Population Research and Policy Review

, Volume 31, Issue 4, pp 465–496

Going Back Home? Changing Demography and Geography of Mexican Return Migration

Authors

    • Department of SociologyMcGill University
    • Department of SociologyMcGill University
  • Bryan R. Roberts
    • Department of SociologyThe University of Texas at Austin
Article

DOI: 10.1007/s11113-012-9243-8

Cite this article as:
Masferrer, C. & Roberts, B.R. Popul Res Policy Rev (2012) 31: 465. doi:10.1007/s11113-012-9243-8

Abstract

Return migration has been a constant feature of Mexico–US migration patterns, but its characteristics have changed sharply with time. We use the Mexican censuses and counts of 1995, 2000, 2010, and the complete set of individual and household records of the 2005 Population Count to explore the demographic characteristics of returnees in the context of tighter border control and rising levels of forced return migration. Involuntary and therefore unplanned return is likely to mean greater difficulties of incorporation into the community of origin. The study of the effects of the militarization of the US–Mexico border on migration patterns has focused on the US side. We contribute to this literature by focusing on the Mexican side. We consider the intensity and type of previous migration to the US as compared to current return migration, and State of origin and destination. Our data suggest that particularly attractive destinations for returnees are border cities, prosperous communities and growing metropolitan areas. Findings suggest changes in the demographic composition and geographic distribution of returnees. The discordance between the patterns of outmigration and return is a telling indicator of changes in the overall migration relationship between Mexico and the US. The patterns for 2005 are also observed in 2010 even if the absolute number of inter-censal returnees increased threefold over the period. Finally, we argue that focusing on destinations of return instead of areas of emigration will contribute to a more comprehensive understanding on the nature of future return migration to Mexico and its policy implications.

Keywords

Return migration Deportations and immigration enforcement Mexico United States Emigration

Introduction

Research on Mexico–US migration has neglected return migration. The Mexican perspective has concentrated on emigration and the United States on immigration. Two factors help explain this neglect. The Mexican census counts show, as we will see, a decline over time in return migration relative to the outmigration of Mexicans to the US. Also, the geography of return migration, the places to which people return, has until recently been uncomplicated, since the census counts show that most return migrants have gone back to their places of origin, which, in the majority of cases, are small communities in the sending states of the Center-West of Mexico. In the first decade of the new millennium, various factors have combined to make return migration a less predictable process than in the past, raising important policy issues concerning the reintegration of return migrants in Mexico. These factors are changes in the geographical pattern of Mexican migration to the United States (Riosmena and Massey 2012), a tightening of border controls (Cornelius 2005; Massey 2005), insecurity on the Mexican side of the northern border (París Pombo 2010) and the US economic crisis at the end of the decade (Rendall et al. 2011).

Regions and places, such as the larger cities, that in the past had little international migration experience have now become major sources of migrants to the US. Also, Mexican migrants find work in areas in the East, North-West and Center of the US that had little previous Mexican immigration. The changing geography of economic opportunities in the United States and the impact of uneven economic development in Mexico create new destinations and new origins (Riosmena and Massey 2012). The changes in the geography of emigration are likely to change the geography of return migration towards cities rather than rural areas and away from the traditional sending states of the Center-West of Mexico. Migrants from the new places of origin in Mexico have had less time to establish households and migrant communities in the US than migrants from traditional areas of out-migration. Consequently, their propensity to return may be higher than migrants from traditional areas when faced by economic difficulties both because of lack of community support and lack of family commitments in the US. Consequently, we hypothesize that the new geography of emigration will result in a different geography of return despite the fact that traditional sending places have much larger stocks of migrants who could potentially return than do the newer sending places.

New geographies of return migration are likely to pose difficulties for the reintegration of migrants into Mexican society, since they include involuntary return through deportation and also the return of people to places that may not easily be able to support them. Thus, the new places of origin include some of the poorest rural areas of Mexico, as well as cities where supportive social relationships may not be as readily available as in rural communities. Even in traditional sending areas, changes in the rural economy brought by commercialization and competition from imported foodstuffs can make reintegration more difficult than it was in the past, potentially resulting in a greater number of returns to places other than those of origin, as returnees seek available employment opportunities.

We argue that changes in the geography of return migration are likely to be more consistent and more revealing of fundamental changes in the Mexico–US migration relationship than are changes in the volume of migration. Trends in the volume of migration are complicated by the contradictory effects of the tightening of border control and increased immigration enforcement. The tightening of control, combined with the insecurity of the northern Mexican border can deter undocumented Mexican emigration, but also deter the return migration of undocumented migrants who had habitually moved back and forth between Mexico and the US (Cornelius 2008; Fuentes et al. 2008; Massey 2005). Likewise, even the US economic crisis of the end of the decade might not necessarily lead to a major increase in return migration if migrants can rely on support in the US, have family commitments there and prefer to wait out the crisis rather than risk not being able to return when economic opportunities improved. Offsetting these retaining forces is the substantial increase in deportations in the years up to 2010.

A qualitative study conducted in 2010 in Tlacuitapa, a rural town in Los Altos de Jalisco region with a long history of migration to the United States, illustrates the impact of these factors in reducing return migration in face of economic crisis in the US. Migrants from this traditional sending community have support from social networks to go through adverse economic situations, many of the migrants consider their home to be in the US because most of their family members are there, a large proportion is documented, and a majority has been there more than 15 years in the US (Fitzgerald et al. 2011). The authors do not find that the anti-immigrant policies have had an impact increasing return migration of Tlacuitapeños living in the US (García et al. 2011). They do not find either massive return to Los Altos due to the economic context because individuals in both sides of the border have found ways to go through this and other crises (Blanc et al. 2011; Cabrera-Hernández et al. 2011).

Tlacuitapa represents a traditional sending place with consolidated migrant communities in the US. We hypothesize that it and communities like it will have a lower rate of return migration than do newer sending places. Is Tlacuitapa a representative case of contemporary return at the national level? We will compare the rates and distribution of return migration between traditional and newer sending states in the first decade of the new millennium. We will be particularly interested in whether the changes in the redistribution show a consistent trend despite the fluctuations in the volume of return migration, which the census reports as declining up to 2005 and then increasing sharply by 2010. We use the complete set of individual records of the 2005 Population Count to study the types of places to which migrants return to examine the demography and geography of return. As a measure of change in the geography of return migration we define a rate of return linking the population resident in Mexico who had been living in the US five years before, with the population who had left during the last 5 years to the US and were still away.1 This indicator will be calculated for 2000 and 2005. A similar benchmark cannot be estimated for the Mexican Census of 2010 since the 2005 Count does not report the numbers of Mexicans who left after 2000 and were still in the US in 2005. We will, however, include data from the 2010 Census that provide comparisons from 2010 with the geographical patterns identified in 2005.

The structure of this work is the following. The next section considers the historical and contemporary nature of return migration in Mexico. The section following presents the data, measures and methods used. Then we compare briefly the characteristics of contemporary returnees with previous returnees; we consider age, gender and educational levels. The following sections consider the relation between previous geographical patterns of migration to the United States and the current patterns of return migration. We conclude with a discussion of the implications of the changing patterns of return migration.

The Nature of Return Migration

Historical Characteristics of Return

Historically, return migration has responded to a range of motivations and circumstances: the repeated circular movement of seasonal migrants who complemented subsistence farming with temporary agricultural work in the US; those returning to stay after meeting their targets for savings or learning new skills after a couple of years of work in the US or in old age; those forced to return by deportation or lack of work who had wanted to stay longer or permanently in the US (Massey and Espinosa 1997; Riosmena 2004). For most of the twentieth century, however, return migration from the US to Mexico was heavily circular, moving repeatedly for seasonal work in the United States and returning to villages and small towns where the rest of their family had remained (Durand et al. 2001). It also was a return to the communities where migrants originated, predominantly rural in the states of the Center-West of Mexico, known as the traditional sending states.

The migration flow that originated in the beginning of the 20th century in rural communities of Western Mexico constitutes now a well-established flow such that we can identify states as traditional sending regions. The states that are considered under this category are Aguascalientes, Colima, Durango, Guanajuato, Jalisco, Michoacán, Nayarit, San Luis Potosí and Zacatecas (Tuirán et al. 2002). It is these states that have been the main focus of the Mexican Migration Project (MMP). Although the project gathers data on those who have returned to Mexico, the main research perspective has been emigration, with the exception of Lindstrom (1996) and Reyes (1997). The studies emanating from the MMP have shown that long established migration flows generate a cumulative causation in which earlier migrants facilitate the migration of subsequent ones, making it easier for women or those with less skills or material resources to migrate (Massey 1990). Long-established flows are also likely to mean that a larger proportion of the community’s migrants have documents to enter the US, gained through the provisions of IRCA and family reunification. The presence of well-established migrant communities in the US may encourage migrants to stay, but strong transnational networks and legal documentation may support circular migration.

Various factors changed the dominance of this pattern of seasonal migration from small rural communities in the Center-West of Mexico in the late twentieth century, including the changing basis of the Mexican economy as a result of free trade policies and a continuing urbanization that undermined the viability of subsistence oriented farming and made non-farm employment an increasing feature of rural economies. Accompanying these trends are shifts in the pattern of internal migration in which the importance of rural–urban flows decreases, whereas inter-urban migration increases, particularly to intermediate-size cities (Sobrino 2011). The result has been a changing geography of economic opportunities in Mexico, with the emergence of urban economies linked to tourism and to export industries, such as maquiladoras. The origins of Mexican migrants to the US diversified away from the traditional sending areas, such as the Center-West of Mexico, to new areas, such as Veracruz, and became more urban and metropolitan (Escobar 2008b; Hernández-León 2008; Tuirán et al. 2002). Rural, urban and border regions play different roles in emigration (Fussell 2004a; Fussell and Massey 2004). However, the rural–urban divide has become less clear-cut. Using the 2000 Mexican Census, Hamilton and Villarreal (2011) challenge the traditional rural–urban dichotomy by including the proximity of rural areas to metropolitan areas and showing its importance in emigration from small communities. A recent study analyzing the links between the diversification of origins in Mexico and destinations in the US support the hypothesis that economic restructuring in both sides of the border have had large consequences in the changes in the geography of emigration (Riosmena and Massey 2012). We hypothesize that all these changes in the geography of emigration and internal migration will increasingly shift return migration away from traditional sending areas to the newer sending areas.

Although internal and international movements have traditionally been conceived and studied separately, research suggests both their theoretical and empirical interconnectedness (King and Skeldon 2010). The border region of Mexico and the United States plays a key position in this relationship. Internal migration from the 1990s onwards showed a marked trend northwards towards the US-border region, which drew migrants from all other regions (Roberts et al. 1999). Fussell (2004a, b) finds that Tijuana plays a two-fold role in the migration flow: a destination for internal migrants and a home base for migrants that make repeated trips to the United States. We expect the border to be a region for step-migration both before and after a trip to the United States; in other words, a place for step-return migration as well.

The return destinations of migrants to the US diversified to include the new sending areas and away from the places from which migration originated. Return to places other than those of birth was estimated as making up a quarter of return migration in Mexico in the 1990s, although international migrants originating in the traditional sending regions were the most likely to have returned to their places of origin (Lozano-Asencio et al. 1997). Data from the 1992 to 2002 Mexican National Survey of Demographic Dynamics (ENADID) show that a similar proportion of those returnees that had moved internally before migrating to the United States was found to have migrated internally upon return (Canales and Montiel 2007). Moreover, in a traditional sending state like Jalisco, an important share of those that leave would return to the state but resettle in cities other than their community of origin, especially if they created a new household in the US (Papail and Arroyo Alejandre 1993).

Contemporary Return

With return migration representing a small fraction of Mexican migrants in the US, return migrants are not likely to reflect the characteristics of those remaining in the United States, whether in terms of region of origin or in terms of age or gender. One reason is that involuntary return migration as a result of Department of Homeland Security (DHS) expulsion is now considerable (see Table 1).2 The number of deportees from 1995 to 2009 is close to the total number of returnees reported by the 2000 and 2010 Mexican Censuses and the 2005 Mexican Population Count.3 The Mexican Population Count4 has no information on returnees who had left after 2000 and returned by 2005, but total returnees by 2005 are likely to be either equivalent to or less than total deportees by 2005. The ratio of population residing in the US five years before (inter-censal return migration) captured in the Mexican data over the number of deportations captured by the DHS increase over the five year period suggests an increase of returnees that may have been deported.5 The close fit between deportations and returns suggest that return migration is becoming a forced decision, or at least unplanned or with no preparedness.6 An important caveat is that despite the criminal penalties imposed on deportees who are discovered re-entering, a significant, number may actually return again to the US because they are likely to have family and property in the US (Hagan et al. 2008). A recent journalist account supports the image that increasingly, people trying to cross the border are not first time migrants, but deportees that have left family in the United States and define their home in the northern country (Cave 2011). This is consistent with a decrease over the decade in unauthorized adults that have been less than 5 years and an increase in the population that has been more than 15 years in the US: in 2000, 32 % have been less than 5 years but this was reduced to 27 % in 2005 and to 15 % in 2010 (Taylor et al. 2011).
Table 1

Deportations by criminal status and return migration by type of return in Mexico, 2000–2010

Deportationsa

Return

Return migration/total deportations

Period

Criminal

Non-criminal

Total

Period

Intra-censal migrationb

Year of observation

Inter-censal migrationc

1995–1999

198,357

262,740

461,097

1995–2000

260,412

2000

267,338

0.6

2000–2004

311,054

444,710

755,764

2000–2005

NA

2005

238,331

0.3

2005–2009

395,308

698,962

1,094,270

2005–2010

307,783

2010

985,383

0.9

Sources: Data on deportations were obtained from Department of Homeland Security (2010) and estimates of intra- and inter-censal migration were obtained from the Mexican 2000 and 2010 Census and 2005 Mexican Count

aDHS classifies removals under a criminal or non-criminal charge. Removals based on an order differ from returns which are mostly Mexicans apprehended by the Border control. Period estimates were obtained adding data from fiscal years

bPopulation of 5 years and older that left Mexico during the given period but was living in Mexico by the end of the period at the moment of the Census (2000, 2010) or Count (2005)

cPopulation of 5 years and older that was living in the US 5 years ago but was living in Mexico at the moment of the Census (2000, 2010) or Count (2005)

The border region’s role in US–Mexico migration is likely to have been transformed by the increase in deportations. The border becomes a destination for those deported that try to enter the United States as soon as possible or that remain in border cities for years as an idle floating population waiting for the opportunity to migrate again (París Pombo 2010). While the number of deportations increased over the last decade, especially those with a criminal order (U.S. Department of Homeland Security 2011), the number of apprehensions of Mexicans by the Border Patrol has declined (U.S. Department of Homeland Security 2011). París Pombo (2010) shows how strong stigma caused by deportation and lack of social ties in Mexico deters those that have been deported to return to their communities of origin and how limited human capital deters their reincorporation into the labour market or into the educational system. Difficulties faced for reintegration upon forced return is not exclusive of the Mexican case (Black and Gent 2004; Brotherton and Barrios 2009; Ruben et al. 2009).

As smaller proportions of migrants return, and restriction increasingly deters or expels undocumented migrants, the prevalence of the different types of return migration is likely to alter. Circular migration is likely to be the most affected since it entails repeated entries and exits, whereas target migration is often a once only exercise. Likewise it is less likely that life stage migration by the young to gain experience or the elderly to retire will occur as the costs and risks of undocumented migration increase for the first group and established migrant communities offering care and a familiar culture emerge in the US for the second group. As the voluntary components of return migration become less consequential, involuntary migration becomes a larger component of return migration, with likely negative consequences for ease of adjustment.

Not everyone migrates and not everyone who migrates returns. The phenomenon we are studying is immersed in diverse selection processes, both for outmigration and for return (Borjas and Bratsberg 1996; Cohen and Haberfeld 2001; Lam 1986, 1994). These selection processes are likely to change with time depending on previous migration flows and capital related to migration (Lindstrom and López Ramírez 2010). The overall changes in return migration patterns and the geography of opportunities in Mexico are also likely to affect the individual characteristics of return migrants, such as gender, age, education, poverty and employment (Gitter et al. 2008; Lindstrom 1996; Massey et al. 2006; Reyes and Mameesh 2002). The experience gained due to migration, the acquisition of new traditions, values and lifestyles, have transformed the individual that once left Mexico.

According to a recent study using the 1996–2009 March Current Population Surveys, compared to other migrants in the United States, Mexican emigration is more affected by social and family variables (family ties in both countries) than by short-term unemployment spells in the US (Van Hook and Zhang 2011).7 This is likely to have different effects for Mexican returnees depending on where their family members are living and where they originally left from. We can expect that those from traditional areas are more likely to have family in the US and so would be less likely to return. However, after deportation, they are also more likely to re-migrate to US if family was left behind in the US. Contrary to this, someone deported from a new sending area is more likely to have family in Mexico, and would be more likely to stay in Mexico. Whereas family ties may pull migrants to their communities of origin, going back to these communities may not be an option for the same reasons that made the migrant leave in the first place. The stage in the life cycle at the time of return affects the creation of new households that may be nuclear (for example when a Mexican couple returns with children born in the United States) or formed by someone living alone. Return to a new destination different than the community of origin may use social networks based on extended family members living elsewhere, as well as on households of co-residence with no family relation. Family structures of returnees can be thought of as cause and consequence of the processes of decision to return and the selection processes we have discussed before. The analysis of the family structure of returnees and their position in households inform partially these processes. These effects are likely to shape differently the types of households and living arrangements where migrants return to. That is, we expect an interaction between the demography and the geography of return. Thus, we hypothesize that the demographic characteristics of returnees will vary by type of destination.

With a decline in Mexican emigration to the US and an increase in return, a recent report has found that net migration rate has fallen to zero or below (Passel et al. 2012). The volume of return migration has been affected by economic crises and recession.8 Before the publication of the results of the 2010 Mexican Census, there was an ongoing debate over whether return migration would have increased or not as a result of the economic crisis in the United States and of increased immigration enforcement policies including higher rates of deportation (Alarcón et al. 2008; Passel and Cohn 2009; Rendall et al. 2011). Despite continuing uncertainty about the extent and nature of return migration in the first decade of the new millennium, the 2010 Mexican census shows that the absolute number of individuals whose residence 5 years before was in the US increased threefold from 2005 to 2010.9

Because of their economic interdependence, the pattern of GDP growth in Mexico normally follows that of the United States. Thus, in the years 1995–2005, Mexico’s annual rate of growth of 2.7 % was similar to that of the United States, which had a 3.3 % growth rate. Individual years had wider disparities, but both countries showed slow growth in 2001 and 2002. After 2005 the relative performance of the US and Mexican economies changes with US growth slowing considerably between 2005 and 2010 to a 1.0 % annual average. Mexico, which was less affected by the global financial crisis, grew at a 5.5 % annual average. After 2005, the relative economic conditions in the two countries together with increased deportations would be forces acting to increase return migration. Comparing the two periods shows a dramatic change in the numbers of return migrants, defined as those Mexican residents who were living in the US five years previously. Return migrants increased from 244,426 to 994,474, as did the increase in deportations (Table 1). Despite this increase, return migration from the US remains the choice of a minority of migrants. From the Mexican 2010 Census, only some 30 % of migrants leaving Mexico for the United States in the five year period, 2005–2010, had returned by the end of that period. Return itself may be temporary and those that return may leave again. Data from 2010 show that 14.1 % of those that were living in the US in 2005 but returned by the moment of the 2010 census (what we will call inter-censal migrants) had made an additional move to the US within the five year period (what we will call intra-censal migrants).10

Since individual and community poverty influence migration to the US and remittances from the US (Escobar 2008b; Janssen and Escobar 2008; United Nations Development Programme 2007), poverty is also likely to be selective of return migrants and the communities to which they return. Data has shown that most of the migrants to the US are not coming from the poorest municipalities; the relationship between poverty and marginality and out migration is not linear and the degree of marginality and the index of migration intensity show an inverted U relationship (Zenteno 2008). Also, due to the costs of financing their trip, “the emigration rate among the poor is lower than the non-poor. […But…] during the past 20 years, emigration has risen particularly rapidly in states with high poverty and marginality rates, and especially in the South (Oaxaca, Veracruz, Puebla, Campeche)” (Escobar 2008a). High proportions of returnees in places with economic opportunities are likely to indicate not only a greater propensity of their inhabitants to return to these places, but also the addition of people who are choosing to return to other places than the places of origin if their origins have few or no economic opportunities.

Data, Measures and Methods

We concentrate on a descriptive analysis of return migration using Mexican data from 1995 to 2010 and focusing on the characteristics of returnees and their households, as well as the localities that attract these returnees.11 The 2005 Population Count is a key part of this analysis since it represents the universe of returnees and not a sample because we are working with the complete records of the Count.12 That is, we are working with the complete records of those that declared they were living in the United States in 2000 but were living in Mexico in 2005, allowing us to make a more accurate count of migrant characteristics and of their places of residence than is possible with a census sample. A main approach of this paper is to present a comparison of return migration for 2000 and 2005. The 2005 Population Count will be used to compare return migration with the estimates of the propensity of return by state and locality. The 1995 Population Count is used mainly to define a rate of return for 2000 and we use the 2010 Census to verify if the trends in the patterns observed in 2005 persist.

The only question regarding migration that is available in the 2005 Population Count is “In which state of Mexico or in which country were you living 5 years ago?”13 (state or country of residence 5 years ago for the population of 5 years and older currently living in Mexico).14 We will refer to this as inter-censal migration return migration. Migration is measured in the Censuses and Counts based on a five year period. Therefore, intra-censal migration is defined as the population that left for the United States during the five year period and returned to Mexico within the period.15 So, for example, 1995–2000 intra-censal migration is defined as the population that left in the period of 1995–2000 (after the 1995 Population Count) and returned to Mexico before the census of 2000. Intra-censal migration is captured in the extended questionnaire of the 1995 Population Count and 2000 and 2010 Population Censuses, it cannot be measured for the period 2000–2005 and is sometimes referred to as circular migration, for example in Tuirán et al. (2002) and other Mexican publications.16

By definition, from the questionnaire, intra-censal migration is of someone who was resident in the household previous to migration17 whereas this is not necessarily the case for inter-censal migration since we do not know where the individual was living before moving to the US. The role of the household of return is key for our analysis because it allows us to understand better the living conditions upon return. We acknowledge the underestimation of outmigration and circular migration in census data due to complete households that left18 and this constitutes an inherent limitation of the use of census data. However, there is no clear evidence to support geographic differences in the magnitude of the bias.19

We define and calculate a rate of return as those non-institutionalized individuals who were in Mexico in a given year whose residence 5 years ago was the US, over those who went to the US during the previous five-year period but had not returned by the end of the period.
$${\text{Rate}}\;{\text{of}}\;{\text{return}}\;(y,y + 5) = \frac{{{\text{Population\; that\; were\; living\; in\; the\; US}}\;{\text{in}}\;y\;\text{but\; were\; living \;in \;Mexico}\;{\text{in}}\;y + 5}}{{{\text{Population\; that\; left\; during\; the\; period}}\;y - 5\;{\text{to}}\;y\;{\text{to the US and were still there in}}\;y}} $$
For example, during the period of 1995–2000, 1,469,801 migrated to the US. Of these, 260,650 (17.7 %) people went back to Mexico before the Census in 2000 and 1,209,151 (82.3 %) were still living in the US in 2000. Thus, we have a national rate of return of 0.2 for 2005:
$$ {\text{Rate of return}}\;(2000,2005) = \frac{\text{238,331}}{\text{1,209,151}} = 0.2 $$

This is slightly smaller than the rate of return for 2000 which we calculated as 0.22 = (294,441)/(1,324,817) using the 2000 Population Census (for numerator) and the 1995 Population Count (for denominator). This is compatible with the declining trend in return observed over these years.

The rate of return will also be calculated at the state level. The concept of rate of return, as defined here, for example, relates the total population that was living in the US in 2000 and had returned to Mexico in 2005, and the population that was living in the US in 1995 and had not returned to Mexico in 2000. Note that the populations reported as living in the US in 1995 in the 2000 Census and those living in the US in 2000 in the Population Count of 2005 can include people who migrated to the US many years before either 1995 or 2000, but who have only recently returned to Mexico.20 Thus, places with a large and long-standing stock of migrants in the US have a potentially much bigger base for generating return migrants by 2000 or 2005 than do places of more recent migration experience. Therefore, the rate of return for traditional sending states is likely to be overestimated by being based only on the past 5 years and our calculations for these places are likely to be conservative. Conversely, states for which emigration is a relatively recent phenomenon should have a potentially smaller population who have migrated many years before 1995 and are at risk for return.

The rate of return as defined here attempts to capture the population at risk of returning by linking both processes of out and return migration. The rate of return migration cannot be calculated at the municipality or locality level, due to restrictions of data availability in the 2000 census. It differs in the denominator to other indicators of return like the number of returnees per 1,000 inhabitants, or the percentage of returnees present. We will compare characteristics of returnees in states that have a rate greater than one (i.e. that attract more people than they send) with those in states with rates below one to indirectly estimate the characteristics of those that may not be returning to the place from which they originally left. We will also distinguish states having rate below one in two groups: having rate below and above the national average (which in 2005 was 0.2). Remember that for inter-censal migrants, the household of return does not need to be the same household to the one previous to migration. This allows us to consider the destinations of return as destinations of “immigration” or settlement after migration. Thus, a strategy to detect the changes in the patterns of the geography of return is to compare current destinations for return to the origins of emigrants and analyze the characteristics of those localities that receive more returnees. In order to explore the interaction between the demography and geography of return we will compare the characteristics of returnees in states that have different degrees of rate of return.

How do Returnees in 2005 Differ to Previous Returnees?

In this section we will briefly summarize the characteristics of the inter-censal returnees (population21 residing in the US five years before) between 1995 and 2010. First, comparing the gender distribution of returnees over time, we see that although inter-censal return migration has changed in absolute numbers the gender distribution has remained fairly constant with men making up around two-thirds of the population (see Table 2). In terms of age we can see that the mean age of returnees was increasing up to 2005 and remained relatively stable in 2010. This can be partially explained by the changes in their position in the household: increase in the presence of returnees who are the head of the households and a decrease of returnees who are sons or daughters of the head. The major difference comparing the 2010 to the previous years is the fact of the dramatic increase in the total number of inter-censal migrants that went from the 200,000 to reach almost a million people.22
Table 2

Basic demographic characteristics of those in Mexico with residence 5 years ago in US, 1990–2010

Demographic characteristics

Inter-censal migration from the USa

1995

2000

2005

2010

Total

346,214

280,051

238,331

985,383

Gender

 Men

61.9

60.5

66.1

68.7

 Women

38.1

39.5

33.9

31.3

Mean age

28.9

28.2

31.4

30.9

Position in the household

 Head

38.8

36.1

44.6

43.3

 Spouse or partner

12.5

15.1

14.7

12.6

 Son or daughter

32.3

34.9

29.9

30.9

 Other

16.4

13.9

10.9

13.1

Education (aged 15 and older)

 Mean years of schooling

8.5

8.3

8.5

8.5

Sources: Estimates generated by the authors using the complete set of individual records of the 2000 Population Census and the 2005 Population Count at the individual level. Estimates for 1995 and 2010 were obtained by the authors using the 10 % of the 1995 Count and 2010 Census

aPopulation of 5 years and older that was living in the US five years before (the year referenced) and was living in Mexico at the moment of the Census or Count. To make data comparable to 2005, we do not distinguish by place of birth and include everyone no matter their nationality. Includes non-institutionalized individuals only

Next, we will focus briefly on the characteristics of returnees in 2005. First, if we contrast the comparable Mexican-born population in the US in 2000 (the population at risk of returning) and inter-censal migrant populations, inter-censal returnees are more likely to be male than is the Mexican-born US population, more likely to be over 50 and as likely to be children. The average years of education of both males and females is similar to that of their counterparts in US, as are male years of education compared to those of their non-migrant counterparts in Mexico, but female returnees have higher levels of education than their counterparts in Mexico. This is coherent with previous findings of differences by gender (Lowell et al. 2008; Rendall et al. 2010, 2011; Van Hook and Zhang 2011).

In terms of living arrangements, most of the 2005 returnees live in nuclear families and are actually the head of the household. Returnees’ household structure is quite similar to the national distribution, although returnees are slightly more likely to be in extended families. For 2005, on average, the characteristics in terms of the quality of the buildings and access to basic services of households with returnees are slightly better than the households without returnees no matter the size of the locality. A similar pattern is seen when we compare the goods of households with and without returnees where return households are more likely to possess some goods than are households without return migrants.

Destinations of Return

The main finding about geographical destinations of return is that returnees are returning disproportionately to states and places which, in the past, had low out-migration to the US. States with highest rates of return are not traditional sending states (see Table 3). There is variation within regions, but overall the general pattern is of the highest rates being in states of the northern region and lowest in the central states. The states that have a rate of return greater than one are Quintana Roo, Baja California and Baja California Sur and we will refer to these as having “rates above” and the other 29 states will be referred to as having “rates below”.23 The states having rate below one are divided in two groups: those having rate smaller than 0.2, the national average for 2005, and those having a rate between 0.2 and 1. The states that have a rate of return greater or equal to 0.2 but smaller than one are Aguascalientes, Campeche, Coahuila, Colima, Chiapas, Chihuahua, Distrito Federal, Durango, Jalisco, Nayarit, Nuevo León, Oaxaca, Querétaro, Sonora, Tabasco, Tamaulipas, Yucatán and Zacatecas. The states with rates smaller than the national average are Guanajuato, Guerrero, Hidalgo, Puebla, San Luis Potosí, Sinaloa, Tlaxcala and Veracruz. A rate above one shows evidence of attracting proportionately more returnees than they had emigrants. Note that Sonora, a border state with Arizona, has a rate of return of 0.61 which is higher than many other states. After the increase of border control in California, the Sonoran desert became one of the main crossing points into the US and this has had several implications for undocumented migration (Cornelius 2005). The high rate of return suggests a concentration of people with previous migration history in the US who are likely to include a large proportion intending to cross the border again.
Table 3

State rate of return (2005) and inter-censal return migration by Region (2000–2010)

State

Inter-censal return migrationa

Rate of return migrationb

State

Inter-censal return migrationa

Rate of return migrationb

2000

2005

2010

2000

2005

 

2000

2005

2010

2000

2005

Traditional migration sending state

South and Southeast

 Aguascalientes

4,167

4,922

14,700

0.20

0.29

 Campeche

373

623

2,865

0.94

0.35

 Colima

4,030

3,306

12,589

0.37

0.36

 Chiapas

701

1,636

12,609

0.12

0.21

 Durango

7,186

6,288

21,701

0.20

0.20

 Guerrero

10,238

5,276

42,517

0.14

0.09

 Guanajuato

20,736

14,829

73,800

0.10

0.12

 Oaxaca

7,031

9,633

39,679

0.18

0.20

 Jalisco

36,090

29,429

88,108

0.31

0.23

 Quintana Roo

546

2,064

4,698

0.93

1.31

 Michoacán

29,520

21,351

80,762

0.20

0.17

 Tabasco

313

599

4,247

0.23

0.24

 Nayarit

6,764

6,361

20,020

0.30

0.34

 Veracruz

4,490

8,834

53,407

0.08

0.14

 San Luis Potosí

7,627

8,363

28,061

0.13

0.16

 Yucatán

1,341

1,448

5,439

0.49

0.31

 Zacatecas

10,724

9,737

31,305

0.22

0.20

      

Northern

Central

 Baja California

20,719

19,434

58,764

1.23

1.14

 Distrito Federal

9,295

8,260

21,476

0.49

0.20

 Baja California Sur

723

1,272

6,807

1.04

1.10

 Hidalgo

5,575

5,818

37,459

0.16

0.12

 Coahuila

4,856

3,762

19,621

0.39

0.26

 México

13,597

12,355

60,231

0.14

0.13

 Chihuahua

14,162

12,950

49,841

0.29

0.37

 Morelos

6,127

3,883

25,080

0.19

0.11

 Nuevo León

7,712

6,046

17,422

0.20

0.27

 Puebla

9,748

6,640

40,276

0.18

0.11

 Sinaloa

5,223

5,165

23,081

0.21

0.19

 Querétaro

2,635

3,599

16,683

0.20

0.21

 Sonora

6,881

6,290

37,104

0.58

0.61

 Tlaxcala

969

938

6,779

0.22

0.14

 Tamaulipas

7,239

7,220

34,238

0.25

0.29

      

National

267,338

238,331

985,383

0.22

0.20

      

Source: 2000 and 2010 Mexican Population Census, and 1995 and 2005 Population Counts

aPopulation of 5 years and older that lived in (1995, 2000, 2010) in the United States and was living in Mexico in (2000, 2005, 2010) at the moment of the Census or Count

bThe rate is calculated as the number of inter-censal return migrants in (2000) 2005 over the number of non returnees of the period (1990–1995) 1995–2000

Of the 238,331 returnees, 90.4 % are in states with a rate below while only 9.6 % (22,770) are in states with a rate above. There is evidence that the geography of return influences the demography of return, with states with rates above differing from states with rates below on several demographic indicators.24 In order to explore how the characteristics of returnees vary with type of destination, we examine the selected characteristics of returnees and their households for those living in the three types of states according to their rate of return (see Table 4). In terms of the gender composition of returnees, we see that female participation among returnees decreases along with rate of return. Returnees living in states with rate above are less likely to be men (61 %) than those in states with rate below (64 % above the national rate and 69 % below the national rate). Age does not differ much, although the mean age is slightly greater in places with rate above one and those of 15 years and older are more likely to have higher levels of education since their mean years of schooling (10.1) is greater than for those in states with rate below (8.7 and 7.9).
Table 4

Selected characteristics of returnees and their households by rate of return of the state where they are living in 2005

Characteristic

States “above one”

States “below one”

Rate ≥ 1

0.2 ≤ Rate < 1

0 ≤ Rate < 0.2

Individual level

Total

22,770

122,109

93,452

 %

9.55

51.24

39.21

Sex

 Male

60.9

63.8

68.8

 Female

39.1

36.2

31.2

Age groups

 5–19

22.4

22.6

19.2

 20–34

33.9

40.2

43.8

 35–49

26.7

24.4

26.1

 50–64

12.6

9.0

8.0

 65+

4.5

3.7

2.9

Mean age

33.0

31.3

31.3

Position in household

 Head

43.6

42.8

47.1

 Spouse or partner

19.4

15.4

12.5

 Son or daughter

24.2

30.9

30.0

 Other

12.8

11.0

10.3

Education (aged 15 and older)

 Mean years of schooling

10.1

8.7

7.9

Urban level of locality of residence: population

 >500,000

62.0

25.2

13.3

 100,000–500,000

9.8

18.3

12.5

 15,000–100,000

14.1

14.8

16.6

 2,500–15,000

6.3

15.0

21.0

 <2,500

7.9

26.7

36.7

Household level (household with at least one returnee)

Total

14,318

82,895

69,446

 %

8.6

49.7

41.7

Type of household

 Nulcear

54.2

62.9

64.5

 Extended

23.0

25.9

28.3

 Mixed or other familiar

3.6

2.0

1.5

 Unipersonal

15.9

7.9

5.2

 Non-familiar corresidence

3.4

1.3

0.5

Average number of returnees

1.6

1.5

1.3

Source: 2005 Population Count

Note: Includes 238,331 non-institutional individuals living in 166,659 private households. The returnee population refers to the population that was living in the US in 2000 and in Mexico in 2005 at the moment of the Count

The position in the household presents some differences between states with rates above and below. Possibly due to a larger female participation in states with rate above one, we see that in these states there is a larger proportional presence of returnees who are spouses or partners of the head of the household. Returnees in states with rates above are more likely to be head of households compared to states with the lowest rates, but less likely compared to the middle range states. In states with rates above one, returnees are less likely to be sons or daughters of the head25 and more likely to have no family relationship with the head. Related to these, the types of households are different. In states with rate above, households are less likely to be nuclear, less likely to be extended family, more likely to be mixed families, more likely to be unipersonal and more likely to be co resident with no family relationship than in states with rate below. The average number of returnees present in a household is slightly larger in states with rate above than below. The patterns on living arrangements suggest that places attracting return migrants are attracting people and whole households who did not originate there such as people living alone or as non-family member of a household.

Increasingly, a lower proportion of returnees go back to traditional sending areas and a larger proportion is found in northern, south and southeastern states. Figure 1 shows the regional distribution of inter-censal return migration to the United States for the period 1995–2010. We can see that of the total numbers of people returning to Mexico after being in the US five years previously, there is a decrease of returnees to traditional migration states from 53.6 % of total returnees in 1995 to 43.4 % in 2005; and although we had in 2010 a dramatic increase of returnees, the declining trend persists with 37.7 % of returnees in traditional sending states. In contrast, the northern region increases from 15.9 % to 26.9 % of the total in 2005 and then proportionally declines for 2010 to 24.4 %. The increase in the proportional presence of returnees in the South and Southeastern states observed in 2005 is larger in 2010. Note that, in 2005, more than half (137,315 or 56 %) of the returnees are not coming back to traditional sending states, while this was the opposite in 1995.
https://static-content.springer.com/image/art%3A10.1007%2Fs11113-012-9243-8/MediaObjects/11113_2012_9243_Fig1_HTML.gif
Fig. 1

Regional distribution of inter-censal return migration, 1995–2010 (percentages)

A similar pattern occurs in the case of intra-censal migration (see Fig. 2a, b). The traditional sending areas decrease their share of total intra-censal return migration to the United States between the three periods of observation (from 63.4 to 38.3 %) while decreasing the share of intra-censal non-returnees (49.5–37.7 %). Traditional sending states decrease their share of migrants returning within the five year period more rapidly than their share of the migrants who remain in the United States. By this measure, traditional sending states are becoming states that are less likely than in the past to attract back the migrants that leave them. People are not returning to traditional regions as much as would be expected given their past patterns of international migration.26 We see that the trend presented in 1990–1995 and 2000–2005 continues in the period 2005–2010 and is greater in magnitude. We also see in these figures that there was a proportional increase in intra-censal return and non-return in the central and south and southeastern states. Relative to traditional areas, the new areas of origin account by 2010 both for a greater share of migrants who return within 5 years and a greater share of migrants who stay in the US. This suggests a changing pattern of return migration due to the appearance of new areas of origin and the development of new areas of attraction for returnees.
https://static-content.springer.com/image/art%3A10.1007%2Fs11113-012-9243-8/MediaObjects/11113_2012_9243_Fig2_HTML.gif
Fig. 2

Intra-censal a return and b non-return migration by region, 1990–2010 (percentages)

Part of the literature gives special attention to the characterization of origin by the rural–urban distinction. For the case of return, we see that there is a rural-urban27 duality with a bipolar distribution: 44.7 % of the returnees in 2005 are in rural localities with less than 15,000 habitants. The pattern of return migration by size of locality (results not shown here) in 2005 is similar to that of 2000, but with a consistent decrease in absolute numbers, for all type of localities. If we look at the rate of return migration by size of the locality, we see that the only areas with rates above average (0.2) are medium sized localities: 0.34 for localities with a population size of 15,000 to 99,999 habitants and 0.6 for localities of 100,000 to 499,999 habitants. The rate of return for localities with a population under 2,500 is 0.14 while the rate for localities with a population between 2,500 and 14,999 as well as for localities over 500,000 habitants is 0.18.

There is a proportional increase in inter-censal returnees in rural localities with less than 15,000 habitants. These returnees are concentrated mainly in states with a large number of rural localities and with recent emigration, like Oaxaca, Veracruz, Puebla and Chiapas. These are states that incorporated later than the traditional sending areas into the Mexico–US migration stream and who have been migrating into new destinations in the US (Riosmena and Massey 2012). Therefore, migrants from these states are less likely to have established strong communities in the United States with ample resources and strong social networks that could help migrants sit out the US financial crisis and the economic recession of 2008–2009 and this could be a possible explanation for the increase of inter-censal returnees in these localities. For intra-censal migrants who left Mexico between 2005 and 2010, there is a sharp contrast between localities of different sizes in the proportions returning and not returning. Overall, 69 % of migrants in this period have not returned to Mexico. Of all those that have not returned, the majority (60.7 %) are from rural localities of less than 15,000 habitants, but of all those that have returned by the end of the period, a minority (46.4 %) are from localities of less than 15,000 inhabitants. In the 5 years previous to the 2010 census, households in small places send the most migrants to the US, but fewer of them return compared to those migrating from larger places.

Data show that most of the migrants are not coming from the poorest municipalities. The degree of marginality and the index of migration intensity show an inverted U relationship (United Nations Development Programme 2007). That is, municipalities with greatest and least marginality have lower migration intensities than those with a moderate degree of marginality. If we graph return migration and the marginality index28 in 2005 (results not shown here), a general finding is that even if return migration is larger in localities with low marginality and in states with low levels of poverty, it is not exclusive to such places. In this sense, we see that return migration follows a similar inverted U pattern with marginality as does out migration itself. Regarding the relationship between the rate of return migration and capacities poverty at the state level, we observe a similar pattern to that of marginality. However, states with low levels of capacities poverty differ in their rates of return migration, which indicate that there are other conditions associated with choosing a place to which to return.

New Destinations of Return

Why is return migration to traditional migration states declining? At least two possible reasons could explain this. Established migrant communities in the United States retain migrants. It is also possible that those leaving traditional regions do not come back to where they left even if they return to Mexico.

There is a change in the localities with most returnees from 2000 to 2005 (see Table 5). In 2005, 25 localities concentrate 24 % of the returnees. These are all urban areas with the following characteristics: they are border cities, metropolitan areas, localities in traditional migration states or new tourist areas. 25 % of all the returnees are moving from the US into border states. The top localities are Tijuana, Guadalajara’s Metropolitan Area, Ciudad Juárez, and Mexicali. In 2000, 25 localities concentrated almost 20 % of the returnees. The top localities were Guadalajara’s Metropolitan Area, León, Morelia and the Mexico City Metropolitan Area. Notice that the pattern of the localities with most returnees changes in the period of 2000 and 2005. In 2000 there is a larger presence of localities in traditional migration states and near Mexico City. The position in 2000 of the localities with most returnees in 2005 is very different; although 16 are both in the top 25 in 2000 and 2005.
Table 5

The 25 localities with most returnees in 2000, 2005 and 2010

Position in 2000

2000a

Position in 2005

2005b

Position in 2010

2010

Locality

Returneesc

Position in 2005

 

Locality

Returnees

Position in 2000

 

Locality

Returnees

Position in 2005

1

Guadalajara

6,551

3

1

Tijuana

10,843

25

1

Tijuana

32,864

1

2

Zapopan

3,330

5

2

Juárez

5,414

12

2

Juarez

16,198

2

3

León

3,102

14

3

Guadalajara

4,031

1

3

Mexicali

11,520

4

4

Morelia

2,548

7

4

Mexicali

3,526

81

4

Matamoros

9,830

12

5

Gustavo A. Madero

2,476

24

5

Zapopan

2,996

2

5

Guadalajara

9,090

3

6

Cd. Nezahualcóyotl

2,472

36

6

Chihuahua

2,472

9

6

Durango

8,433

9

7

Aguascalientes

2,451

8

7

Morelia

2,335

13

7

S.L. Rio Colorado

8,433

18

8

Ecatepec

2,427

23

8

Aguascalientes

2,196

7

8

Leon

8,240

14

9

Chihuahua

2,236

6

9

Durango

1,907

10

9

Chihuahua

8,147

6

10

Durango

2,207

9

10

Iztapalapa

1,828

11

10

Ensenada

7,771

13

11

Iztapalapa

2,173

10

11

San Luis Potosí

1,787

14

11

Morelia

7,486

7

12

Juárez

1,878

2

12

Matamoros

1,713

216

12

Culiacan Rosales

6,609

20

13

El Pueblito

1,800f

300

13

Ensenada

1,594

45

13

Hermosillo

6,542

21

14

San Luis Potosí

1,741

11

14

León

1,570

3

14

Aguascalientes

6,399

8

15

Monterrey

1,682

16

15

Tepic

1,504

20

15

Zapopan

6,384

5

16

Naucalpan

1,631

50

16

Monterrey

1,447

15

16

Reynosa

6,326

19

17

Puebla

1,485

22

17

Nuevo Laredo

1,416

129

17

Ecatepec

5,657

23

18

Torreón

1,342

43

18

S. L. Río Colorado

1,368

189

18

Puebla

5,377

22

19

Cd. Guadalupe

1,267

31

19

Reynosa

1,358

65

19

Nuevo Laredo

4,945

17

20

Tepic

1,213

15

20

Culiacán Rosales

1,273

36

20

Tepic

4,740

15

21

Querétaro

1,192

28

21

Hermosillo

1,271

133

21

San Luis Potosi

4,620

11

22

Tlaquepaque

1,192

26

22

Puebla

1,175

17

22

Cancun

4,343

25

23

Tonalá

1,144

38

23

Ecatepec

1,108

8

23

Iztapalapa

3,217

10

24

Tlalpan

1,120

56

24

Gustavo A. Madero

1,100

5

24

Monterrey

2,315

16

25

Tijuana

1,098

1

25

Cancún

1,084

137

25

Gustavo Madero

2,125

24

Total (20 % of all returnees)

49,958

 

Total (24 % of all returnees)

52,578

 

Total (18 % of all returnees)

180,991

 

Source: 2005 Population Count at the locality level (ITER2005) and 10 % sample of the 2000 and 2010 Population Censuses

aThe estimations of 2000 are generated using the 10 % sample of the 2000 Population Census; the weighted data is aggregated at the locality level since the database ITER 2000 does not contain the number of returnees in each locality

bThe ITER 2005 contains the variable of number of returnees in the locality for 2005 and it was used for these estimations

cUnlike the data for 2005 which is provided by the ITER2005 database, this number is an estimator using the weighted 10 % sample of the 2000 Population Census

We can note new places of attraction for returnees in 2005: Mexicali, Matamoros, Ensenada, Nuevo Laredo, San Luis Río Colorado, Reynosa, Culiacán, Hermosillo and Cancún. These findings suggest the growing importance of border locations and tourist destinations for return migrants, most likely due to an increase in involuntary deportations from the United States or an increase in employment opportunities product of tourism or the maquiladora industry, for example. This is coherent with what has been found in intermediate cities in Jalisco (Papail and Arroyo Alejandre 1993) and growing metropolitan areas like Tijuana and Puerto Vallarta (Serna Enciso 2008) where migrants settle in cities that are attractive both for returnees and internal migrants. This change certainly raises questions for further research looking, for example, on the processes or mechanisms that made these cities attractive for those with US migratory experience or alternatively, we could ask why these returnees were not attracted to the same places as had previous return migrants. One possible explanation could be given by the literature of the New Economic Geography where agglomeration effects, social networks and selectivity of migrants explain different patterns of settlement (Kanbur and Rapoport 2005; McKenzie and Rapoport 2010).

When we analyze the localities with most of the returnees in 2010 (see Table 5), the pattern is very similar to that found in 2005 although these localities concentrate less returnees (18.3 %) than before, indicating more dispersion. Overall, we find an increase of return in border places and a decrease in traditional sending areas. Tijuana and Juárez continue to be the localities with most returnees but we see that other border cities like Matamoros, Tamaulipas and S.L. Río Colorado, Sonora are now in the top ten localities, while they were not in 2005. Guadalajara, Jalisco falls in position compared to 2005, while Leon, Guanajuato is the only place in a traditional sending state that gains importance in the top localities. Overall, although the number of inter-censal returnees in 2010 increased dramatically, the geographic pattern does not change from the one observed 5 years ago after a decline in return.29

We will briefly compare some characteristics of the returnees that live in 2005 in the 25 localities listed in Table 5 and those who do not (see Table 6). Returnees in the “top 25” localities have a slightly lower mean age, are less likely to be men in contrast to the general population of returnees, and to those in traditional areas, and are likely to have higher levels of education than those living in other localities. Returnees in these “top” localities are less likely to be living in nuclear families, almost as likely to be in extended families, more likely to be unipersonal households, more likely to be co residing with non-family members and more likely to be in mixed families than returnees in the rest of the localities.
Table 6

Selected characteristics of returnees and their households by locality where they are living in 2005

Characteristic

Not in top 25 localities

Top 25 localitiesa

Individual level

Total

182,723

55,608

 %

76.67

23.3

Sex

 Male

67.1

60.2

 Female

32.9

39.8

Age groups

 5–19

20.2

24.6

 20–34

42.1

37.6

 35–49

25.2

25.6

 50–64

8.9

9.2

 65+

3.6

3.1

Mean age

31.6

30.9

Position in household

 Head

45.1

41.2

 Spouse or partner

13.9

17.2

 Son or daughter

30.3

28.7

 Other

10.2

13.0

Education (aged 15 and older)

 Mean years of schooling

8.1

10.0

Urban level of locality of residence: Population

 >500,000

5.8

83.9

 100,000–500,000

14.9

16.1

 15,000–100,000

20.1

 2,500–15,000

21.6

 <2,500

37.6

Household level (household with at least one returnee)

Total

130,533

36,126

 %

78.3

21.7

Type of household

 Nulcear

64.4

57.3

 Extended

26.7

26.7

 Mixed or other familiar

1.7

2.9

 Unipersonal

6.5

10.8

 Non-familiar corresidence

0.8

2.3

Average number of returnees

1.4

1.5

Source: 2005 Population Count

Note: Includes 238,331 non-institutional individuals living in 166,659 private households. The returnee population refers to the population that was living in the US in 2000 and in Mexico in 2005 at the moment of the Count

aTo see the full list of the localities included in the group named “Top 25 localities” see Table 5

Discussion

Comparing the characteristics of returnees over time, we found an increase in male participation, an increase of head of households and, most importantly, a reversal of the trend in declining return until 2005 and a sharp increase by 2010. However, despite the changes in the volume of return migration, we find that the trends in the changes in the geography of return hold. That is, our main finding is that we observed a change in the geography of return migration in Mexico. Overall, we find a decline of relative importance of return in traditional sending states. We detect states and localities that are more likely to receive returnees and comparing their characteristics we find evidence for the interaction between geography and demography.

The change in the geography of return suggests differences in the stage of migration. For example, the stage in the migratory process of Veracruz (a new important source of migrants) is very different to the case of Jalisco (a state with migratory tradition) or to Quintana Roo (where a city like Cancun has attracted large amount of returnees, as well as internal migrants). This suggests that the customary distinctions between traditional and non-traditional sending states should be extended to include distinctions between receiving states. Just as there are new origins of migration, there are new destinations of return. Some destinations appear to be less dependent on prior patterns of out-migration than on emerging patterns of economic opportunity within Mexico. The new destinations are new partly because they are in states that only recently had begun to send migrants to the US and partly because returnees are going back to places other than their places of origin.

At the beginning of this article we wondered whether the case of Tlacuitapa was representative or not of national patterns of return. In the context of a changing geography of emigration and return, we expect Tlacuitapa to differ from what is observed in other communities. Our findings suggest the need for qualitative research conducted in non-traditional emigration areas with high levels of return migration that help understand the relationship between emigration and return, as well as the challenges for reincorporation upon return. Our findings also suggest the need for the gathering of data related to the migration experience of returnees in the destination of return in nationally representative surveys and censuses, and not only in the household of origin (as in the extended questionnaire that asks about intra-censal migration).

What may be driving the changes in the patterns in the geography of return is a combination of new and old models of migration. Whereas the old model of short-term workers leaving families behind may not now be the predominant one in traditional sending areas, it may have more predominance in recently introduced places of emigration. Further, we speculate that these changes are affected by the fact that those returning to Mexico are returning to different places. Complicating these patterns is the impact of rising levels of formal deportation from the US. Unfortunately, this data does not allow for a thorough analysis of these issues but we hope this project stimulates further research. The return of migrants has an impact on the demographic composition of a community. In the community of return, new households may have been created through time not only due to transitions into adulthood, but because people may live alone when they do not return to live with their original family.

Migrant return also has economic implications in the labor market. On the one hand, returnees may be disadvantaged in the labor market because of the time spent outside the country and the weakening of their work-related networks in Mexico, but, at the same time, the skills they acquired may give them an advantage compared to locals that do not have the migratory experience. But deportees may lack references for the number of jobs held in the United States making it hard to prove the acquisition of these skills, for example. Returnees in 2005 were less likely to be employed in the formal sector (jobs with social security coverage) and they are less likely to have access to public health and social security. This does not necessarily indicate lower incomes, since small-scale entrepreneurs and the self-employed will not have coverage, but different types of protection. The Survey of the Reinsertion of Migrants (EREM from the Spanish name) in 2000 shows that an important proportion of returnees that were salaried employees before living in the United States start a small business and stop being employees after finishing their migration cycle. This entrepreneurial shift usually raised their income. Second, there are a large number of jobs created directly by return migrants, using skills learned in the US, such as English, that are mainly in the service sector, such as tourism, but with no access to social security. But transfers of skills are not limited to language acquisition (Hagan and Demonsant 2010). The study of the various implications of return in the Mexican labor market deserves further exploration.

Deportees face many other difficulties on reincorporation besides entry into the labor market. Special attention should be put on the incorporation into the educational system of the children of migrants that were born in the United States from Mexican parents or who migrated young. Given the increasing importance of families of mixed documented status in the US, deportation will disrupt family life either when the deportee decides to return with the whole family or alone. Both family separation and the return with children challenge reincorporation in Mexico. Moreover, cross-border policy discussions should consider the fact that a proportion of the Mexican return migrant population has stronger family ties in the United States than in their country of birth and that families are broken apart after deportation.

The Mexican government needs to take seriously the problem of return migration. Social policy is needed to protect migrants with social security and with job creation that enable reincorporation. Special programs are needed to enable the transfer of skills between migrants and non-migrants to improve the human capital of the community. This co-operation and transfer can potentially facilitate integration into communities, whether the original ones from which migrants left or the new ones to which they have returned. In addition, returnees constitute a mobile population that has already been exposed to migration and may easily move again, either back to the United States or elsewhere in the country. Their mobility may make difficult social policies designed to re-integrate migrants into their communities of origin.

Return migration may contribute to new types of inequalities. This may be the case regarding access to social security and health services (or implicitly access to employment in the formal sector) where returnees are at a disadvantaged position compared to non-migrants. There is a need for a mobile insurance that could guarantee access to health services, insurance and pensions to this mobile population based on contributions and their medical insurance history in Mexico and the United States.

The pattern of migrant return has important implications for local development, both in the communities of return and maybe most importantly in the communities of origin. First, out-migration combined with migrants returning to different communities may be emptying some places, changing their demographics and their economic base as remittances decline. In these places, local development projects may not be a solution for re-integrating people into their original communities. These implications are not limited to the Mexican case but to other migrant sending countries. Ultimately, our understanding of the interaction between the demography and geography of return could provide areas of opportunity for better re-integration and the well-being of returnees and their families in both sides of the border.

Footnotes
1

The benchmark is inexact since those returning by 2005, for example, may have been living in the US before 1995, suggesting that regions and places with a long migration tradition should have proportionately more return migrants.

 
2

The DHS lists separately the number of inadmissible or deportable aliens who are returned. This is the only source of official US records on exits of Mexicans by year. For their classification, return is distinguished from deportation, which is based on an order of deportation and the DHS notes that most returns are Mexicans apprehended by the Border control who voluntarily agree to return. Returns have consistently been higher than deportations (based on an order). Multiple counting of an order of deportation/removal for the same individual in any 1 year can occur, but only between categories as when someone removed earlier without criminal charge seeks to return to the US and is then deported with a criminal charge—that of violating the earlier order of removal (U.S. Department of Homeland Security 2010).

 
3

Although returns captured in the census are not the same thing as returns captured by the DHS, they provide a benchmark for comparison between sources of data from both countries.

 
4

The Population Count or Conteo is a mid-decennial census count of the entire Mexican population with a reduced questionnaire compared to the decennial censuses.

 
5

The number of inter-censal returnees is likely to be underestimated by the fact that a proportion of deportees are likely to re-enter the US. Their re-emigration from Mexico is an indicator of the involuntariness of return associated with deportation. Another possible source of undercount is the fact that respondents in Mexican households may not declare that a member of the family was living in the US five years ago because they were deported and of the stigma associated to it. Again, this would support the claim of the involuntariness of return by deportation.

 
6

For a discussion on the concept of preparedness, see Cassarino (2004).

 
7

This analysis of US data uses residual methods or the matching method proposed by Van Hook et al. (2006) to estimate return migration, or the emigration of the foreign born. From the perspective of the US, it is impossible to talk to the geography of return in Mexico and limits to the estimations of the magnitude of return. For example, Hill and Wong (2005) have used jointly data from both countries studying from a binational perspective estimated age and gender distributions.

 
8

We have no information on intra-censal migration during the period 2000–2005, but compared to the period 1995–2000, the period of 2005–2010 shows a decline in outmigration (from 1.47 to 0.99 million emigrants in the intra-censal period). A possible explanation for this could be related to the negative effects of the economic crisis and recession in the US and Mexico.

 
9

After the publication of the 2010 Mexican census, the debate on the explanations for their impacts is likely to continue, especially because the 2010 census showed a dramatic increase of inter-censal migrants while Rendall et al. (2011) found that return declined over the 2005–2009 period using the ENOE. Passel and Cohn (2009) have argued that ENOE may over-represent return migrants who move to the US for short periods of time. The measures of return migration used by Rendall et al. and by the census are not comparable. The census measure is based on whether an individual returned from the US within a five year period, but does not count multiple returns by that individual in the period. ENOE reports a return within a 15 month period. The ENOE sample could thus capture up to four instances of a repeat return by an individual migrant within the census’s five year period. Migrants who make multiple trips to the US in a five year period will thus be oversampled by ENOE. One possible explanation for the decline in return migration showed by Rendall et al. (2011) is consequently a decline in circular migration by 2010, which is credible in face both of declining job opportunities in the US and the increasing security risks facing by migrants travelling to the northern border area. Another possible explanation is the underestimation of the return of complete households in ENOE, as explained in a recent report from Pew Hispanic Center (Passel et al. 2012).

 
10

As noted by one of the anonymous reviewers, a possible explanation for the increase in the number of inter-censal returnees in 2010 is declining re-emigration over the period. Due to data limitations, it is impossible to compare re-emigration rates of inter-censal returnees over time. However, an alternative way of showing a decrease on re-emigration is by showing an increase in duration. Comparing data from 1995, 2000 and 2010 we see that the mean number of days of the duration of the trip of intra-censal returnees increased considerably from 414 to 442 to 620 days for those who left and came back in the periods 1990–1995, 1995–2000 and 2005–2010, respectively.

 
11

Contrary to other data sources, the complete set of records of the 2005 Conteo provides us data that is representative at the locality level. For example, although the National Survey of Demographic Dynamics (ENADID) contains richer data and questions related to migration, it is only representative at the national, state and locality size level. Other survey data available present limitations in terms of their nationally representativeness or comparability with the census and other nationally representative data. For example, the Mexican Migration Project, a major effort of data collection, does not provide us with data that is representative of Mexico or its emigrants and returnees. The National Survey of Occupation and Employment (ENOE), a quarterly household survey recently used by Rendall et al. (2011) to study return migration, started in 2005 and is preceded by the National Survey of Employment (ENE) and National Survey of Urban Employment (ENEU) that capture a different migrant population in the time period we are studying. For a larger comparison between ENOE and ENADID, see Rendall et al. (2010), for a comparison between the Survey of Migration at the North Border of Mexico (EMIF) and ENADID, see Rendall et al. (2009).

 
12

Of the 244,426 returnees, only 2.49 % live in collective dwellings and 199 are homeless. We will only include in the analysis at the individual and household level, non-institutionalized individuals, i.e. 238,331. The population of interest will be specified accordingly.

 
13

The definition of a returnee as someone that was in the United States five years ago implies that we are limiting the study to recent returnees. This consideration should be taken into account when comparing this paper with others using other data sources; for example, studies using the Mexican Health and Aging Study (MHAS), a nationally representative, prospective panel study of Mexicans aged above 50 in 2000. Two-thirds of those male that had returned and are captured by the MHAS had done so 20 years ago or before.

 
14

Other limitations of the data are that the data does not include place of birth, so some individuals considered as returnees may be American expatriates or previous migrants from Central America and other countries that are trying to re-enter the United States. The Population Count of 2005 does not have the date of departure and arrival, or the place of origin or last emigration so we cannot determine if they returned to the original place they departed from. Also, it does not include the question of cause of emigration like the 2000 Census.

 
15

The intra-censal migrant could have moved to the US more than once during the 5 year period. However, the census only asks information (date of departure, date of return, etc.) about the last trip made during this period.

 
16

The definition of a household changed in the 2010 Census. Previous data sources defined households based on the common consumption of food of the members that reside together. This allowed for multiple households residing in the same dwelling. However, the 2010 Census does not capture data any more at the household level and groups together all members living in the same dwelling. This is likely to affect the definition of unipersonal and extended households and non-kin living arrangements that are more likely to have different food expenditures.

 
17

The 2010 Census limits the question to the members that were living in that dwelling 5 years ago. This is more likely to affect states and localities undercounting intra-censal migrants that may be more likely to move from one household to another before and/or after living in the United States.

 
18

For a discussion on the coverage of Mexican migrants in census data, see Hill and Wong (2005) and Ibarraran and Lubotsky (2007).

 
19

The emigration of complete households will lead to an underestimation of intra-censal outmigration. But for our analysis, the issue is whether the complete household emigration is disproportionately concentrated in certain places, thus underestimating the emigration from those places. Complete household emigration is likely when there are no subsistence opportunities in the community of origin, and is reinforced by the absence of extended family members to support those left behind. The first condition is more likely to occur in small, marginal rural communities and the second in cities. This balancing of precipitating factors makes it unlikely that complete household emigration varies by size of place, although this requires research.

 
20

A proper rate of return should include in the denominator the whole set of population at risk to return. However, since our interest will be to calculate this rate at the state level, for this we would need to know the state of origin of the Mexican born population living in the United States and there is no US data source, to our knowledge, that gives a nationally representative estimation of the state of origin in Mexico.

 
21

In order to maintain comparability with 2005, we do not distinguish by place of birth; thus, this includes people born in the United States. However, the immigration of U.S.-born individuals is strongly related to return migration. In 2010, 71 % of the U.S.-born individuals where actually living in Mexico with at least one Mexican parent in the same dwelling. In the period 2000–2010, the number of people born in the US duplicated from 342,000 to 740,000 and the proportion of minors increased from 73 % in 2000 to 77 % in 2010. The number of people born in the US who were living there 5 years before (counted with the inter-censal returnee population) almost tripled from 58,000 in 2000 to 152,000 in 2010. This increase was mainly due to the increase in the number of minors born in the US who arrived to Mexico in the five previous years.

 
22

It is beyond the scope of this article to explain this increase. However, possible explanations could be related to the adverse economic situation, as well as an increase in deportations, an increase in return of documented migrants that went to the US temporarily or the return of green-card holders who have lived longer periods and can re-emigrate to the US in the future.

 
23

These labels (“rate above” and “rate below”) are appropriate because of the way we are defining this rate. They differentiate the states that had greater or lower return migration in 2005 given the level of their out (non-return) migration in 2000.

 
24

Since we are dealing with the universe of returnees, the differences reported are real not estimated ones.

 
25

There could be a bias created with emigrant sons that returned to different places from which they left, so we could be inflating the return of sons in villages. In order to control for this and for families going back together to different places, these figures should be compared with the information of the 10 % sample of the 2000 Census.

 
26

This result is even stronger if we remember that out migration from traditional states was very high in the past.

 
27

The classification of rural and urban localities in Mexico is done with the population size of the locality. A rural locality is defined by INEGI if it has a population under 2,500 habitants. However, it is also common to use the 15,000 habitants’ cutoff to define a rural area, just like the authors do. The authors will use both for different purposes but will always specify the definition of rural locality that is used each time.

 
28

The Migration Intensity Index (MII), and its categorical version (Degree of Migration Intensity) at the state and municipal level are generated by the Mexican Population Council (CONAPO). The MII includes information of the presence of remittances, emigrants, circular and return migrants (Tuirán, Fuentes and Ávila 2002). The Marginality Index (Índice de marginación) produced also by CONAPO is generated at the locality level using the technique of Principal Components and summarizes educational and dwelling characteristics at the state and locality level. The Degree of Marginality is the categorical version of the index (Anzaldo and Prado 2007). The index of marginality has been the indicator of social exclusion most often used in Mexico (Cortés 2002; Hernández and Székely 2005) and is our measure of poverty. The index of marginality differs with other definitions of poverty currently used in Mexico (Consejo Nacional de Evaluación de la Política de Desarrollo Social 2010) since it is more infrastructure based.

 
29

Although the geographic pattern in 2005 and 2010 are similar, the reasons associated to those patterns are likely to be different. Conditions at the Northern Mexican border changed over the period 2005–2010 with an increase in violence and insecurity related to organized crime and narco traffic, and a decline in employment related to the global crisis of late 2000s and the decline of the maquiladora industry and business closures. As a result, there has been an increase in poverty related to earnings in the border states (Consejo Nacional de Evaluación de la Política de Desarrollo Social 2011). Although the border is less attractive for these reasons, increases in deportations may still counterbalance the adverse economic effects making returnees to spend some time there before trying to go back to the US.

 

Acknowledgments

The authors would like to thank the Secretaría de Desarrollo Social (SEDESOL) for sharing the restricted data after the agreement between the University of Texas at Austin and SEDESOL, and the Population Research Center (PRC) at UT-Austin for allocating the data and providing technical computational support. Parts of the empirical research were funded through the C.B. Smith Sr. Chair in US-Mexico Relations # 1. Claudia Masferrer would like to thank the PRC and the Teresa Lozano Long Institute of Latin American Studies for their support as a graduate trainee and graduate research assistant. C. M. was partially funded by Consejo Nacional de Ciencia y Tecnología who granted financial support from 2007 to 2009 and by the Fonds Québécois de la Recherche sur la Societé et la Culture through the PBEEE award. Preliminary versions of this work were presented in 2009 at the Population Association of America meeting in Detroit, the Seminario Permanente de Estudios Migratorios del Occidente de Mexico held at CIESAS in Guadalajara, and at the conference “Migration during an era of restriction” held at UT-Austin. Earlier versions where presented in 2008 at the Joint Statistical Meetings in Denver and the conference “Mexico–U.S. Migration: Rural Transformation and Development” held at UT-Austin. We appreciate enormously the interesting comments and feedback provided by the participants in these meetings. We are grateful to Agustín Escobar and Jacqueline Hagan for reading complete versions of earlier manuscripts. Last but not least, we would like to thank the anonymous reviewers for their careful revisions and recommendations.

Copyright information

© Springer Science+Business Media B.V. 2012