total outmigration for foreign-born, shares a similar demographic profile with return migration in respect to sex, age, origins, and follow a similar trend along periods, responding to similar context determinants too. The only feature that tears them apart concerns the effect of citizenship: while Spanish citizens have larger odds to report their return, non-Spanish citizens have larger odds to leaving Spain unnoticed. Except for this feature, our findings point to outmigration with unknown destination as an

25

• utmigration with a high share of return migration. However, this is no more than a

hypothesis, and further research should be carried in order to contrast it. Finally, we can conclude that return and outmigration with unknown destination are more associated to the economic recession, while internal migration –though it declined during the onset of the crisis- was the first reaction to the economic collapse, and remigration to EU was only preferred to internal migration in the years before the crisis. This result confirms our hypothesis to a certain extent. First, internal migration was preferred to return

• r outmigration at the beginning of the economic crisis, despite descriptives have shown

that this type of migration has declined since 2009. Second, return and outmigration with unknown destination –which behaves similar to return in the multivariate the analysis- were the main migration forces during the recession (2011-2013).

26 References Aysa-Lastra, M. & Cachón, L., 2012. Latino Immigrant Employment During the Great Recession : A Comparison of the United States and Spain. Norteamérica, 7(2), pp.7– 42. Bernardi, F., Garrido, L. & Miyar, M., 2011. The Recent Fast Upsurge of Immigrants in Spain and Their Employment Patterns and Occupational Attainment. International Migration, 49(1), pp.148–187. Cebolla-Boado, H., Miyar-Busto, M. & Muñoz-Comet, J., 2015. Is the Spanish Recession Increasing Inequality? Male Migrant-native Differences in Educational Returns Against Unemployment. Journal of Ethnic and Migration Studies, 41(5), pp.710–728. Cerrutti, M. & Maguid, A., 2014. Crisis y retorno. Los sudamericanos en España. Notas de Población, 99, pp.1–36. Cuadrado-Roura, J., 2015. El turismo, motor del crecimiento y de la recuperación de la economía española, Alcalá: Instituto Universitario de Análisis Económico y Social de la Universidad de Alcalá. Czaika, M., 2012. Migration in times of uncertainty. On the role of economic prospects, Oxford: IMI University of Oxford. Czaika, M. & de Haas, H., 2012. The Role of Internal and International Relative Deprivation in Global Migration. Oxford Development Studies, 0818(July 2015), pp.1–20. Czaika, M., Haas, H. De & De Haas, H., 2013. The effectiveness of immigration policies (WP). Population and Development Review, 39(3), pp.487–508. Durand, J. & Massey, D.S., 2003. Clandestinos: migración México-Estados Unidos en los albores del siglo XXI, Mexico: Miguel Ángel Porrua. Gil, F., 2010. Análisis de dos propuestas metodológicas para estimar las salidas de extranjeros en España: las bajas por caducidad padronales y la renovación de las tarjetas de residencia temporales. Estadística Española, 52(174), pp.277–309. Gonzalez-Ferrer, A., 2014. Retorno y reintegración de Latinoamericanos en Europa. In FIIAP, ed. Propuestas para vincular las políticas de migración y empleo. Madrid: FIIAP, pp. 53–90. Gurak, D. & Kritz, M., 2000. The interstate Migration of U.S. Immigrants: Individual and Contextual Determinants. Social Forces, 78(3), pp.1017–1039. IOM, 2012. Impactos de la crisis sobre la población inmigrante, Madrid: IOM. King, R., Skeldon, R. & Vullnetari, J., 2008. Internal and International Migration : Bridging the Theoretical Divide. Russell The Journal Of The Bertrand Russell Archives, (52), p.49. Available at: http://www.sussex.ac.uk/migration/documents/mwp52.pdf. Koolhaas, M., 2015. Migración internacional de retorno en Uruguay : magnitud, selectividad y reinserción laboral en tiempos de crisis económica internacional.

27 Universidad de la República. Koser, K. & Kuschminder, K., 2015. Comparative Research on the Assisted Voluntary Return and Reintegration of Migrants, Geneva: IOM. Larramona, G., 2013. Out-migration of immigrants in Spain. Population, 222(68), pp.249– 271. Mas-Giralt, R., 2016. Onward Migration as a Coping Strategy? Latin Americans Moving from Spain to the UK Post-2008. Population Space and Place, in press. Mateos, P. (Editor), 2015. Ciudadanía Múltiple y Migración. Perspectivas Latinoamericanas, México DF: CIESAS-CIDE. Muñoz Comet, J., 2013. ¿Qué trabajos ocupan quienes abandonan el desempleo? Diferencias entre españoles y extranjeros en un contexto de cambio económico. Revista Internacional de Sociología, 72, pp.353–376. Ortega, E., Domingo, A. & Sabater, A., 2013. El empadronamiento de la población extranjera en los municipios andaluces de 1998 a 2010. Cuadernos Geograficos, 52(1). Pajares, M., 2010. Inmigración y mercado de trabajo. Informe 2010 O. P. de la Inmigración, ed., Madrid: Observatorio Permanente de la Inmigración. Pandit, K., 1997. Demographic cycle effects on migration timing and the delayed mobility

• phenomenon. Geographical Analysis, 29(3), pp.439–450.

Pereira, S., 2012. Immigrant workers’ (im)mobilities and their re‐emigration strategies. Employee Relations, 34(6), pp.642–657. Prieto Rosas, V. & López Gay, A., 2015. Push and Pull Factors of Latin American

• Migration. In A. Domingo, A. Sabater, & R. Verdugo, eds. Demographic Analysis of

Latin American Immigrants in Spain. From Boom to Bust. Springer International Publishing, pp. 1–27. Prieto, V., 2016. Las consecuencias de quedarse y de volver en el empleo de los uruguyos que migraron a España. Notas de Población, 102, pp.149–176. Prieto, V. & Koolhaas, M., 2014. Retorno reciente y empleo. Los casos de Ecuador, México y Uruguay. In L. Gandini & M. Padrón, eds. Población y trabajo en América Latina: abordajes teórico-metodológicos y tendencias empíricas recientes. Montevideo: ALAP. Quintero-Lesmes, D.C., 2015. Las migraciones internas de latinoamericanos en España: evolución, determinantes e impacto territorial. Universitat Autonoma de Barcelona. Ramos, R. & Matano, A., 2013. Remittances, education and return migration. Evidence for immigrants in Spain, Barcelona: SEARCH. Recaño, J. & Jáuregui, A., 2014. Emigración exterior y retorno de latinoamericanos desde España: una visión desde las dos orillas (2002-2012). Notas de Población, 99, pp.177– 240.

28 Recaño, J. & de Miguel Luken, V., 2012. The Internal Migration of Foreign-born Population in Southern Europe: Demographic Patterns and Individuals Determinants. In N. Finney & G. Catney, eds. Minority Internal Migration in Europe. Farnham: Ashgate, pp. 239–262. Recaño, J., Roig, M. & de Miguel Luken, V., 2015. Spain: A New Gravity Centre for Latin American Migration. In A. Domingo, A. Sabater, & R. Verdugo, eds. Demographic Analysis of Latin American Immigrants in Spain. From Boom to Bust. Springer International Publishing, pp. 181–209. Rodríguez-Planas, N. & Nollenberger, N., 2014. A precarious position: The Labor Market Integration of New Immigrants in Spain, Geneva: MPI. Torre-Cantalapiedra, E. & Giorguli, S., 2015. Las movilidades interna y de retorno de los varones migrantes mexicanos a Estados Unidos en perspectiva longitudinal (1942- 2011 ). Revista Estudios Demográficos y Urbanos, 30(88), pp.7–43. Torres-Pérez, Francisco Moncusí, A. & Esteban, F., 2013. Ciudadanía, convivencia multicultural y crisis. Análisis comparativo de dos barrios de Valencia. In XI Congreso FES. Madrid. Vela Peón, F. & Cabezas, G., 2015. La inserción laboral de la migración de retorno: los casos de Brasil, Ecuador y México. In II Congreso Migración Internacional en América Latina y México. Toluca. Vidal-coso, E., Gil-alonso, F. & Domingo, A., 2012. La destrucción de empleo de migrantes y españoles (2007-2011): factores demográficos, sectoriales y territoriales, Bellaterra: Centre d´Estudis Demográfics. Zimmermann, K.F. & Zaiceva, A., 2012. Returning Home at Times of Trouble ? Return Migration of EU Enlargement Migrants during the Crisis. IZA DP, (7111), p.25.