Abstract
Research on Mexican migration to the United States has long noted how the characteristics of sending communities structure individuals’ opportunities for international movement. This literature has seldom considered the concentration of indigenous residents (those with origins in pre-Hispanic populations) in migrant-sending communities. Drawing on data from 143 communities surveyed by the Mexican Migration Project, and supplemented with data from the Mexican Census, this article uses multilevel models to describe how the share of indigenous residents in a migrant-sending community relates to different aspects of the migratory process. We focus on (1) the decision to migrate to the United States, and (2) the documentation used on migrants’ first U.S. trip. We do not find that the concentration of indigenous residents in a sending community is associated with the decision to migrate to the United States. However, we do find that people in communities with relatively high indigenous populations are more likely to migrate as undocumented rather than documented migrants. We conclude that the concentration of indigenous peoples in communities likely indicates economic and social disadvantage, which limits the residents’ possibilities for international movement.
Keywords
About 11.6 million Mexicans lived in the United States in 2013; half this population was undocumented (Passel, Krogstad, and Gonzalez-Barrera 2014; Migration Policy Institute 2018). Demographers note how characteristics of individuals’ origin communities—such as their levels of economic development or varying histories of U.S. migration—structure opportunities for international migration, including the documentation status held upon entry (e.g., Massey and Espinosa 1997). However, few have considered how these characteristics overlap with the share of indigenous residents—those persons with pre-Hispanic ancestry—in a community. We argue that the context of indigeneity—the extent to which places have relatively high shares of indigenous residents—is an important consideration for understanding migratory processes (Fox 2006; Fox and Rivera-Salgado 2004). Indeed, ethnographers and historians note that the context of indigeneity represents a primary dimension of place stratification in Mexico in that it is strongly associated with the uneven distribution of economic and social resources (Batalla 1996; de la Peña 2006; Novo 2006; Zapata 2000). This article takes a demographic approach, examining how this axis of inequality may influence Mexico-U.S. migration.
We investigate how having origins in an indigenous place relates to different aspects of the migratory process. Specifically, we use data from 143 communities surveyed by the Mexican Migration Project (MMP), supplemented with additional information on respondents’ larger economic and social contexts, to examine if the context of indigeneity is associated with (1) the decision to take a first trip to the United States, and (2) the documentation status of migrants on their first U.S. trip. Given ethnographic and historical research suggesting that indigenous density is associated with an uneven distribution of economic and social resources across communities, we expect migrants from indigenous places to be less likely than their counterparts in nonindigenous places to migrate to the United States, and, among those who migrate, to be more likely than their counterparts in nonindigenous places to do so without documentation.
We find that the context of indigeneity is associated with divergent pathways of movement to the United States, but not always in straightforward ways. Although we do not find that the context of indigeneity is related to the likelihood of first migration to the United States, we do find that people who migrate from high-indigenous places are more likely to do so in undocumented than in documented status. We conclude that the share of indigenous residents in a community is related to the reproduction of social inequality in Mexico-U.S. migration: places with relatively high indigenous populations are likely to be sites of concentrated economic and social disadvantage, and they limit residents’ options for international movement.
Mexico-U.S. Migration: Economics, Networks, Policy, and the Context of Indigeneity
Here, we review the dominant micro- (i.e., individual or household) and macro- (i.e., ecological) level explanations for international migration, specifying how economic, social, and policy considerations may contribute not only to individuals’ U.S. migration decisions but also to their documentation status on the first U.S. trip. We then describe how the context of indigeneity is associated with the various determinants of migration and suggest the importance of studying the context of indigeneity as an axis of stratification in Mexico-U.S. migration flows.
Micro- and macro-level economic considerations
Neoclassical and new economics theories of migration describe how micro-level factors such as individual income and household wealth affect the migratory process. In neoclassical economics, individuals migrate when they expect to earn higher wages in a destination country (Harris and Todaro 1970; Sjaastad 1962). In the new economics perspective, households in developing economies use migration as part of a strategy to diversify risks to income (Stark and Bloom 1985). In addition to mobilizing U.S.-bound flows, these micro-level factors may limit access to the documentation required for lawful migration because applications for visas and other migration papers require the payment of fees and proof of financial resources that many individuals and households lack. As a result, not all potential Mexican migrants can afford to secure documents, particularly given the visa backlogs and long wait times that have come to characterize the process since the enactment of per-country visa quotas after 1965 (Massey and Espinosa 1997, 949). 1 These individuals instead turn to undocumented entry as the only means of gaining access to U.S. jobs, often with the help of border smugglers known colloquially as coyotes (Donato et al. 2008). Overall, studies suggest that persons lower on the socioeconomic ladder are both more likely to migrate, and more likely to do so without documents, compared to those higher in the status hierarchy.
The neoclassical and new economics traditions also imply that macro-level factors—labor supply and demand as well as local distributions of income and wealth—are associated with the migratory process (see Massey, Durand, and Malone 2002, 13). In more economically developed communities, individuals have access both to institutions useful for mobility, such as schools, and to industrial and service jobs that employ high-skilled workers (Durand et al. 1996; Massey 1988; Massey, Durand, and Malone 2002). In communities typified by low-wage agricultural work, however, access to jobs and resources is relatively lacking (Taylor and Yunez-Naude 2000). Given limited access to stable, high-paying jobs outside of agriculture, residents of these communities are vulnerable to the fluctuations of a volatile economy. As a result, residents from less-developed communities are more likely to migrate—and to do so without documents—than those from more-developed areas.
Micro- and macro-level social considerations
Cumulative causation theory considers how network connections contribute to the migratory process (Massey et al. 1993, 449). It outlines how migration attains a self-feeding character over time, whereby each migration act expands the range of ties that connect nonmigrants in an origin community to migrants in a destination country, with this expansion of ties serving to mobilize additional migrants (see Garip and Asad 2016). The theory suggests that direct connections to current or former U.S. migrants—particularly family members—facilitate international movement. Given the family-reunification entitlements of U.S. immigration law (Jasso and Rosenzweig 1986), the theory also implies that those with family connections are more likely to depart with documentation. Holding constant socioeconomic status, we therefore expect that persons lacking family members with documented migratory experience are not only less likely to migrate but also less likely to migrate in a documented status compared to those with family ties to documented family members.
Cumulative causation theory also identifies a number of macro-level pathways associated with the migratory process. As more individuals from an origin community migrate, the community is likely to experience distributional transformations that alter its economic structure (Fussell and Massey 2004, 154). Increasingly visible signs of wealth—in the form of land purchases or home improvements—may stimulate additional out-migration (Garip and Asad 2015, 2016; Stark and Taylor 1991). In addition, as the prevalence of migration increases within communities, international migration may become a normative expectation (Kandel and Massey 2002). The concentration of community members who can transmit knowledge and information about how to legalize may, in turn, contribute to a higher likelihood of taking a documented trip, even among those lacking direct family ties (Corona and Romo 2008; Hagan 1994; Garip and Asad 2016). Overall, then, migration and especially documented migration are hypothesized to be more likely in places with extensive migration histories; individuals from places with limited migration histories are less likely to migrate overall but more likely to depart without documents when they do choose to head north.
Policy and timing considerations
Whatever economic and social circumstances may prevail in households and communities, at any point in time the likelihood of migration is also conditioned by policy (Massey, Goldring, and Durand 1994). A temporary labor arrangement known as the Bracero Program brought millions of Mexicans to the United States for farm labor on lawful short-term visas from 1942 to 1964 (Cornelius 1986). Ongoing demand for labor after 1965 (Massey and Pren 2012)—coupled with two peso devaluations in Mexico, as well as the United States’ implementation of visa restrictions—mobilized additional flows from Mexico while ensuring that they would enter the country without legal authorization (Jasso and Rosenzweig 1990; Massey and Espinosa 1997). About 80 percent of the 5.7 million Mexicans who entered the United States between 1965 and 1986 were undocumented (Bean and Stevens 2003; Durand and Massey 2003; Massey, Durand, and Malone 2002).
The Immigration Reform and Control Act (IRCA) heightened border security and granted amnesty to 2.3 million undocumented Mexicans already in the United States. Amnesty recipients’ relatives then migrated to the United States themselves, both with and without documents (Massey and Espinosa 1997, 960; see also Massey, Durand, and Pren 2016). Additional attempts to bolster immigration enforcement in 1990 and 1996 followed, but they had limited success in stemming undocumented Mexico-U.S. migration (Donato and Armenta 2011).
The context of indigeneity
Both theory and research explain how the availability of economic and social resources at the micro and macro levels affect the migratory process. But they seldom grapple with how the context of indigeneity, and its association with the uneven distribution of resources across communities in Mexico (Batalla 1996; de la Peña 2006; Flores and Telles 2012; Novo 2006; Villarreal 2014; Zapata 2000), relates to migrant decision-making (but see Asad and Hwang 2018). Mexico is home to some sixty-two recognized indigenous groups that make up the largest indigenous population in the Western hemisphere (Hamilton 2011; Layton and Patrinos 2006). Indigeneity is a feature of place, with ethnographic and historical accounts describing how residence in an indigenous community—regardless of indigenous language proficiency or ethnic identification—constitutes a powerful axis of inequality within Mexico (Batalla 1996, 22; Novo 2006, 7; see also Friedlander 2006). 2 Even when individuals speak an indigenous language, or self-identify as indigenous in one setting, they can shed their community membership and achieve fluency in Spanish to gain access to coveted economic and social resources in other communities (Nagengast and Kearney 1990). A community’s relatively high indigenous density thus represents a potentially critical but overlooked factor in explicating the process of Mexico-U.S. migration (Fox 2006).
The limited attention to the context of indigeneity in demographic studies of Mexico-U.S. migration is surprising in light of the relatively large ethnographic and historical literature on Mexico’s indigenous populations. Although a few studies describe early waves of migrants out of indigenous areas that participated in the Bracero Program—including P’urépechas from Michoacán (Durand 1994), Mixtecs and Zapotecs from Oaxaca (Nagengast and Kearney 1990; Stephen 2007), and Nahuas from Tlaxcaltec (Raúl and Romo 2008)—the vast majority of indigenous communities remain unstudied. Once the Bracero Program ended in 1964, the already-mobilized indigenous communities continued to send migrants to the United States for farm work, motivated by endemic poverty and successive Mexican economic crises. IRCA’s passage in 1986 allowed the relatively few migrants from indigenous communities with a history of U.S. migration during the Bracero Program the opportunity to legalize (Durand and Massey 2003a, 89), but most Mexicans from indigenous communities remained untouched by migration. Aligned with the expectations of the cumulative causation theory of migration, we therefore expect the context of indigeneity to be negatively associated with the likelihood of migration and, among migrants, to be positively associated with the likelihood of undocumented migration.
The outflow of workers from the largely mestizo communities in west-central Mexico created labor shortages in the historical heartland for U.S. migration. Indigenous communities often supplied workers to meet this demand, using internal migration within Mexico as a strategy for socioeconomic mobility until the 1980s. Some left subsistence farming in their home communities to pursue seasonal work on large, export-oriented farms in northern Mexico (Hamilton 2011; Psacharopoulos and Patrinos 1994), while others moved from the countryside to urban hubs in Mexico to work in factories (Nolasco 1995, 125). Following economic crisis in Mexico during the 1980s, members of these itinerant indigenous communities looked to join still-mobilized nonindigenous Mexican communities in migrating to the United States for new employment opportunities.
Although some groups, such as the Mixtecs (FitzGerald, Hernández Díaz, and Keyes 2013), had crossed into the United States before IRCA, most did so after IRCA’s passage when access to documentation was limited, security along the Mexico-U.S. border was fortified, and the economic outlook for indigenous communities in Mexico grew bleak in the wake of trade-induced changes. Thus, we expect U.S.-bound migration from high-indigenous communities to be less likely compared to communities with few or no indigenous residents. Among persons who do migrate, however, we expect those from high-indigenous communities to display a higher likelihood of migrating in an undocumented rather than documented status.
At around the same time that U.S.-bound flows from indigenous communities began to mobilize, Mexico entered a period of agricultural and industrial development that exacerbated economic disparities between the country’s indigenous and nonindigenous areas (Gordon 1997, 422). Residents of nonindigenous places benefitted from infrastructural improvements that facilitated access to human and financial capital, but these benefits did not extend to residents of indigenous areas (Hamilton 2011, 131). A constitutional amendment in 1992 permitting the privatization of landholdings upended the communal system of farming (known as ejidos) on which the livelihoods of many indigenous communities depended (Jung 2003).
The North American Free Trade Agreement (NAFTA), which sought to eliminate barriers to trade between the United States, Mexico, and Canada, followed in 1994 against the strong opposition of indigenous communities (Fernández-Kelly and Massey 2007; Kelly 2001). The Zapatista rebellion in Chiapas exemplifies this opposition in many ways (see Harvey 1998; Zapata 2000). In the end, NAFTA not only exacerbated wealth disparities between indigenous and nonindigenous places but also increased rates of poverty within indigenous communities (Hamilton 2011, 131; Fernández-Kelly and Massey 2007; Kelly 2001; Meza 2006). These transformations prompted new waves of internal migration throughout the 1990s as members of indigenous communities searched for viable sources of income.
Contemporary U.S.-bound migration from indigenous areas occurred against this economic backdrop (Holmes 2013). The few indigenous communities whose histories of U.S.-bound migration allowed them to legalize under IRCA, depicted in ethnographies of Mixtecs (Kearney 1996/2018; Novo 2006; Smith 2003; Stephen 2007) and the Nahuas (see Raúl and Romo 2008), could rely on networks to secure documents for family members adversely affected by NAFTA. Indigenous communities without such extensive histories, depicted in ethnographic or census-based analyses of the Hñähñú from Hidalgo (Quezada Ramírez 2008; Schmidt 2012; Schmidt and Crummett 2004), could only journey to the United States in undocumented status. With little recourse to legalization and an increasingly securitized Mexico-U.S. border, pioneer migrants from these nascent sending communities served as conduits for ongoing undocumented flows. Anthropological and historical accounts thus point to the late 1980s and early 1990s as a new epoch of U.S.-bound migration from indigenous areas.
Methodology
Data
In this analysis, we merge surveys of migrants and nonmigrants from the MMP 3 with data on municipalities of residence from the Mexican Census compiled by IPUMS-International (Ruggles 2013). The MMP survey records information on respondents’ demographic and socioeconomic characteristics and, among migrants, the timing and legality of their first and last U.S. trips. The dataset includes 122,836 individuals, 15.9 percent of whom had been to the United States by the time of the survey. Altogether, they lived in 22,374 households located in 143 communities throughout Mexico. Around 200 households were randomly sampled in each community, and, within each household, the household head reported on the U.S. migration history of each household member. The communities in MMP143 were surveyed during winter months between 1982 and 2013 (when migrants traditionally return for the holidays) and cover 24 of Mexico’s 32 states.
Comparisons of the MMP data with nationally representative surveys show that the former provides an accurate profile of U.S. migrants who return to Mexico (Durand, Massey, and Zenteno 2001; Zenteno and Massey 1999). But the MMP is not representative of all Mexican communities, including indigenous ones. The survey initially focused on Mexico’s historical migrant-sending regions, particularly those in the Central-West where indigenous communities are sparse relative to the central and southern parts of the country. Though the MMP has since expanded to survey communities with larger shares of indigenous residents, these communities remain underrepresented. Whereas communities in which 10.0 percent of all residents speak an indigenous language constitute one-third of all municipalities in Mexico, they represent only 10.0 percent of those included in the MMP sample. Although speakers of indigenous languages are contracted for fieldwork for the MMP survey in indigenous areas, the MMP survey questionnaire is written and usually administered in Spanish, thereby excluding indigenous-language respondents and communities, which are generally among the most disadvantaged in Mexico (CONEVAL 2011; INEGI 2009). However, most indigenous speakers also speak Spanish. According to the 2010 Mexican Census, only about 7 percent of Mexico’s population report speaking only an indigenous language. We thus expect the MMP data to provide a conservative test of the link between the context of indigeneity and U.S. migration.
Municipalities in Mexico are roughly equivalent to a U.S. county and constitute the smallest geographic units of data available from the Mexican Census. In this study, we combined data on respondents surveyed by the MMP with information on their municipalities of residence derived from the census. Some MMP communities are villages or towns in small municipalities, whereas others are neighborhoods in mid-sized and large urban municipalities; in some cases, two MMP communities may lie in the same municipality. Unfortunately, we cannot identify whether MMP respondents lived in the survey community when they first migrated to the United States and therefore assume that characteristics of the survey place proxy the context from which the migration occurred. About 14.0 percent of respondents reported taking a domestic trip at some point prior to the survey, and 19.5 percent were surveyed in a municipality different from that of their birth. Since variation within municipalities along these contextual indicators is likely, we further assume our estimates represent a lower bound for the variance of place effects.
Measures
We examine the relationship between the context of indigeneity with two key features of the migratory process. Our first dependent variable is whether a respondent chose to undertake a trip to the United States by the time of the MMP survey. Respondents who ever took a U.S. trip are coded as “1,” with nonmigrants coded as “0.” Our second dependent variable is whether respondents who chose to migrate did so without documents, with undocumented migrants coded as “1” and those who entered with one of the several documented statuses captured by the MMP coded as “0.” Our focus on first trips to the United States limits the analytic complications that arise as the characteristics of people and places change as a result of migration itself. Although research suggests that gradations of documentation (e.g., permanent versus temporary status) matter for patterns of migration and immigrant incorporation (e.g., Asad 2017; Donato and Armenta 2011; Menjívar 2006; National Academies of Sciences 2015), we lack the sample size to consider separately the various categories of documentation contained in our data.
Our primary independent variable is a categorical indicator of a community’s municipal-level indigeneity calculated from Mexican Census data. We classify communities in municipalities where at least 10.0 percent of residents reported speaking an indigenous language as “high-indigenous,” those communities in municipalities where fewer than 1.0 percent of residents speak an indigenous language as “low-indigenous,” and those with shares in between as “moderate-indigenous.” 4 Our classification mirrors that used by Mexico’s National Population Council and National Commission for the Development of Indigenous Pueblos (CONEVAL 2012, 39). About 70.0 percent of the MMP communities are in low-indigenous municipalities; approximately 9.0 percent are in high-indigenous municipalities, with the remaining 21.0 percent falling into the moderate-indigenous category.
Data on whether residents speak indigenous languages are available from the Mexican Census beginning only in 1990. In contrast, first U.S. trips recorded in the MMP range from 1906 to 2012 and average 1975; and nearly two-thirds of the first-time migrants in our sample left before 1990. Assessing the context of indigeneity only in 1990 thus minimizes the possibility that a municipality’s history of U.S. migration affects its indigenous language-speaking density. The degree of indigeneity decreased by about 2.0 percent, on average, between 1990 and 2010, and the change was greater in more indigenous places. We thus expect our 1990 indigeneity measure to be conservative.
We first test, separately, if the context of indigeneity is related to migration and documentation status. Then, in a sequence of models, we adjust for community characteristics particular to the MMP, as well as individual-, household-, and municipal-level characteristics that tend to be correlated with the context of indigeneity based on the literature reviewed. These variables are observed at the time of the survey unless otherwise noted. Our models adjust for the survey year to account for differences in how the MMP selected its communities over time (i.e., beginning mainly in nonindigenous communities), as well as the type of community (i.e., metropolitan area, smaller urban area, town, or rancho, based on population sizes determined by the MMP).
Using information on all respondents included in the MMP data, we constructed a measure for the average number of rooms in the household’s dwelling and the average years of education in each community to compare the socioeconomic status of individuals within the community to the community average. We also include a dummy indicator for whether respondents lived in a state on the Mexico-U.S. border at the time of the MMP survey to account for the possibility that these individuals were recently deported and had not yet returned to their home communities, may be preparing to migrate, or may migrate more frequently than residents of nonborder states (Durand and Massey 2003a, 78–9). No MMP community in a high-indigenous municipality lies in a border state.
We also control for individuals’ demographic characteristics, including birth year, sex, household position, marital status, and household size; as well as socioeconomic characteristics, such as education, employment status, and occupational sector. 5 We measure household-level socioeconomic status with characteristics of respondents’ dwellings: the number of rooms in each individual’s home and whether the household has finished flooring (Massey et al. 1990).
To capture the circumstances of migration, we include dummy variables for whether individuals ever migrated within Mexico and whether they moved away from their municipality of birth, as internal migration may serve as a “springboard” to international migration (Durand 1994). We control for the year of first U.S. trip to account for potential differences in the policy context at the time of migration.
To control for household-level migration dynamics—specifically, that migrants from households with longer migration histories are more likely to enter the United States and to do so with documents obtained through migrant family members—we include a dummy variable for whether another household member ever made a documented trip. In models predicting undocumented migration, we further control for the legal context of migration by including a measure of the probability of apprehension at the border during the time of the trip and a measure of visa accessibility in the year of migration based on the number of Mexicans who received a green card relative to all Mexican migrants observed (variables constructed by the MMP and available from the project website).
Finally, we incorporate indicators for respondents’ communities and municipalities. We measure a community’s macro-level economic development by using the share of residents in its municipality in the survey year that earns less than the minimum wage. This variable is highly correlated with multiple measures of economic development, including municipal-level rates of self-employment, agricultural work, low education, and a lack of basic infrastructure. To capture domestic migration experience in the community, we control for the share of adults who reported taking an internal trip by the year of the survey. In models predicting undocumented migration, we account for community migration histories by including a control variable for the share of a community’s adults with U.S. migration experience in the year of U.S. migration, calculated by the MMP. We do not include this variable in models predicting migration relative to nonmigration since it is a summary measure of the outcome variable.
Table 1 compares means of the above measures for all respondents in communities in high- (8.0 percent of the sample) and low- (75.0 percent of the sample) indigenous municipalities. Respondents in these municipalities are statistically distinguishable on almost every indicator. Relative to respondents surveyed in communities in low-indigenous municipalities, those in high-indigenous municipalities are more likely to be male, to be younger, to be the spouse of the household head, to be more highly educated, 6 and to be employed. They live in smaller homes with unfinished flooring, are more likely to live in their birth community, and are less likely to have taken a domestic trip. Respondents in high-indigenous municipalities have shorter histories of U.S. migration in their households, are less likely to have a member of the household who is documented, and are less likely to live in a border state.
Descriptive Statistics for All Variables Used in Multilevel Models
Notes: Standard deviations in parentheses; p-values test statistical significance of differences between respondents in communities in high- and-low indigenous municipalities. Measures of “Larger Community Context,” except indigeneity, observed in survey year. Indigenous levels are based on the share of residents who report speaking an indigenous language in 1990 (Source: Mexican Census compiled by IPUMS-International).
p < .001, **p < .01, *p < .05, †p < .1.
Migrants compose 15.9 percent of our data (15.9 percent of the entire sample). A lower share of individuals in high-indigenous municipalities (13.4 percent) are migrants compared to low-indigenous municipalities (17.1 percent). Compared with migrants in low-indigenous places, migrants in places with a relatively high indigenous presence are more likely to have migrated more recently, to have migrated without documents, to have migrated during periods when visas were less available, and to have departed when the probability of apprehension was slightly lower.
When we examine average community- and municipal-level characteristics, there are large differences between indigenous and nonindigenous places. On average, indigenous municipalities were surveyed later and were less likely to be located in metropolitan areas relative to small urban areas or villages. Indigenous communities also had smaller homes and lower rates of education and were less economically developed with shorter histories of internal migration. Thus, the context of indigeneity is highly correlated with many characteristics that are negatively associated with the likelihood of migration to the United States and, among migrants, making a trip with legal documentation.
Analytic strategy
Our goal is to examine the association between a sending community’s context of indigeneity and their residents’ decision to initiate migration to the United States. To accomplish this aim, we estimated a series of multilevel models to account for the hierarchical structure of the MMP data, where individuals and households are nested within survey communities. To estimate the odds of taking a trip to the United States, we fit a three-level logistic regression model, with individuals constituting the first level, households the second, and communities the third. Our outcome of interest is whether an individual took a U.S. trip by the time of the MMP survey. The primary independent variable is the categorical indicator of a community’s municipal-level indigeneity, with communities in low-indigenous municipalities omitted as the reference category.
To estimate whether individuals sought to enter the United States as undocumented migrants, we consider only migrants and fit a two-level logistic regression with individuals at the first level and communities at the second. 7 We cannot include a separate level for households here because many households contain only one or two migrants, making for sparse data structures that complicate estimation (Bell, Ferron, and Kromrey 2008). Our outcome of interest is whether a migrant made their first U.S. trip in undocumented status and our principal independent variable is the same as that defined above.
In both analyses, we first test our primary hypothesis that the context of indigeneity is associated with the migratory process (model 1), assessing the degree to which communities in high-indigenous municipalities are associated with the odds of migration and the odds of migrating without documents. In subsequent models, we add community characteristics at the time of the survey, including its metropolitan category, fieldwork year, whether it is in a state on the Mexico-U.S. border, as well as each community’s average household size and education level. Then we add individual- and household-level demographic characteristics; individual- and household-level socioeconomic indicators; and national migration characteristics. Last, we evaluate if respondents’ larger economic and social contexts explain any observed association between the context of indigeneity and the migratory process. For brevity, and for both analyses, we simply present the model without control variables and the fully adjusted model. Results from intervening models are available upon request.
We centered all indicators around their mean values for the analytic sample so that the level-1 intercept in the models predicting first U.S. migration represent the mean log odds of migrating to the United States for an individual who has average characteristics across all independent variables. Likewise, in the models predicting documentation, the intercept represents the mean log odds of undocumented migration for individual migrants who have average characteristics across all variables in the model. We drop all respondents missing data for any variables used in the analyses, removing about 5.2 percent of the sample. An examination revealed that missing values were not systematically related to either our primary dependent or independent variables.
Results
Resources, migration, and documentation
Before estimating our multilevel models, Table 2 presents average characteristics for MMP communities located in low- and high-indigenous municipalities. This contrast supports the view that the context of indigeneity is associated with access to fewer of the economic and social resources that commonly predict migration to the United States. Relative to MMP communities in low-indigenous municipalities, those in high-indigenous municipalities are less likely to be found in metropolitan areas and more likely to have residents who work in agriculture and who are self-employed. High-indigenous municipalities have a larger share of residents living in poverty and who lack finished flooring and plumbing in their homes. They also lack access to the social resources that predict U.S.-bound migration, containing fewer residents who have ever migrated either within Mexico or abroad, shorter histories of migration, and larger shares migrating without documentation.
Average Characteristics for Communities in Low- and High-indigenous Municipalities in the Mexican Migration Project Data
Note: Two-tailed t-tests indicate different from communities in high-indigenous municipalities. Indigenous levels are based on the share of residents who report speaking an indigenous language in 1990 (Source: Mexican Census compiled by IPUMS-International). All other variables except average household size and years of education—which are authors’ calculations of the MMP data—are from the MMP “COMMUN” file.
p < .001, **p < .01, *p < .05, †p < .1.
Table 3 displays odds ratios and 95 percent confidence intervals estimated using robust standard errors for the first set of models predicting migration to the United States. The odds ratios for level-1 variables represent the mean expected change in the odds of U.S. migration relative to nonmigration associated with a one-unit change for the variable in question, holding other characteristics constant and accounting for the hierarchical structure of the data. The level-2 and level-3 odds ratios indicate the expected average change for the average migrant and average household in the odds of migrating to the United States associated with a one-unit change in the variable of interest. Odds ratios greater than one indicate an increase in the odds of migrating compared to not migrating, whereas odds ratios less than one indicate a decrease in the odds of migrating relative to not migrating.
Odds Ratios from Three-level Logistic Regression Models Predicting Mexicans’ First Migration (Relative to Non-migration) to the United States
Notes: All variables are grand-mean centered. All indicators observed at survey year unless otherwise noted. Confidence intervals are calculated using robust standard errors.
p < .001, **p < .01, *p < .05, †p < .1.
The unconditional model, which has no independent variables, has an intraclass correlation coefficient of .194, indicating that around 19.4 percent of the variation in the likelihood of migrating by the time of the survey is due to between- rather than within-community differences. Odds ratios estimated from the fully adjusted, three-level logistic model predicting first U.S. trips do not support the hypothesis that the context of indigeneity is associated with the decision to migrate. In models with or without controls, the odds of initiating U.S. migration are statistically the same in communities in low- and high-indigenous municipalities. Instead, we find that other factors described in prior research better explain variation in individuals’ decisions to migrate to the United States. As shown in Table 3, individuals from rural areas, small towns or villages, and mid-sized cities have higher odds of migrating compared with individuals from metropolitan areas. Individuals from more-recently surveyed communities have slightly higher odds of migrating compared with those from communities surveyed in earlier years, although this relationship is not significant at conventional levels (p < .10).
Demographic characteristics are also associated with the U.S. migration decision. Males, older people, household heads and spouses, and married individuals all have higher odds of initiating U.S. migration than do females, younger people, nonhousehold heads or spouses, and unmarried individuals. When we consider individual and household socioeconomic characteristics, we find support for both the neoclassical and new economics views of migration. Higher education levels, better housing quality, and employment in manufacturing relative to agriculture are positively associated with migration. In contrast, persons working in the professional sector, who are not in the labor force, and who are unemployed are less likely to migrate than those working in agriculture.
The results also support the cumulative causation theory of migration in that an individual’s access to current or former U.S. migrants significantly increases the odds of an individual’s own first migration. Respondents surveyed outside their birth municipality have higher odds of U.S. migration relative to those surveyed at their birthplace; but respondents who have ever taken a domestic trip in Mexico evince lower odds of U.S. migration compared with those who have not taken a domestic trip. We do not find that economic development is associated with U.S. migration in our sample, perhaps because there is substantial variation in migration rates in areas with both low and high levels of economic development. Residents may not migrate if development in their community brings economic opportunity, but they may also not migrate if they lack access to resources useful for doing so (see Fussell and Massey 2004).
Although some of the factors that predict U.S. migration relative to nonmigration overlap with communities in high-indigenous municipalities, others do not. For example, indicators positively associated with U.S. migration—being male and having completed only primary or secondary schooling—are more prevalent in the communities in high-indigenous municipalities in our data, and most communities in high-indigenous municipalities are small urban areas or towns. Indicators negatively associated with U.S. migration—being younger and not being the household head or spouse—also characterize high- rather than low-indigenous municipalities in the MMP data. Other indicators that predict U.S. migration—being married or college educated, living outside one’s birth municipality, having at least one member of the household who is documented, coming from a larger household, living in a home with finished flooring, and residence in a rancho or metropolitan area—are overrepresented in communities in low-indigenous municipalities. Some factors characteristic of low-indigenous municipalities—domestic migration at the individual and community levels—are negatively associated with U.S. migration.
Table 4 presents odds ratios from logistic regression models estimated to predict whether migrants took an undocumented versus a documented first U.S. trip. The unconditional model with no independent variables has an ICC of .275, indicating that approximately 27.5 percent of the variation in undocumented migration by the time of the survey is due to between- rather than within-community differences. Here, we find evidence that the context of indigeneity is positively associated with undocumented relative to documented migration on the first U.S. trip. In the model including only the indigenous context indicators, migrants from high- rather than low-indigenous municipalities have 2.29 times the odds of migrating undocumented relative to migrating documented (p < .001).
Odds Ratios from Two-level Logistic Regression Models Predicting Mexican Migrants’ Undocumented (Relative to Documented) Status on the First Trip to the United States
Notes: All variables are grand-mean centered. Confidence intervals are calculated using robust standard errors. Indicators observed at survey year unless otherwise indicated. Indigenous levels are based on the share of residents who report speaking an indigenous language in 1990 (Source: Mexican Census compiled by IPUMS-International).
p < .001, **p < .01, *p < .05, †p < .1.
In the fully adjusted model, however, the odds ratio for high- compared with low-indigenous municipalities drops to 1.90, which is not significant at conventional levels (p < .10). Nonetheless, the coefficient for communities in municipalities of moderate indigeneity becomes statistically significant with the addition of independent variables (p < .05); it is not significantly different from the coefficient for communities in high-indigenous municipalities, suggesting that the relatively small number of communities in the high-indigenous category yields insufficient power to detect a significant effect. Of course, other factors correlated with the context of indigeneity may also explain some of the observed relationship between indigenous places and migrants’ documentation.
As in the models predicting U.S. migration, we find that other factors explain variation in undocumented relative to documented migration. In the fully adjusted model, migrants surveyed outside of metropolitan areas have higher odds of leaving as undocumented migrants compared to those in metropolitan areas. However, neither survey year nor community socioeconomic characteristics significantly predict the likelihood of undocumented migration. Instead, it is the addition of individual demographic, socioeconomic, and migration characteristics that appears to mediate the relationship between high-indigenous municipalities and documentation status. As can be seen, males and younger migrants, as well as those from larger households, display higher odds of undocumented migration. In contrast, migrants with higher levels of education evince lower odds of migrating without documents.
We also find that migrants who live outside their birth municipality and those who migrated during a time when the probability of apprehension was higher have higher odds of undocumented migration. As one would expect, those who migrated during a time of greater visa availability have much lower odds of undocumented migration. Consistent with the cumulative causation theory of migration, we further find that migrants in households where another member is documented have lower odds of undocumented migration. However, we do not find a relationship between undocumented migration and the share of the municipal population earning less than the minimum wage, the share of adults with U.S. migration experience in the community, or the share of adults with domestic migration experience in the community in our models. Indeed, documentation status, at least on migrants’ first trip to the United States, varies evenly across communities with high and low levels of these characteristics.
Discussion and Conclusion
Demographers describe the context of indigeneity as a primary dimension of place stratification in Mexico that is strongly associated with the uneven distribution of economic and social resources across communities. Ethnographic and historical accounts of indigenous communities suggest that this axis of inequality may be related to the prevalence of migration to the United States, but demographers have seldom examined this possibility in the research done to date. Using survey and contextual data from the MMP, supplemented with additional information from the Mexican Census, we find that the context of indigeneity is related to divergent pathways of movement to the United States.
Specifically, our results suggest that many of the economic and social characteristics of origin communities known to be predictive of U.S. migration (e.g., social class, economic development, and network connections) are lacking in areas of relatively high indigeneity. Nonetheless, we did not find a significant association between the context of indigeneity and first U.S. migration. While some factors that are negatively associated with migration—such as nonmetropolitan areas and homes with unfinished flooring—may concentrate in high-indigenous municipalities, many other factors that predict migration overlap with low-indigenous communities—such as being married, having a college education, living outside one’s birth municipality, having a larger household size, being related to a household member who is documented, and living in a rancho or metropolitan area.
Models predicting documentation status on migrants’ first U.S. trip nevertheless suggest that the unequal concentration of economic and social resources in indigenous versus nonindigenous places still matters for migratory outcomes. We find that the context of indigeneity is associated with the likelihood of migrating undocumented as opposed to documented on the first trip for U.S.-bound migrants from communities in moderate-indigenous municipalities. Although the coefficient for those migrants living in high-indigenous municipalities is not statistically significant at conventional levels, its magnitude is similar to the coefficient for moderate-indigenous municipalities and is significant at the p < .10-level. Altogether, we take these results to suggest that U.S.-bound migrants from communities in municipalities characterized by greater indigeneity are more likely to take an undocumented rather than documented first trip when compared with their peers in low-indigenous municipalities.
These findings likely do not apply to all of Mexico’s diverse indigenous communities, particularly those in west-central Mexico as opposed to central or southern Mexico. Some indigenous communities in west-central Mexico—such as the P’urépechas in Michoacán (Durand 1994)—maintain longstanding ties to the United States, while others in central and southeastern Mexico—such as the Hñähñú from Hidalgo (Quezada Ramírez 2008; Schmidt 2012; Schmidt and Crummett 2004)—have recently begun to migrate to the United States more intensively. Nonetheless, our results suggest that a bundle of factors may underlie between-community differences in migrants’ documentation status in the central and southeastern regions of Mexico. High-indigeneity in those regions is associated with limited economic development and a low prevalence of U.S. migrants, and together these circumstances appear to limit access to documentation.
Our analyses should be taken as a first step in uncovering the role the context of indigeneity plays in shaping aggregate patterns of Mexico-U.S. migration. Our results need to be extended to consider how the context of indigeneity and its resources relate to migratory decisions taken later on in the migratory process after the initial U.S. trip (see, e.g., Asad and Hwang 2018). Research focusing on migrants’ last trip to the United States may illuminate whether and how individuals accumulate the financial and social resources useful for not only migration but also for documentation (Orrenius and Zavodny 2005, 220; see Aptekar 2015; Massey, Durand, and Malone 2002). Future survey research should also make a greater effort to collect information on individual- and community-level measures of indigeneity to assess its effect on migratory outcomes across levels of analysis. This sort of large-scale research should be supplemented with additional micro-level studies of indigenous communities migrating to the United States to identify what mechanisms underlie the association between the context of indigeneity and migratory outcomes.
Overall, our examination of the context of indigeneity and U.S. migration reveals at least two pathways through which indigenous places are associated with inequality in Mexico-U.S. migration flows. First, relative to nonindigenous places, high-indigenous municipalities are more likely to be sites of multiple forms of economic and social disadvantage. Second, related to this point, residents of indigenous areas have limited access to resources useful for international movement with documentation. These results suggest that demographers should pay greater attention to the different ways the context of indigeneity and its resources affect the migratory process.
Footnotes
Notes
Asad L. Asad was a postdoctoral fellow at the Center for the Study of Inequality at Cornell University before joining the faculty of Stanford University as an assistant professor of sociology in 2019. His work focuses on social stratification, migration and immigrant incorporation, and race/ethnicity.
Jackelyn Hwang is an assistant professor in the Department of Sociology at Stanford University. Her main research interests are in the fields of urban sociology, race and ethnicity, immigration, and inequality.