Economic recession and the reverse of internal migration flows of Latin American immigrants in Spain

ABSTRACT Fluctuations in internal migration flows have been closely associated with changes in economic processes. However, foreign immigrants’ spatial responses to changing national economic conditions have not yet been sufficiently studied. This paper aims to analyse the impact of the Spanish economic downturn on Latin American immigrants’ mobility by examining their territorial patterns of internal migration. In order to elucidate the intensity and directionality of Latin Americans’ internal flows between 2004 and 2013, divided in two periods – before and during the crisis–, we relied on the data of the Residential Variation Statistics, and the creation of a typology of provinces according to each province’s socio-economic characteristics. We calculated the net migration rates for the main immigrant groups and generated origin–destination matrices. This study shows that Latin Americans’ internal flows in Spain have been affected by the economic downturn, revealing a decrease in intensity and a change in directionality. The economic crisis in Spain has caused a change in the previous dynamic of geographical dispersion and has become a new factor of differentiation in the distribution of the Latin American population in Spain.


Introduction
Previous academic research has provided evidence that changes in the economic processes, such as the business cycle, economic restructuring, and uneven development, have an impact on the internal migration flows in advanced capitalist countries (Fielding 1993). The study of the relationships between economic processes and internal migration reveals that, in periods of expanding economy, the volume of internal migration rises (higher migration rates) and, on the contrary, in times of economic contraction, the number of people who move declines (lower migration rates), as demonstrated in the US (Saks and Wozniak 2007;Molloy, Smith, and Wozniak 2011), Japan (Fujita et al. 2004), Nordic countries (Graversen et al. 2001), France (Courgeau 1985), and Spain (Pujadas and García Coll 1995;Recaño and Cabré 2003), among other countries. Furthermore, changes in national economic conditions 'affect different regions differently, thus triggering migration' (Greenwood et al. 1991, 239) and also changing the direction (origins and destinations) of domestic flows, creating dramatic reversals of migration trends in developed and advanced developing countries (Cochrane and Vining 1988) over the last 40 years.
However, these general statements refer to the total number of movers, and data should be further analysed to identify group differences in migration behaviour according to nationality or country of origin because advanced nations' internal flow composition is becoming more heterogeneous due to the increasing participation of newly arrived foreign populations. This is particularly so in many countries of the European Union and the post-industrial nations of eastern Asia. Thus, in these countries, the flows are composed not only of native, but also of an increasing number of foreign immigrants, who are generally more mobile than natives, in particular, the more recent arrivals (Bélanger 1993;Newbold 1996Newbold , 2007Rogers and Henning 1999;Zorlu and Latten 2009). Foreign immigrants show a different spatial behaviour, as recent research has also confirmed (Schündeln 2007; Catney and Finney 2012;Vidal and Windzio 2012;Lamonica and Zagaglia 2013;Calnan and Painter 2016). Immigrants tend to migrate less from the most economically active regions, as well as from those with a larger percentage of co-nationals (Kritz and Gurak 2000). However, the impact of the crisis on immigrants' migration intensities is viewed differently in the discipline. Thus, whereas Papademetriou and Terrazas (2009) and Calnan and Painter (2016) argue that foreign immigrants' mobility can increase during economic crises because they may be more willing to relocate for new jobs, other researchers such as Ellis, Wright, and Townley (2014) observed a decrease in the intensity of their internal migration in the US.
In order to contribute to that discussion, and using the Spanish case as an example, this paper focuses on the hypothesis that the economic crisis produces changes in the internal migration dynamics, particularly in the intensity and direction of migration, and that those changes will affect natives and foreign immigrants differently. We will show the recent Spanish migratory situation by quantifying the differences in internal mobility in Spain between the Spanish native population and the foreign immigrant population, concentrating on the foreign-born from Latin America and the Caribbean. This is the main foreign group in Spain, which has become the leading receiving country of the increasingly diversified Latin American inflow to Europe during the last two decades (Durand and Massey 2010;Domingo, Sabater, and Verdugo 2015). According to Eurostat's statistic data, this inflow amounts to more than 50% of Europe's 4.5 million Latin American residents. Their immigration to Spain was encouraged by laws, regulations, and programmes that responded to labour market needs during the expansive economic cycle in the country, and by Spaniards' ethnic preferences (Izquierdo Escribano, López de Lera, and Martínez Buján 2002;Vono 2010;Avila-Tàpies and Domínguez-Mujica 2015;Hierro 2016).

The specificity of the Spanish case
The Spanish migrations are paradigmatic for the testing of our hypothesis because of the strength of socio-economic factors affecting the recent spatial patterns of internal mobility, which could be enumerated as follows: first, a remarkable economic prosperity between 2000 and 2008, when the country experienced the highest rates of activity and the lowest levels of unemployment ever recorded, with a 30% increase in the number of employed persons, going from 15. 8 million in 2001 to 20.7 million in 2007. Second, a large influx of immigrants (including Latin American immigrants) -which increased the foreign-born population from 1.4 million in 2000 (3.6% of the total population) to 6.4 million in 2009 (13.8% of the total population) and made Spain the world's second most important receptor of immigrants in 2007, after the United States. Third, a domestic real estate bubblewhen the number of new housing units shifted from 300 thousand per year in the 1990s to 760 thousand in 2006, multiplying their prices (Burriel 2008) -, whose related industries were highly localised in the Spanish territory and employed many foreign immigrants. Fourth, a great recession started in mid-2008, related to the global economic crisis that affected Spain severely, with unemployment rates reaching 27% for the total of the country, and more than 40% for non-EU foreign residents. All these exceptional factors have changed the population dynamics of Spainwhich is a country characterised by 'relatively wide historical disparities in regional development' (Cochrane and Vining 1988, 237) and its patterns of internal mobility, as will be discussed later in this paper. Moreover, this global economic crisis of 2008 affected international migration dynamics, reducing the number of immigrants to Spain and increasing the number of returns as well as the exit of young Spaniards in search of work opportunities abroad (González-Ferrer 2013;Domingo, Sabater, and Ortega 2014), resulting in a negative migration balance for Spain.

Literature on foreign-born immigrants' internal flows in Spain
In the context of the Spanish immigration boom that began at the end of the 1990s, various studies were published analysing the intensity, direction, and characteristics of the internal migration of foreigners in Spain. One of those first studies is that by Recaño (2002), who identified the main characteristics of these movements, such as their high intensity, the diversity of profiles between nationalities, the complexity of migration routes, and the role of distribution and reception played by various provinces for certain nationalities (e.g. sending Latin Americans away from Madrid, or Moroccans' attraction to Catalonia). These characteristics were reproduced and gained importance over the years (Reher and Silvestre 2009;Recaño and Domingo 2006). As an example of these high migration intensity values, Reher and Silvestre (2011)making use of the National Immigrant Survey (Encuesta Nacional de Inmigrantes) of 2007have found that one in three foreigners had changed municipalities during their permanence in Spain. In some years (2007)(2008)(2009), foreigners' mobility even exceeded 30% of the total internal mobility in Spain, despite that their proportion of the Spanish population was only about 12%. The importance of the phenomenon is such that some scholars have identified it as a 'structural factor of the Spanish migratory system' (Recaño and De Miguel 2012;Recaño, Roig, and De Miguel 2015).
Other studies have detected a more important impact of the immigration boom in the internal mobility of the two largest metropolitan areas of the country -Madrid and Barcelonadue to the participation of foreigners in the process of urban deconcentration and suburbanisation (Pozo and García Palomares 2011;Sabater, Bayona-i-Carrasco, and Domingo 2012;Bayona-i-Carrasco, Gil-Alonso, and Pujadas 2013). In many cases, as noted by Quintero (2015), it consists of a spatial redistribution of those migrants who arrived in the metropolitan areas and who, after securing employment, moved towards the periphery as they demanded better housing. Those movements were related to foreigners' territorial dispersion and new re-concentration patterns. That would be the case of the Barcelona metropolitan area (Thiers, Bayona-i-Carrasco, and Pujadas 2015), where the emergence of areas of concentration in the peripheries of the core city are largely due to internal migration.
With regard to the decline in international inflows from the last quarter of 2008 due to the financial and economic crisis, new recent research has identified four major internal migration patterns (Gil-Alonso, Bayona-i-Carrasco, and Pujadas 2015) for all foreigners, using the criterion of nationality.
(1) A return of foreign nationals to the large metropolitan areaspreviously net-expellers of internal migrants to the rest of the countrybecause of the lower impact of the crisis in these areas due to an economic structure that was more resistant to the crisis (González-Romero, Caravaca, and López-Lara 2015; Rubiales 2016; Méndez, Abad, and Echaves 2015). (2) A strong out-migration of the foreigners from the areas most affected by the economic crisis, especially those where the real estate boom had a greater impact in the past, with Valencia as the best example. (3) Increased in-migration to the provinces and cities with low levels of immigrants, such as the case of some Andalusian provinces and the north of Spain. (4) In-migration of some foreign national groups to areas with certain ethnic niches in the labour market, such as intensive farming, in particular, to regions of Extremadura and Andalusia.
Furthermore, and referring to the particular case of Latin Americans, Quintero (2016) also found an increase in the distance of their internal moves as a result of the new pattern of migration outflows produced by the strong impact of the economic crisis in the residential markets of some particular regions, such as the flows from the east of Spain to the Basque Country.

Data and methods
We have used the micro-data from the Residential Variation Statistics (RVS) compiled by the INE (National Statistics Institute), basically from the exploitation of the information of registrations and cancellations due to changes of residence recorded in the Continuous Municipal Register (Padrón Continuo de Población), which annually registers the population residing in each Spanish municipality. According to the Law of the Local Regime Bases (Ley de Bases de Régimen Local, 1996), all persons residing in Spain have the right and the obligation to register in the Municipal Register in which they habitually reside. Foreigners, including illegal residents, are usually recorded in the Municipal Register as well, as this allows them to have access to the public health and education system and, especially, to prove their presence in Spain. It is considered that its figures provide a good approximation of the real number of foreign residents, particularly after several correction processes of the registers have been performed. 1 The RVS refers then to the annual domestic and international migration, although the availability of variables is limited: gender, age, nationality, and country of birth, and the place of origin and destination of the flows. This source provides a good quality flow data among Spanish municipalities (Recaño 2002;Martí and Ródenas 2006)on the contrary, more doubts are generated by the figures of the population exchange with foreign countries. The RVS collects data on moves and thus, it includes possible re-emigrations in the same year, although its incidence does not seem to be important (Martí and Ródenas 2012). Its main problem is the existence of 'false migrations' in order to achieve some benefits only conferred to residents of a particular municipality. There is also the problem of time lag between moving and reregistering. The Migration Statistics (Estadística de Migraciones) recently published by the INE address these problems, although their time series starting at 2008 does not allow us to cover the entire study period. We analysed the last 10 years, from 2004 to 2013, distinguishing two periods: a pre-crisis period (2004)(2005)(2006)(2007)(2008) concerning the growth of international immigration flowsparticularly of the Latin American inflowsand a continued increase in internal mobility; and a crisis period (2009)(2010)(2011)(2012)(2013) referring to the beginning of the decrease in the influx of migrantsincluding the Latin Americansand to the pattern of falling internal mobility rates.
Unlike other approaches (Gil-Alonso, Bayona-i-Carrasco, and Vono 2012), the criterion adopted in both cases for selecting the population is the country of birth, not the nationality. 2 This is due to the growing number of nationalisations that have occurred in Spain in recent years, which has particularly affected the Latin Americans 3 because of the preferential treatment offered by the Spanish legislationthey only need 2 years of legal and continuous residence in Spain to acquire Spanish nationality. 4 For operational reasons, the term 'Latin American' refers here to all individuals who were born in the Americas except for those born in Canada and the United States.
Finally, a qualitative typology of Spanish provinces has been developed to analyse internal mobility. The aim of this analysis is to identify the patterns of internal migration in relation to the socio-economic context. 5 The classification has been built using as a basis the following four criteria and assumptions: (1) The proportion of foreign-born in 2009, because of the assumption that, in times of crisis, the internal flows would move from the highest immigration areas to those with lower immigration, redistributing the population within Spain. The Mediterranean Coast, Madrid province 6 and its immediate surrounding provinces, and the Canary Islands account for most immigrants (Maza, Villaverde, and Hierro 2013). (2) The unemployment rates in 2009 (first trimester), because of the assumption that the areas of highest unemployment are less attractive than those with lower rates. A north-south division is clear, with a less affected north (a minimum of 8.3% in Orense) and a more affected southwith the maximum rates observed in Europe, over 25% in Cádiz, Málaga, and Las Palmas. (3) The employment rates in the agricultural sector (4.4% of total employment) in 2009, because of the assumption that the weight of the agricultural sector in economic activitydue to the existence of certain ethnic nichescan trigger some migration flows. The year 2009 was selected for the analysis because this is when the crisis started showing its effects. A high concentration is registered in the west of Spain, and a low concentration in the north (with the exception of Galicia), the centre (Madrid), and most of the Mediterranean coast.
(4) The size of the provincial population is also considered because large cities like Barcelona and Madrid respond better to the crisis due to their economic activities, and because they acted as the main gateways of Latin Americans into the Spanish territory, having stronger and longer established immigrant social networks.
As a result, a typology was created, where the 50 Spanish provinces were classified into five major groups or types (Metropolitan, North, South, East, and Northwest). 7 The clustering serves to abstract randomness in provincial behaviour. The resulting types of provinces can be generally characterised as follows: the first type (Metropolitan) is composed of Madrid and Barcelona, which are the two most populous provinces of the country, encompassing nearly one million Latin Americans, who constitute 42.5% of Spain's Latin American residents in January 2015. They are the most urbanised provinces in the country and have better resisted the negative impact of the crisis due to their large and diversified service sector. Both provinces show an important demand for Latin American female labour, focused on the domestic service sector and on the hospitality industry. The second type (North) involves 11 provinces, all of them located in northern Spain. They are characterised by a lower level of unemployment than the national average rate in 2009, and by a slight increase in foreign immigration also during the crisis period. The North hosts the 13.1% of the Latin Americans population in Spain, with just over 300 thousand residents. The third type (South), which groups 15 provinces, presents the opposite characteristics, such as very high levels of unemployment, low levels of foreign immigration, and relevant economic weight of agriculturewhich served as a gateway to Spain for some nationalities. The South has also a population of 330 thousand Latin Americans, which represents 14.3% of all Latin Americans living in Spain. The fourth type (East) encompasses seven provinces, including three insular provincesdeveloped particularly as residential tourism areaswhich showed a high economic growth at the start of this century, and a strong economic downturn during the crisis in the construction industry and associated industries. The economic weight of agriculture is very low and the foreigners' proportion of the whole population is high. There are 513 thousand Latin Americans settled in the East, which account for 22.1% of the total Latin American population in the country. The fifth type (Northwest) involves 15 provinces where the impact of the crisis has been more discrete due to their lower economic dynamism. Their foreign population is small, with only 188 thousand Latin Americans, representing the 8.1% of the Latin American residents in Spain.
Internal migration was analysed by calculating in-migration, out-migration, and net migration rates. Results are presented using net migration, that is, the difference between in-migrants and out-migrants in each of the five created groups of provinces in a year. The reference population is based on the Continuous Municipal Register statistics (mid-year population). As the flows of the RVS are derived from the municipal registers, no problem of consistency arises. The rate is calculated as follows, where n = period of time: Internal Net migration rate = (Immigration − Outmigration)/n P final − P initial 2 * 1000.
In Table 1 the Gross Migraproduction Rate (GMR) has been also calculated, which measures more accurately internal migration because it considers the effect of age structure. It is calculated as follows: where m t, t+n x, x+n are age-specific internal migration rates for the t period, a is the length of the age group, and w is the last age group analysed.

The distribution of Latin American immigrants in Spain
An issue to consider in our analysis is the territorial distribution patterns of Latin Americans in Spain. By studying the distribution in the last decade (2004-2014), we found that Latin Americans show a propensity to settle in the most urban and dynamic areas of the country. A high concentration of Latin Americans can be seen in Madrid and Barcelona, as well as in their respective metropolitan areas, and to a lesser extent in Valencia and Zaragoza, with few variations over the last decade. As in the case of natives, Latin Americans show a significant presence throughout the coast because of the existence of Latin American immigrant niches in the labour market. Some nationalities, such as the Ecuadorians, tend to concentrate in Murcia province because of its intensive agriculture, whereas others are more present in tourist areas such as the Spanish archipelagos. However, some changes have occurred over the last few years, such as a greater concentration in cities of the north of Spain and in some large cities of the south (Seville). Furthermore, the presence of conationals in Galicia and the Canary Islands is explained by old historical ties with certain Latin American countries, such as Venezuelans in the Canary Islands (Domínguez-Mujica, Díaz Hernández, and Parreño Castellano 2000) or Colombians in Galicia (Lamela, López de Lera, and Oso 2005).
The Latin American group is characterised by its feminisation, especially in early stages of incorporation into Spain (Pérez-Caramés 2004;Domingo 2006). That is, when the flows emerge, the proportion of women is higher; later on, they try to regroup their families in Spain, balancing the sex ratio (in recent years, large numbers have obtained the Spanish nationality and been allowed to apply for family reunification of their spouses and descendants under 18 years). The Latin American group is also characterised by high economic activity in the service sector (Gil-Alonso and Domingo 2008;Cacopardo, Maguid, and Martínez 2007), which locates them more preferably in the large urban areas in relation to other foreign groups. This is due to the large proportion of women in the flows 8 and their economic activities concentrated in the tertiary sector (Domingo and Martínez 2006), particularly in the domestic service sector: child/elderly care and housekeeping services, and, to a lesser degree, in the hospitality industry. The rest are mostly engaged in construction and agriculture. Despite that these are jobs with low educational requirements, an important social characteristic of the Latin American immigrants is their language proficiency (mostly, Spanish native speakers) and their relatively high level of schooling.

The focus on the territorial analysis
In order to elucidate the Latin Americans' internal movements, firstly, the intensity and spatial distribution of flows are analysed using absolute numbers and net migration rates. Secondly, the typology of provinces is used to analyse foreigners' internal migration flows according to migrants' region of birth at continental scales, and at national scales in the case of Latin Americans. For the latter, we also present the results for the top twelve countries of birth. Thirdly, we generate an Origin-Destination matrix, which shows the variation in the flows between the pre-crisis period and the crisis period. Lastly, we compare these findings with the trends of Spanish native population to contextualise our results.

Latin American internal migration in Spain
The volume of internal moves of Latin Americans is very large, as it exceeded two million in a decade, divided almost equally in the two study periods: 1,049,282 moves between 2004 and 2008; and 1,039,506 moves between 2009 and 2013. There was a peak in 2007 with 238,156 moves ( Figure 1). Eight foreign-born groups account for 82.5% of the moves in the decade: the Colombians (more than 400 thousand), the Ecuadorians (381 thousand), the Argentines (215 thousand), and the Bolivians (202 thousand) were the ones who had more registered movements. These numbers are related to a stock population of 2,459,089 Latin Americans in 2010, which has experienced an intense and continuous growth in absolute numbers until 2009, and a mild slow-down and decline later on, coinciding with the crisis, reaching 2,358,131 at present.
The lower intensity of internal migration more faithfully reflects the impact of the economic crisis (Figure 1, right). The rate of mobility reached a peak in 2007 with 110‰, began to decrease in 2008 until the present 77‰, a rate nearly 30% lower than the maximum established rate. In internal mobility, two types of movements are distinguishable: those that occur between provinces (inter-provincial) and those produced within the same province (intra-provincial). The former are usually associated with employment, while the latter are more related to housing and life-course factors. Both migration rates had followed similar trends over the decade analysed. The rate of inter-provincial mobility reached a peak in 2007 with 48.1‰, and its evolution had been characterised by a pattern of falling mobility (28.6‰ in 2013, a decrease of 40.5%). Likewise, the rate of intra-provincial mobility reached a peak in the previous year, in 2006, with 64.2‰, later decreasing to 48‰ in 2013 (a decline of 25%). Thus, the decrease is more important for inter-provincial moves. If we compare these values with those of the natives, we find that they are very high for both periods.
By country of birth, the rates follow a similar pattern (Table 1), with a general decline when comparing both periods. Paraguayans, Bolivians, Dominicans, and Colombians show higher rates, 9 which generally decrease between the two periods. Only the Cubans and Dominicans show similar rates in both periods. This is due to the characteristics of their migratory flows. The Dominicans were among the earliest migrations to Spain. They settled in large cities, working primarily in the domestic service sector. Their demographic profile is closer to that of the Filipinos than to the rest of Latin Americans. Cubans and Venezuelans are also particular cases, due to the restrictions to emigration for Cubans and the political situation for Venezuelans, which explain the lesser importance of economic migrants in both groups. The use of the GMR which considers the effect of age structure, confirms these trends.

Territorial patterns of internal migration in Spain
The territorial structure of Latin Americans' flows in Spain is roughly similar in the two periods, reflecting the distribution of Latin Americans in the country. Nevertheless, movements of different kinds are found: Firstly, the movements of suburbanisation, that is, of urban sprawl between a metropolitan centre and its neighbouring provinces, such as the case of Madrid with Guadalajara and Toledo, 10 which are the most important ones detected. Secondly, the redistributive effect of Madrid, which used to show a return movement. Except for the rest of Castile and Extremadura, this diffusion occurs throughout Spain, with special preference for the east coast. Thirdly, Barcelona's ambiguous function: it attracts migrants from Madrid, but at the same time, it sends them to its closest provinces. Finally, the exchange of migratory flows between the two Spanish archipelagos and the peninsula is important, occurring in the three most populated provinces: Madrid, Barcelona, and Valencia. In the case of the Canary Islands, the existence of old migratory ties with Latin America explains the situation, while the attractiveness of the Balearic Islands can be explained by their tourismdriven economic dynamism.
Thus, the territorial structure of the flows has not changed much prior to and after 2008. However, the intensity of the flows has changed considerably, and the trend has turned from one of net out-migration to one of net in-migration in many provinces. If we look at the provincial migration balances for the two periods (Figure 2), we can find two divergent situations: First period [2004][2005][2006][2007][2008] Madrid has the highest negative migration balance. A negative migration balance is also visible in the western half of Spain, especially Galicia (Pontevedra, La Coruña, and Orense, among the five areas with the most negative balances), and around the east coast (Albacete and Almeria). Positive balances are found around Madrid (Toledo and, to a lesser extent, Guadalajara); the Catalan coast (Barcelona and Tarragona), the Balearic Islands and, particularly, Valencia and Alicante. Furthermore, the Basque Country and Seville also have small positive balances. Second period [2009][2010][2011][2012][2013] In the following five-year period, many provincial migration balances change signs. Now, Madrid is the largest attractor of internal migration. Barcelona remains similar to the previous period, and the pull power of the Basque Country provinces increase, particularly Biscay. The changed sign of the Galician migration balance (from negative to positive) is remarkable. But perhaps the greatest change occurs on the east coast (from Tarragona to Murcia), which becomes a net expeller of migrants. The same changes occurred in the Canary Islands, which become a net expeller of migrants.

A typology of internal migration movements. An overview
Using the typology mentioned in Section 3, we represented in Table 2 the absolute net migration and the net migration rates for each type, by country of birth. Results show that just before the crisis, the Metropolitan type areas were characterised by net outmigration for all origins; the north and the east of Spain were clear attraction areas, particularly the latter; the south of Spain expelled migrants of northern African origin, particularly Moroccans, who, after having worked in the Andalusian agriculture sector, migrated to other areas with more labour opportunities (Cohen and Berriane 2011); the northwest, however, lost Latin Americans and other foreign-born people and also presented a negative or almost negative balance. This can be explained by the function of Table 2. Net migration balances in absolute numbers and migration rates, by periods, typology, and immigrant's place of birth. the Galician provinces, as Galicia is the first port of entry to Spain due to historical linkages with Latin America (Lamela, López de Lera, and Oso 2005) as a result of the large number of Galicians who settled in Latin America from the early 1900s. The crisis modified the scenario in the second period, to the extent that more than half of the migration balances changed sign. First, we can observe a decrease of the net migration rates with respect to the first period. Second, we see areas that showed drastic changes, such as the Metropolitan type areas or the east, and others that present almost no changes. In the Metropolitan type areas, three of the four foreign groups (all except the Africans) changed sign from negative to positive net migration rates. The greatest change was among Latin Americans, who showed by far the largest increase. In the east, the dynamic was the opposite of the previous period: all four groups lost population. In the south, the dynamic partly changed, and the Africans' migration balance became positive, meaning that the agricultural sector recovered importance in times of crisis.

A typology of internal migration movements, by Latin American origins
The analysis of the top twelve countries of origin in Latin America and the Caribbean confirms the patterns described so far for the whole of the Latin American group (Table 3). Only Dominicans, Venezuelans, and Cubans do not always follow the same pattern, due to the peculiar demographic characteristics of these groups, as mentioned above. Meanwhile, among the rest, the Bolivians are those who show higher migration intensities, as generally expected, because they are more recent arrivals. On the whole, Latin Americans' migration behaviour is more similar in the second period, that is, during the crisis times, than in the first period, in terms of migration balances and intensities.
Latin Americans, however, left the south in greater numbers as a result of high unemployment ratesamong the highest in Europe, along with the south of Italy or a large part of Greece. In the rest of the areas, there were few changes. In the northwesta historical gateway for some foreign immigrant groups-, the loss of Latin Americans was minimised. Small changes are observed in the north, which continued its slow growth as a pull centre.
In some instances, the spatial behaviour is almost identical for all countries of origin, as is the case of the net migration change for the Metropolitan type areas (from negative to positive) or for the east during the present economic downturn (from positive to negative). It can be said that the vast majority of the Latin American-born population follows the general patterns shown in Figure 2. Thus, the differences found among types of regions are not due to a different nationality compositionor, more properly, to a different population composition by country of origin in Latin America or the Caribbean.
Finally, graphs in Figure 3 represent the rate of change of the flows between the first and the second period. The horizontal axis represents the destinations, and, as we can see, the Metropolitan type areas increased inflows from all regions, without exception, particularly from the south. This means that the large Spanish metropolises work as 'refuge areas' for Latin Americans from any region. Furthermore, flows between the Barcelona and Madrid provinces also increased, as well as inter-municipal moves inside these metropolitan provinces.

Contrasting patterns of migration: Spaniards vs. Latin Americans
Two major factors changed the internal migration patterns in Spain from the mid-1970s: on the one hand, the advent of democracy and with this, the investments of the governments of the 17 Autonomous Communities in the territory; on the other hand, the crisis of the domestic economy, which particularly affected the industrial sectorthe most important attractor for labour force and highly localised in major metropolitan areas.
The new political and economic situation produced a change in the spatial pattern of the Spanish inter-provincial migration, altering the importance and directionality of the till then dominant migration flows (Avila Tàpies 1993). Since then, and before the arrival of the current crisis, internal migration gradually increased, with patterns characterised by a higher intra-provincial mobility and lower inter-provincial mobility than in the past. The latter only recovers in the 1990s, as a result of the greater participation of foreign immigrants in the migratory system. At this point, the major poles of attraction and expulsion disappear, and most of the municipalities become involved in internal migrations. Hence, a higher intensity is found as a result of an increase of intra-metropolitan migrations in the case of Spaniards (Stillwell and García Coll 2000) and interregional migrations in the case of foreigners (Franch 2009). Demographically, the largest metropolitan areas and their nearest provinces are those that show the highest growth, along with the Mediterranean coast, while the north of the country loses population. Afterwards, the intensities descended, and areas of attraction and expulsion changed ( Table 2). The Metropolitan type areas shifted from senders to attractors, both for Spaniards and foreigners, while the eastthe most affected by the economic crisis due to the higher weight of the construction sector, which is also a strong ethnic niche experienced the greatest changes. In general, net migration intensities descended and directionality changed, with higher intensities among foreigners (see Table 2). In the 10 years analysed (2004-2013), internal migration intensities increased between 2004 and 2007, reaching a maximum of 1.795 million moves in year 2007. Thereafter, the intensities gradually dropped to 1.551 million in 2013. During this period, the evolution of the movements of Spaniards hardly changed over the years, being always around 1.1 million movements per year (with a maximum of 1.235 in 2006). Meanwhile, foreign immigrants' moves reached a peak in 2007 of around 571 thousand movements, and fell thereafter. In the year 2013, the figure was 431 thousand. Currently, foreign immigrants account for 27.8% of all internal moves, a decrease from the maximum of 32.3% in 2008.
Finally, and regarding the demographic profile of internal mobility, foreign immigrants show higher levels of mobilitythree times higher than the nativeswith high mobility in all productive agesmaximum rates near 140‰ for [25-29] age groups and rates higher than 60‰ for all ages under 65. In contrast, the Spaniards present their highest rates of 60‰ in the [30-35] age group and related to emancipation and family formation (Gil-Alonso, Bayona-i-Carrasco, and Pujadas 2015).

Conclusions
The internal migration patterns of Latin American immigrants in Spain reflect the abrupt change in the economic situation of the country since 2008: on the one hand, in a decrease in intensity, estimated around 30%, and, on the other hand, in the change of the directionality of the flows.
However, Latin Americans are not the ones most transformed by the crisis; Africans' internal migration patterns were more affected, as they were engaged in construction (strongly impacted by the crisis) and agricultural activities (especially in the early stages of insertion in Spain). In any case, the changed sign of Latin Americans' migratory balances in most areas is remarkable, as it represents a break in the dynamics and flow trends of the previous period. From a regional perspective, there have been clear changes in the provinces of destination. Those provinces highly affected by the crisis show a decline in their attraction power, and are now sending migrants to other provinces. Spain's east and the archipelagos, but also Spain's entire south, are the most affected by these dynamics.
In all types of provinces, net migration rates are now lower than prior to the crisis. This means greater uncertainty in the current territorial trends. Reversely, large cites had acted as refuge centres from the crisis. In the first period, they were areas of arrival and subsequent redistribution, especially Madrid. Now, they show a decrease of outflows (because there are fewer 'newcomers' from Latin America) and an increase of inflows from other areas of the country. Therefore, the economic crisis in Spain caused a change in the previous dynamic of geographical dispersion. Although this dispersion trend remains in a part of Spainin the north, less affected by the crisis-, it disappeared in the south due to the lack of job opportunities. Thus, the crisis has become a new factor of differentiation in the distribution of the Latin American population in Spain. Commonalities among all the Latin Americanborn in the use of the Spanish space (traditional gateways, suburbanisations, 'refuges', etc.) and recent spatial responses to the crisis (returning to the metropolises) reflect their historical linkages with Spain and the effects of a shared language, norms, and culture on networking, employment, and mobility.