Introduction

In the early 2010s, return migration received increasing attention among Mexico’s social, governmental, and research actors. It was mainly due to the increase in return migration that arose during the 2008 economic crisis and the deportation policies that affected the immigrant population in the United States, among which the Secure Communities program (SC), implemented in 2008, stands out. According to the 2010 census in Mexico, 826 000 Mexicans were identified as residing in the United States five years earlier, an increase of 237% over 2005. This increase had enormous implications in the receiving regions. There were even very unforeseen situations, such as the one observed in Tijuana, where a group of returnees opted to settle in El Bordo (a river channel near the border), living in holes in the ground or shelters built with scrap material. Most of them returned from the United States due to deportation (91.5%), where the majority spent at least six years (78%) (^{Velasco & Albicker, 2013}). The population of El Bordo reflects a very specific case of return migration that cannot be extrapolated to the entire phenomenon but helps to illustrate the diversity of situations.

Associated with the return there was also immigration of Americans linked to return migrants. ^{Canales and Mesa (2018}) estimate that between 2005 and 2010, 107 726 of these Americans immigrated to the country. Among these, school-age minors predominated, who, upon arrival in Mexico, faced several obstacles in their incorporation into the educational system. ^{Vargas (2018}) identified that, when looking for a place in schools, these minors faced management and economic barriers due to the enrollment of the transnational student being conditioned on the availability of space or the possession of identity documents. In addition, they had to face sociocultural barriers associated with the use of Spanish and the different school contexts and curricular content between the countries’ educational systems.

Five years after the increase recorded in the 2010 census, the Intercensal Survey showed a considerably lower number of return migrants (443 000), so it could be suggested that the phenomenon was in decline (^{Canales & Mesa, 2018}). However, shortly after that, there were warnings of a possible upturn and even a massive return, as the immigrant population in the United States was the target of threats from the then-presidential candidate, Donald Trump. From the beginning of his campaign in mid-2015, Trump followed a strategy of verbal attacks on migrants to direct the social dislike of Americans towards migrants. He particularly accused Mexicans of being criminals, rapists, and drug traffickers─unfounded assertions─; he also promised to expel undocumented people without distinction (with and without criminal records) and limit documented income. In addition, in his first months in charge, he took decisions and actions against immigrants, such as the reactivation of the SC; the adoption of an aggressive stance against sanctuary city policies; the revocation of the Deferred Action for Childhood Arrivals (DACA) and Temporary Protected Status, as well as the cruel and inhumane separation of migrant families at the border (^{Verea, 2018}).

In this context, the Mexican government and part of Mexican society wondered how Mexican compatriots in the United States would react to the constant harassment, how many would be deported, if there would be a similar or higher return than during the 2008 economic crisis, or if there would be an increase in the return of families, either due to deportation or fear of being deported. However, as the first years of Trump’s term passed, the indicators that usually give clues about return did not show significant increases, and there was even a decrease, for example, in the number of people expelled who were already residing in the United States, or the number of new residents in Mexican dwellings from abroad. Thus, it was possible to suggest that return migration did not show signs of increasing (^{Calva & Orraca, 2019}), but it remained to know its characteristics and to evaluate the magnitude of the phenomenon with more robust sources.

This paper proposes some answers to the above questions based on a quantitative methodology that analyzes the return of Mexicans, taking advantage of the publication in 2021 of the Population and Housing Census 2020, which was carried out from March 2 to 27 of that year. This source provides a more precise measurement of the volume of return migration. The period it covers, 2015-2020, coincides with Trump’s election campaign period and much of his presidential term. In addition, it provides a more detailed characterization of the individuals’ profile and an approximation of the return process based on typologies and the inclusion of the question on the reason for migration.

Return migration as an object of study

The proposed analytical framework for this study is based on analyzing changes in the volume of the phenomenon by return type for the period 2015-2020 and comparing it with what was observed in 2005-2010. The 2010 and 2020 population censuses are taken as the main source. Therefore, two axes are proposed: defining return migration and a strategy to analyze its characterization.

Regarding the definition of return migration, the basic aspects used to define it can be recovered. The first aspect to highlight is that two terms are regularly used when discussing this phenomenon: “migration” and “migrant population”. Migration refers to a move from one place to another that involves a change of residence, while the migrant is the one who makes this move. Conceptually and methodologically, it is not the same to study migration versus the migrant population since migration is an “instantaneous” event, which only lasts for the duration of the move (renewable and reversible) and can be canceled by return (^{Delaunay & Santibáñez, 1997}). In this study, the approximation is based on the migrant population, which is captured possibly up to five years after experiencing the return migration event.

The definition of migration is usually discussed along two axes: spatial and temporal. The first refers to the necessary distance between the place of origin and destination to be considered a change of residence. However, there is no room for confusion in the case presented here since it is a change of residence between countries. The second axis is temporality, and here the problem is to define the time necessary for the person to be considered an immigrant and not a temporary visitor. It is usually recommended that the minimum stay is one year. However, another position would be to try to include the variable on length of stay as a basic element of data collection and analysis, and that the evidence characterizes the situations about each type of stay. For example, in the Survey on Migration at the Northern Border of Mexico (Encuesta sobre Migración en la Frontera Norte de México [Emif Norte]), in the flow of Mexicans from the United States, particular characteristics are observed according to the length of stay. One characteristic is age, as those who spent less than a year tend to be young men and women or at very advanced ages. In contrast, those who spent more than a year are mainly concentrated at the center of the population pyramid (^{El Colegio de la Frontera Norte/Unidad de Política Migratoria-Segob/Consejo Nacional de Población-Segob/Secretaría de Relaciones Exteriores/Secretaría del Trabajo y Previsión Social/Consejo Nacional para Prevenir la Discriminación/Secretaría de Desarrollo Social, 2018, p. 28}).

As for the definition of return migration, Izquierdo (2011, cited by ^{Mestries, 2013, p. 177}) points out that the return migrant is the person who returns to his or her country of origin to remain in it after having been in another nation for a long time, while the event of returning to the country is the migration return. Another definition is presented by King (2000, cited by ^{Gandini et al., 2015}, p. 32), who indicates that return is the process by which people return to their country or place of origin after a significant period in another country or region. Both definitions are very similar and produce the same doubts: what does a long time imply? What is a significant period? Nevertheless, trying to answer these questions will lead to the same conclusions as above.

París Pombo and collaborators (^{2019, p. 24}) point out that it cannot be assured that the most recent migration will be the last. Therefore, a return does not necessarily imply the end of a migration trajectory or the end of the migratory process but a stage of continuum. In addition, ^{Jáuregui Díaz and Recaño Valverde (2014)} point out that by inquiring into migratory history, it can be discovered that the definition of return migration encloses a great complexity, as it is not subject to simple back and forth movement between the country of origin and destination, since the migrant may have transited through a third country or countries before reaching the final destination. As a reaction to the proposals analyzed, return migration can be defined based on the migration trajectory, the latter being understood as the sequence of places of residence and migrations that an individual experiences in his or her life. Return migration will be the event or process by which a person changes their residence to another place where they previously resided. In this way, a person may have one or more returns in their migration trajectory.

Regarding the study and characterization of return migration, authors such as ^{Stark (1991)} and ^{Mendoza Cota (2013)} use an economic perspective to explain return for labor reasons. Based on the postulates of neoclassical theory, they explain that return migration occurs when migrants fail to meet their expectations of benefit derived from migration and fail to maximize their income and achieve permanence in the destination country. Moreover, from the so-called “new economics theory of labor migration” they state that return migration results from a calculated strategy established in the household, modifying the focus from maximization of the individual to maximization within the household.

The above perspective focuses on labor migration. However, other situations, such as forced return derived from deportation or family reunification, require other approaches. One way to order and analyze the diversity of situations is the typologies that are part of the tools used to study diversity in human mobility. ^{Heyman (2015, p. 420)} indicates that the fundamental step in creating, using, and evaluating a typology is selectivity: what is selected and what is not, why, and with what effects? The same author argues that specific characteristics must be selected in a typology that should lead to a detailed analysis, a process of selection, and an association (clustering), which often results in obtaining one or several ideal types.

There are several exercises of typologies on return migration. ^{Cerase (1974)} presents one characterized by reflecting on return migration based on the difficulties or expectations immigrants face during integration in the destination country. He explains four possible types of return. The first type, return of failure, deals with immigrants who do not manage to integrate into the place of destination; and is explained because, in the process of incorporation, immigrants may feel disconcerted and even ashamed of their origin due to prejudices and stereotypes that the residents have towards them. In this situation, immigrants form social networks only with relatives or people from the same group. Over time, most overcome this stage; however, some do not and decide to return to their homeland. The second type, return of conservatism, is characterized by a process in which already settled immigrants face the choice between spending their income in the destination country or saving money to buy goods in their place of origin, with the plan to return at a later date. The third group, return of innovation, is characterized by immigrants who obtain new knowledge and skills in the country of destination but decide to return to the country of origin because they consider they will have greater satisfaction fulfilling goals formed after emigrating, that is, derived from the experiences in the country of destination. The fourth type is called return of retirement and is observed among older immigrants who finish their working careers but feel detachment towards the receiving society mainly because they have no descendants in that place, so they return to the country of origin.

^{Durand (2006)} presents a typology characterized by broadening the range of possibilities. He proposes six types of return: voluntary return of the established migrant; voluntary return of the temporary migrant; transgenerational return; forced return; return of failure; and the official programmed return. The first difficulty this typology faces is the inclusion of a group in which there is no return migration: the “transgenerational return,” which deals with descendants of the immigrant born in the destination country. ^{Gandini and collaborators (2015}, p. 36) point out that this group makes up another target population correlated with return, but it is not justified to include it in a typology of return. In the case of the voluntary return of the established migrant, ^{Durand (2006)} includes those who return voluntarily and after a long stay, perhaps after having documents in order and having adopted another nationality. In the case of the return of the temporary migrant, it refers to temporary workers subject to specific government programs where the contract requires or forces the return. This last group can again generate debates since its members may not be considered immigrants in the destination country. Perhaps because of such situations, ^{Cerase (1974)} emphasized the analysis of cases of migrants who intended to settle in the destination country. However, in the case of the mobility of Mexicans to the United States, there is an important flow of people who travel with the intention of staying only for a while, working, saving, and returning to their country. On the other hand, the return of failure, as defined by Durand, is very similar to the one proposed by ^{Cerase (1974)}. The last type he presents is the official programmed return, in which the characteristic is that there is a government program to relocate a group of people. There are other typology proposals with similarities, such as that of ^{Mestries (2013)}, which is composed of five types: 1) forced by deportation or unemployment; 2) periodic as part of the cycle of a pendular migration; 3) rest or pleasure returns: who return temporarily to their region of origin; 4) the definitive return of the successful migrant; and 5) the definitive return of the unsuccessful migrant.

The presented proposals illustrate the challenges of proposing a typology to study return migration. ^{Gandini and collaborators (2015}, p. 36) note that part of the difficulties arise because the proposals result from an intersection of different criteria for their elaboration since the same classification seeks to include motives, causes, different temporalities, voluntary or involuntary character, and the origin of the person (or that of their ancestors). In addition to the challenges in proposing a typology as an analytical tool, there is also the feasibility of obtaining empirical evidence, especially when using survey and census data.

In the context of the return of Mexicans from the United States, there is a special interest in distinguishing between forced return via deportation, forced return due to lack of work, and return of family members of people who were deported. The latter group is of particular interest due to their possible situation of vulnerability and difficulties in the process of (re)integration upon returning to Mexico. Trying to include all these elements without falling into a rigid typology, the proposal of ^{Canales and Meza (2018}) is taken up again, which analyzes three groups of migrants according to the process: individual return, accompanied national return, and accompanied binational return. This proposal seeks to measure and give a dimension to the collective/family versus individual configuration that the return process may assume. The advantage of this strategy is that the criterion that defines it is quite clear: to compare the collective configuration of the return versus the individual one. It does not take as criteria the motive or cause of migration for the construction of groups; however, this does not imply that these elements are irrelevant for the analysis. In this paper, within each type or group, the characteristics of the migrants involved are identified to distinguish the different return processes: forced by deportation, family reunification, labor issues, or other reasons.

Methodology and sources of information

The main data sources were the 2010 and 2020 Population and Housing Censuses. In both cases, there were two methodologies for the collection of information: the first census is exhaustive and uses a basic questionnaire including questions on place of birth and place of residence five years earlier, and the second uses an extended questionnaire applied to a sample of dwellings and adds a module on international migration. The tabulations of the basic questionnaire were consulted, while for the extended questionnaire, the databases available on the electronic portal of the Instituto Nacional de Estadística y Geografía (Inegi,) were used. An important characteristic of both censuses is that the concept of “census household” replaces that of simply household. The “census household” is the unit formed by the people residing in the same dwelling, regardless of their parental ties (^{Inegi, 2011, p. 27}).

In addition to the census, results from other sources are used to argue some ideas. For example, to analyze the return trend, the 2015 Intercensal Survey and the National Survey of Demographic Dynamics (Encuesta Nacional de la Dinámica Demográfica [Enadid]) 2014 and 2018 are used. These sources are surveys and therefore convey a statistic error, but they maintain the same logic to capture the phenomenon as the census, which is to ask about the place of residence five years earlier and therefore refer to the return observed in a specific five-year period. Data from Emif Norte, a set of independent surveys on flows of people and counts events, are also used. This work uses data on the flow from the south to argue about the trend of emigration of Mexicans to the United States and the flow from the United States.

Both the censuses and Emif Norte are approximations to the phenomenon of return migration. For space reasons, it is not possible to go into a detailed explanation, but the Emif Norte captures events of people whose return collects data on their migration trajectory and intentions about the future, specifically whether they will return to the United States (^{Calva & Coubès, 2017}). For its part, the censuses come close to identifying people who resided in the United States five years before, but it is not known how long they were there since the length of stay is not asked. In addition, it is unknown how long they have been in Mexico since their return (it can be from almost five years to a few days). Some clues as to how these temporal factors might be are obtained from the same census, from the module on migration. This module asks if in the last five years, someone in the household “went to live somewhere else”, the year of departure and, if applicable, the year of return. The results show that of the total number of people who emigrated during the months of 2015 covered by the census, 25% returned to Mexico before the census date in 2020, and 22% did so during 2015 itself, meaning that their stay abroad was less than a year. This shows that some of the census respondents also consider stays of less than one year as “living elsewhere”. So, it should be considered that, when using the census data to analyze returns, those who were away for a short stay, less than a year, could be captured as returnees, for example, the thousands of temporary workers or circular migrants traveling on work visas such as H2A and H2B. In the fiscal year 2015 alone, 153 475 of these visas were granted to Mexicans.

When analyzing census data, it is complicated to show the specificities described above, sometimes because the nature of the data does not allow to know in detail the processes or because, as ^{Masferrer (2021, pp. 20-21)} suggests, there are usually common myths or generalities that sometimes limit a more specific analysis. Usually, it is conceived that in the return, there is a predominance of men in productive ages; furthermore, since most of them are men, women are excluded from the analysis, so fewer experiences are known about women than about men. To address these situations, when analyzing the 2020 census data, the proposal of ^{Canales and Meza (2018}) is recovered by considering, on the one hand, the case of foreigners, mainly Americans who immigrate to Mexico with their relatives born in this country. And on the other hand, an approach to the diversity of migratory processes is presented based on typologies that, in this case, are constructed using the census household unit of analysis (dwellings). This approach identifies three types: dwellings with only one return migrant (individual return); dwellings with at least two return migrants but without the presence of international immigrants (accompanied return), and dwellings with the presence of at least one return migrant and a U.S. immigrant (binational accompanied return). Subsequently, to have clues about the migratory processes, the characteristics of the migrants and the relations between them, as well as the reasons for migration, are analyzed within each group. Particular attention is paid to differences in sex and age (see Table 1).

Return migration of Mexicans

Before presenting the return data for 2015-2020 and comparing it with previous five-year periods, it is important to remember that each period is characterized by economic, social, and political particularities. The 2005-2010 period was characterized by the economic crisis that affected the labor market, but there were also government actions that affected immigrants; for example, the SC created in 2008 by George W. Bush, through which undocumented migrants serving prison sentences were identified for deportation (^{Alarcón & Becerra, 2012}; ^{Ramos, 2018}). Obama canceled the SC in 2014, but Trump reactivated it in 2017. From its inception in 2008 through the fiscal year 2014, and from its reactivation on January 25, 2017, and through the end of that fiscal year, SC interoperability led to the removal of more than 363 400 foreigners (^{U.S. Immigration and Customs Enforcement (ICE), 2021}). As a result of this period of economic crisis and efforts to deport migrants, some indicators showed an increase in return migration. For example, in the Emif Norte there was an increase in the flow of Mexican nationals from the United States returning to Mexico due to lack of employment or having been deported and who also stated no intention of returning. The highest numbers of events were observed in 2007, 2008, and 2009, with more than 250 000 per year; subsequently, the number started to decrease (^{Calva & Coubès, 2017}). Consistent with the above, the census in Mexico recorded in 2010 a considerably high number of Mexicans who resided in the United States five years earlier: 825 609 (^{Canales & Meza, 2018}).

The 2015-2020 period was framed by the fear of an increase in the return of immigrants due to Trump’s anti-immigrant stance, who encouraged a discourse against the arrival of migrants, promoted his border wall project, and took actions against immigrants already residing in the United States. Two examples are the suspension of the DACA program in September 2017 and the reactivation of the aforementioned SC. As for DACA, Trump’s action implied not accepting new applications, the case went to trial, and in June 2020, the Supreme Court blocked Trump’s attempt; however, in theory, no new applications were received during that time, although there were renewals. Undoubtedly, Trump had a totally anti-immigrant stance; however, his actions were not accompanied by increased expulsions or deportations. If one compares the average annual events of Mexicans repatriated during Obama’s second presidential term and Trump’s first three years, a decrease is observed: 261 000 removed and 59 000 returned with Obama, and 206 000 and 43 000 with Trump (^{U.S. Department of Homeland Security [DHS], 2014-2019}).

Among the factors for the lack of an increased number of expulsions during the Trump administration is the decline in undocumented migration of Mexicans to the United States since, to a large extent, expulsions are of people attempting to enter without documents. According to Emif Norte’s data flow from the south, the number of events of Mexicans arriving at the border to enter the United States dropped from 748 000 to just 47 000 between 2008 and 2017. Another factor associated with the decrease in expulsions, especially from the interior, is the changes in the Mexican population residing in the United States. At the beginning of the five-year periods 2005-2010 and 2015-2020, the number of immigrants was very similar (11 164 770 and 11 906 325, respectively). However, there were changes in their characteristics associated with the probability of being deported. The examples are a decrease from 25.7 to 7.9 in the percentage of recently arrived immigrants; an increase from 23.3 to 30.1 in the percentage of those with U.S. citizenship; and an increase from 16.0 to 29.0¹ in the percentage of those in the age group of 50 years or older. In addition, the number of undocumented Mexicans decreased from 6.95 to 5.8 million between 2007 and 2014 (^{Passel & Cohn, 2019}).

Therefore, despite the anti-immigration context promoted by Trump, it is not surprising that the 2020 census in Mexico showed a decrease in return migration. It identifies 286 037 Mexicans who resided in the United States five years earlier, a fairly low level compared to 2010. Taking the return rate as an indicator─calculated as the number of people returning in the reference period, divided by the number of Mexican immigrants residing in the United States associated with the same period─reveals that the indicator increased from 24 per thousand in 2000-2005 to 71 per thousand in 2005-2010. However, in the most recent period, it dropped to practically the level observed at the beginning of the century (25 per thousand in 2015-2020) (see Figure 1).

Source: created by the author based on data from II Conteo de población y vivienda 2005 (^{Inegi, 2005}); Censo de población y vivienda 2010 (^{Inegi, 2010}); Censo de población y vivienda 2020 (^{Inegi, 2020}); Encuesta intercensal 2015 (^{Inegi, 2015}); Enadid 2014 and 2018 (^{Inegi, 2014}, ²⁰¹⁸) and the American Community Survey (^{Ruggles et al., 2021})

Figure 1 Five-year return rate and number of Mexican nationals returning from the United States, 2000-2020 

According to the 2020 census, among the returnees, there is a diversity of profiles: women and men of different age groups and educational levels. Nevertheless, men predominate (75%), and among them, the majority are between 20 and 59 years old (85%). Initially, this simple result suggests that individual return is more frequent than accompanied or binational return since, among the latter two types, there would be an equal distribution between men and women, as shown more clearly in the following section. Focusing on the greater presence of men, it can be seen that at the national level, there are 296 men for every 100 women; however, there are important differences by entity, which also reflect part of the migratory process. Among the entities that register high levels in the index are Chiapas (576) and Guanajuato (452), while in other entities, it is much lower, such as Mexico City (175), Chihuahua (196), and Nuevo Leon (207). The position in the census household of returnees shows that they are mainly heads of household (57.3%), more frequent in men (66.3%) than in women (30.1%), and among women, it is more frequent the situation of spouse or wife (39.3%) or daughter (17.7%) (see Table 2).

A comparison of the characteristics of migrants in 2010 and 2020 reveals important differences that denote the changes in the types of return that characterize each five-year period. There is noticeable aging in the population analyzed: in 2010, only 4.1% were 60 years of age or older, while in 2020, this rose to 14.2%, and the situation is more accentuated among women. There is also a higher concentration of migrant women who are heads of household; their percentage increased from 19.2 to 30.1. Finally, there are also changes in education; in general, there is an increase in the level of schooling; for example, among women, the percentage of those with undergraduate or graduate studies increased (from 9.8% to 15.4%). These data show changes in return migration in which some profiles are accentuated, as shown below when dividing the population by type of return.

Diversity in the return process

An approximation to the return process of Mexicans from the United States can be obtained from the analysis of the different types of dwellings according to the presence of migrants: individual return, national accompanied return, and binational accompanied return. ^{Canales and Meza (2018}) use these three groups to analyze the 2000 and 2010 censuses and the 2015 intercensal survey and identify, among other aspects, that return as a collective or family process is not preponderant and tends to be less frequent in recent years. In their analysis, the authors also include foreigners who are family related to a Mexican return migrant. Although they do not detail how they identify them, they classified 152 593 international immigrants, divided according to migratory modality: 107 726 return immigrants and the complement associated with other modalities. To follow up on what happened in the 2015-2020 period, an analysis of the 2010 and 2020 census data is presented below.

Identifying return migrants with census data is relatively straightforward; however, the case of return immigrants can generate some variants. ^{Canales and Meza (2018}) start from a population of 152 000 immigrants. The figure is obtained by identifying those born in the United States who in 2005 resided in that country; however, this procedure leaves out 203 000 Americans who in 2010 were between 0 and 4 years old (since they were not asked about the place of residence five years earlier). These children likely immigrated to Mexico because of the return of their parents, so in this analysis, they are considered the base population to identify those who cohabit with at least one return migrant (see Table 3).

In 2020, the results show that 253 227 dwellings were home to at least one return migrant; this implies a 60% decrease compared to 2010. The number of return migrants decreased 64%, from 824 436 to 294 203. However, the greatest decrease is observed among those born in the United States who cohabit with at least one return migrant, from 188 058 to 44 684, which implies a decrease of 76%. By type of return, individual return decreased by 54.1%, national accompanied return by 70.5%, and binational accompanied return by 81.5%. These unequal declines show an increasing concentration towards individual return, a process already warned by ^{Canales and Meza (2018}), but which was accentuated especially in the 2015-2020 period, and which allows at least suggesting that there is a greater relationship of return migration with the previous emigration of people who leave seeking employment and return after a certain time.

To deepen the previous analysis, the characteristics of the migrants that make up each type of return are presented. For this purpose, three tables are presented. The first table compares the general profile according to the return type in 2010 and 2020 (see Table 4). The second presents an approach to the differences in the profiles that may provide clues about the migration processes captured in the 2020 census, which includes the question on the reason for migration (see Table 5). The third table shows the combinations of family relationships among the returnees with respect to the head of the household (see Table 6).

The data in Table 4 show two relevant elements. The first is the heterogeneity by return type, and the second is the great similarity between what was observed in 2010 and 2020. The similarity is important because it denotes the consistency within each type of return. Regarding the differences, when focusing on the distribution by sex, it is verified that in the individual return, the majority is composed of men. This happens in both years (89.7% and 85.8%), while in the accompanied national return, their participation decreases to almost half (57.4% and 53.5%), and in the accompanied binational return, they are no longer the majority (45.7% and 45.8%). Among people born in the United States, the distribution by sex is practically equal in both years (49.2% and 49.3%), and they are mostly children of the head of household (70.6% and 71.6%), although there is also an important percentage of grandchildren (20.5% and 15.0%).

Table 5 was constructed to deepen the analysis by type of return, dividing some characteristics by type of return and sex. Although most individual returns are male, this group has great diversity. The first aspect that stands out is the age difference between men and women, since nearly half of the women are 50 years old or older (47.2%), while among men, it is a quarter (26%). This implies that in the individual return of women, nearly half are of an atypical age with respect to what is usually observed in labor flows and closer to the profile of the retired population. In addition, their marital status stands out due to the high percentage of widows or separated women (38.8%), which is more than double that of men (15.6%).

Regarding the reasons for return, 53.1% of women return to reunite with family, a higher percentage than men (42.2%). The reasons associated with other causes are in second place (16.9%), and looking for work or changing jobs (14.9% and 3.0%) in third place. In the case of men, the second most frequent reason is to look for work (21.4%), which suggests that in the United States, they lost their jobs or did not obtain the income necessary to achieve their objectives and returned to Mexico. One of the reasons for return that, due to its characteristics, is of greater concern is that associated with deportations; the census results show that this happens mainly among men, particularly in individual returns, since 17.0% returned for this reason. Among the less frequent motives, the one that stands out is education, especially among women, which is high compared to men (5.5% and 1.5%). It would be necessary to deepen if these are people who emigrated to do an exchange program and return to Mexico to continue their studies or are young people or adults who, faced with the difficulties of studying in the United States, preferred to return to Mexico.

Regarding its distribution in Mexico, individual return migration is present in all Mexican states. For this paper, they were grouped into regions only to simplify the analysis. Women return more to the Northwest (24.9%), and men return more to the Center-North (18.8%) and West (17.9%) and, to a lesser extent, to the Northwest (17.3%). When considering the reasons for return, deportees returned more to the Northwest (28.8%). This is relevant because it shows how the northern border, especially Baja California (12.5%), continues to be an important option for the population that is forced to return and that seeks to be close to their family that remains in the United States. On the other hand, return for labor reasons is more concentrated in the Center-North (26.2%), mainly in Guanajuato (13.8%).

In summary, in the individual return─the most frequent in return migration─there is a predominance of men who, due to their characteristics, highlight the profile of relatively young labor migrants. However, there are also older migrants, especially women, among whom the advanced age distinguishes their group and the high proportion of separated or widowed women. If these groups are related to the typologies analyzed, it can be affirmed that there are elements to suggest that most of the processes or types of return are observed, although the return of men and women who went to the United States for labor reasons and possibly on a temporary basis and who return to reunite with the family in Mexico would stand out.

In the national accompanied return, there is greater homogeneity between men and women. In both cases, an older age structure is identified, with a high percentage in the 50 and over group (45.1% and 47.1%), which possibly reflects the return of couples close to retirement age who return without children since they decided to stay in the United States. Nevertheless, there are also men and women of intermediate ages; a quarter is between 35 and 49 years old (24.0% and 25.4%) who probably returned together with their children. To verify the above, the configuration of family relations was identified for each dwelling (see Table 6), and the national accompanied return stands out because nearly half are single couples (51.8% and 54.6%). The return of couples with children also stands out, but to a lesser extent (11.1% and 11.4%), as well as the return of a parent and a son or daughter (12.0% and 12.4%). Thus, the return of two or more Mexicans is almost always associated with nuclear family groups, mainly couples, although there are also cases where other family members are present. As shown by the percentages in Table 6, there are no significant differences in the distributions between men and women. In terms of motives, the scenario is very similar to the individual return, but with a greater concentration of the motive for family reencounter (47.3% men and 66.0% women), and consequently a decrease in the other motives. Deportation, in this case, drops to 11.4% among men and 4.8% among women.

Lastly, the accompanied binational return is exposed. In the case of those born in Mexico, the age profile shows a younger population. Regarding the 50 and over age group, only 16.0% are in this group, and in the case of women, only 10.6%; in contrast, slightly less than half are concentrated in the 35 to 49 age group. In this type of return, deportation is more frequent than in the previous group (16.1% among men and 6.6% among women); therefore, it is very likely that part of these families returned to Mexico due to the expulsion of one of their members. Table 6 presents these family relationships and highlights that the most frequent profile is that of couples with children among both men and women (40.3% and 36.9%) or the return of one parent and one or more children (29.5% and 27.4%). For their part, those born in the United States present a young profile, especially men, since 88.0% are under 20 years of age (82.9% among women), and the majority are the head of household’s children. Thus, the profiles that most stand out in the binational accompanied return are family groups with children born in Mexico or the United States.

Although it was analyzed as a crosscutting issue among the different types of return, it is pertinent to note that according to the 2020 census, there are close to 40 000 people who resided in the United States in 2015 and returned to Mexico during the 2015-2020 period due to deportation. These are mainly men (90%), specifically men identified in the individual return (75%). These results could generate doubts, as the number is low compared to the figures of Mexicans expelled from the United States; however, it should be kept in mind that these are different situations. The figures on expulsions─or repatriations as they are labeled in Mexico’s management records─are event-based due to a person can be expelled more than once, and not all expelled persons would be considered as return migrants since some spent only days in the United States, and not all of them end up staying in Mexico. The census results are, to some extent, comparable with those obtained in the Enadid 2018. This survey follows the same logic as the census by asking about the place of residence five years earlier and the reason for migration, but the period only partially coincides with the census (2013-2018 and 2015-2020). The survey estimates 301 000 return migrants, and 18.4% of them were deported─almost five percentage units more than the census (55 000 people, 93.9% men)─; despite the differences, it can be said that the results of the census and the Enadid are consistent.

Closing remarks

The study of return migration helps to understand and expand the vision of migration between Mexico and the United States. ^{Alanís and Hernández (2022}) indicate that this is not a new phenomenon since, during the 20th century, there were periods of high flows of Mexicans and their descendants who returned to Mexico due to diverse social and economic contexts. As a concept, return migration implies challenges because, in addition to the inaccuracies regarding temporal factors to distinguish an immigrant from a temporary visitor, there is a limitation in the sources of information. Censuses in Mexico do not record the length of stay abroad, so this source approximates the phenomenon in which it is left to the respondent to define whether a stay abroad is a change of residence. The above is important when reflecting on the results of this work and explaining what happened in the 2015-2020 period.

The results show that during this period, the return decreased, and its concentration increased towards the individual return in which women and men participate, although the latter explain most of the phenomenon. The return of families or subgroups of some combination of family members decreased. This raises the doubt of why a period characterized by Trump’s rhetoric, policies, and actions against migrants was not accompanied by a rise in return migration; and why, contrary to expectations, there was a decline. Part of the explanation is that there was not as much population susceptible to return as in 2005-2010. To support this idea, the strong relation between return migration and the previous emigration of men and women for labor issues who do not plan to settle in the United States must be kept in mind. The census cannot directly demonstrate this, but several results point in that direction; for example, the high percentage of situations in which the returnee is partnered or married but his or her partner and children are not return migrants. In addition, the decrease in return migration coincides with a decrease in the emigration of Mexicans to the United States, as demonstrated by the Emif Norte’s data flow from the south and with the migration modules of the 2010 and 2020 censuses that show a decrease in emigration of 37%.

If in the years prior to the Trump administration undocumented emigration had increased, it is not risky to suggest that there would have been a higher level of return, possibly via deportation. Because, as ^{Verea (2018}) and ^{Armendares and Moreno-Brid, (2019}) point out, his presidential term was mainly characterized by actions against migrants, which undoubtedly represents a serious setback with respect to what was observed in the last four years of the Obama management. However, it should also be considered that, in addition to the specificities in migration policies of each presidential term, in the United States there is a whole machinery that has been in charge of managing migration according to its political and economic interests, and this transcends presidential terms (^{Alarcón, 2019}).

The results of the 2020 census should also call attention to the policies proposed in Mexico regarding return migration. It is relevant to note that the age structure of returnees is aging due to the lower presence of minors in the flow and the increase of older adults. As indicated, school-age minors face several barriers in their educational incorporation. However, the good news is that this immigration is decreasing, which should encourage the actors involved in the care of this population to obtain better results since the demand is lower and there is already experience from previous years. The case of the older adult population is in an opposite situation since it is acquiring a greater presence, and this is likely to continue as the population of Mexicans in the United States is aging: between 2010 and 2020, the number of Mexicans aged 50 and over increased from 2.58 to 4.00 million. Ideally, this would be a retired population that returns of their own free will; however, there is also a component of deportees who have a double disadvantage in trying to (re)incorporate in Mexico because of stigma and discrimination due to their age and being deported (^{SantiagoVargas et al., 2021}).

Finally, the analysis presented also allows making some recommendations regarding the capture of information. It is recommended to include two questions in the expanded census questionnaire. The first one should be in the international migration module to inquire about the type of document used by people who migrate to the United States. This question is in the Enadid 2018, although the possible answers should be reviewed because, on that occasion, it was indicated in option two, “Permission to work (green card)”, however, the green card is not a work permit but a permanent resident card. The second question to include is the length of stay of the last residence to analyze the temporalities according to the type of return. When proposing these questions, it should be kept in mind that the population census is a very costly exercise, and adding a question implies a large investment. Nevertheless, the general migration issue will become more important, and these questions would only be asked to a specific population group, which implies lower costs with respect to other questions.