WHY DO PEOPLE MIGRATE?

Identifying Diverse Mechanisms of Migration

THREE STORIES

"I STUDIED hydraulic engineering," Jorge tells us. He finds little use for his degree these days in Buenavista, his small village in western Mexico. He has worked many jobs in his life: collecting cotton, driving a truck, fixing cars, and supervising a farm field. He is 38 now and feels too old to change careers one more time. "She tells me I am already an oldie," he jokes while pointing to his mother-in-law, who smiles in innocent agreement. Jorge already owns a house, one that is connected to his mother-in-law's through a large courtyard. He is now planning to open an auto repair shop in his village. "I have learned many crafts, like car painting and car repair ... which is what we are going to do here," he says, "with God's help."

Mateo lives in the same village as Jorge but seems much less certain about what he wants to do in life. At age 20, he is already married with a toddler son. They all live with his father, Álvaro, who is a fisherman and the head of the Buenavista village. Álvaro talks about how Mateo's path has zigzagged until now. "He really wanted to study. ... He was just beginning high school. He sent an application to the university in Guadalajara [for the technical high school program]. He wasn't admitted. ... He doesn't have a diploma. ... He liked construction work, but that did not work out either." Álvaro tried to encourage Mateo to learn different skills. "When he was little, I got him used to [working outdoors]. Maybe he is not a working professional, but he knows a lot about the fields." These skills helped Mateo land his current job: planting pine trees in the mountains.

Teresa is from San Marco, a town about 100 miles northeast of Buenavista with remarkable industrial activity and concomitant air pollution. She is the second of seven children in her family, tightly spaced between an older brother and five younger sisters. She finished middle school, and shortly thereafter, met her husband-to-be, Tomás, who was visiting the town. "She had a crush," her father remembers, which seemed certain to remain just that until Teresa decided to go where Tomás is. She is now 40, a happily married homemaker with two children.

Jorge, Mateo, and Teresa have little in common except that they were all migrants in the United States at some point in their lives. Jorge got into a fight with his parents over whether he should work or study, and, rather than seek resolution, chose to migrate to the United States with his uncle. He was 14 years old. Mateo was searching for work in his town when a recruiter approached him with an offer for a short-term work visa to go to Oregon. He borrowed money from his father for the trip, left his wife and son to his father's care, and took off. Teresa wanted to be with Tomás, a migrant in the United States, and left with some relatives headed there at the earliest opportunity.

"EVERY PERSON IS A DIFFERENT WORLD. THEY THINK DIFFERENTLY."

This is how a former migrant responded when we asked him, "Why do people migrate?" It is a simple but profound point, one that often gets sidelined in the ongoing debates on migration: Migrants in the United States are a diverse population. We can characterize this diversity in a number of ways. We can, for one, think of different generations of migrants. Given more than a century of migrant flows, there are now multiple generations of Mexicans living in the United States: those who were born in Mexico (first generation), those whose parents were born there (second), those whose grandparents (third) or great-grandparents (fourth) were of Mexican origin, and so on. The migrant experience varies across generations. The first generation often struggles to find work, to learn the language, to adjust to a new culture, and to obtain legal status. By the second generation, many of these issues get resolved, but some continue to leave an imprint. Many children of migrants are U.S. citizens by birth, for example but still work to mold a multicultural identity or to overcome their families' socioeconomic disadvantage.

We can also think of different cohorts of migrants, that is, groups entering the United States in particular historical periods, within each generation. The migrant experience varies across these cohorts too. Consider the first-generation Mexicans in the United States. The migrant cohort of the 1960s included mostly farmworkers (many on short-term work visas under the Bracero program, a temporary labor agreement between Mexico and the United States that ended in 1964). This cohort left a poor Mexico, whose gross domestic product (GDP) hovered around US$3,400 per capita (in constant 2005 values), and arrived to a relatively welcoming United States, where migrants were in heavy demand due to the labor shortages after World War II. The migrant cohort of the 1980s, many of them undocumented workers, faced dramatically different conditions. This cohort departed from a richer — though more volatile — Mexican economy, where the GDP reached US$7,200 (in 2005 values) per capita at one point, only to drop to about US$6,300 after the financial crisis in 1982. The cohort also encountered more restrictions to entry with increasing enforcement on the border. Migrant cohorts of the 1960s and 1980s, then, varied not only in composition, but also in the contexts of origin and reception.

We can, finally, think of different groups of migrants within each cohort (or across cohorts) for whom the circumstances surrounding migration vary. I mean mostly personal circumstances here. The three stories attest to this more elusive type of variability. The reasons Jorge describes for his move to the United States — a desire to study, a fight with his parents on this topic, and an opportunity to migrate with his uncle — seem quite different from those that pushed Mateo across the border — an unfruitful job search in his community, an unexpected job offer from a recruiter for a U.S. company, and encouragement from his parents. We can rationalize both moves as paths to economic opportunities in the United States, but Teresa's story complicates this simple logic. It suggests that migration may have a relational component — one that is borne out of a couple's desire to live together in this case — as well as an economic motif.

FOREST OR THE TREES? CAPTURING THE DIFFERENT REASONS FOR MIGRATION

It is, of course, not possible to account for each and every migrant story. Nor is it desirable. I do not want to lose sight of the forest for the trees. But, I also do not want to see only the forest and miss the trees. What I seek are common patterns that characterize the experience of many migrants, and also reveal the diversity in those experiences. To identify these patterns, I ask three questions in sequence: Who migrates, when, and why?

WHO MIGRATES?

Our first question is deceivingly simple: Who are the Mexican migrants who come to the United States? As human beings, we are drawn to making generalizations to make sense of our world. These generalizations are often organized around simple social categories (e.g., men and women, migrants and natives, us versus them). We seem to be hardwired to single out certain attributes, such as gender and race, in creating social categories and reaching social judgments. When it comes to migrants in the United States, it is perhaps not surprising that we use ethnic or national origin to place migrants into categories (e.g., Asians and Hispanics, Mexicans and Central Americans). It is surprising, however, to see similar categories, and broad generalizations on each category, prevail in scholarly work.

Many academic studies tell us about the average Mexican migrant: a young married man who has few years of schooling and some family ties to migrants who preceded them, earns low income, and comes from a poor rural community with few employment opportunities. Others show how the attributes of the average migrant have changed over time. The average Mexican migrant, one study shows, had nearly eight years of schooling as of 1990, up from six years in 1970.

These general patterns are important; they keep the pulse of the migrant flow. But the patterns also hide tremendous variation within the migrant population. The average migrant today may be more educated than one three decades ago, but that migrant still has many peers who are much less, or much more, educated. How similar is the migration experience for these individuals?

To answer such questions, scholars have investigated migration behavior separately for different groups of individuals, for example, among men and women, rural and urban residents, the more and less educated, and so on. One limiting factor is that studies typically considered a single dimension of social life, such as gender, community, or socioeconomic background, to differentiate migrants and their experience. But these dimensions often work in conjunction to produce different outcomes for different groups of individuals.

Take the case of gender. A woman may face more constraints to moving in a small community, where family and friends more easily enforce traditional gender roles, than one in a large city. Or take education. A man with no education from a rural area may have less incentive to migrate than a similar man from an industrial town, where job opportunities require formal education. These examples suggest how social categories, like those based on gender or education, may differentially influence one's migration propensity in different contexts.

This idea is not new, at least not in theory. Migration scholars have long noted "the situational and relational character" of social categories, like gender, race, and class. Giving the idea perhaps its most elaborate theoretical treatment, gender scholars have similarly written on intersectionality "the relationships among multiple dimensions" of social life.

These theoretical points chart a distinct approach in empirical work: We need to focus not just on single attributes, like gender or education, but on configurations of multiple attributes to understand the heterogeneity in human experiences. But — and herein lies the main problem — how do we identify and study those configurations?

Migration scholars have deemed this task "difficult without recourse to qualitative methods." These methods allow researchers to become immersed in a setting — through in-depth interviews, participant observation, or a historical case study — and to reveal the complexities of social life there. We can see the payoff to this approach in numerous examples in migration research. It is through ethnographies, for example, that we learn how gender relations not only assign differential roles to men and women when it comes to migration, but also lead to differential access to key resources like migrant networks. We also see how gender relations vary by social class, or across rural or urban settings, and how these relations themselves get transformed through migration experience.

Similar insights emerge from quantitative data and methods as well. Scholars have used quantitative methods — analyzing large-scale social surveys, for example — to demonstrate the interactions between different personal and contextual attributes that lead to distinct migrant experiences. Massey, Goldring, and Durand, for example, show that personal attributes, such as gender, education, or family wealth, matter less for migration behavior as migration becomes more prevalent in a community. Women, the less educated, and the less wealthy — groups that are less inclined to migrate initially — become more likely to do so when they can rally support from others in their community who have already migrated.

Both qualitative and quantitative approaches, in other words, can be effective in showing how personal and contextual attributes jointly determine migration outcomes. Both approaches, however, can be equally limiting in how many attributes they allow us to consider. This is because each new attribute increases the complexity of the analysis exponentially. Consider a simple example. We want to study the heterogeneity in migration behavior across two binary attributes: sex (man or woman) and age (young or old). In combination, the two attributes yield four possible categories. It is easy to keep these categories in mind, and also to track any comparative evaluations of each category. Now, add three other binary attributes: education (none or some), wealth (poor or rich), and neighborhood (good or bad). If we cross-classify all five attributes, we get 25=32 possible categories. It is hard for us to make sense of the patterns across all 32 categories even if we were able to identify such patterns with data analysis.

There is an easy way around this problem, however. Not all possible combinations are equally prevalent in the data. This is because many attributes are highly correlated with one another. For example, wealthier individuals tend to be more educated than their poor counterparts; they also often live in better neighborhoods. In real life, in other words, individuals tend to cluster around a few distinct configurations.

Our task, then, is not to consider all possible configurations of attributes, but to discover those most prevalent in the data. Cluster analysis methods, developed in statistics and computer science, allow us to do just that: discover groups of cases with similar configurations of attributes in data. These methods allow groupings to emerge from data without imposing any initial structure. The methods are used widely in fields as diverse as biology, physics, and computer science to produce effective descriptions of typically large and complex data. The methods, however, have been less popular in social science. (Appendix A provides a brief history to illuminate possible reasons for this state of affairs.)

Many social scientists have used cluster analysis and related methods, such as latent class analysis, to characterize the variability in outcomes or in potential determinants of a particular outcome. My approach here resembles that of Andrew Abbott and Charles Ragin, two prominent sociologists who have independently applied methods similar in their philosophy: the former to categorize sequences of events, such as career changes, and the latter to identify configurations of causal conditions for historical outcomes. In recent work, political scientists, Justin Grimmer and Gary King, have underlined the use of cluster analysis as a tool for conceptualization. This is how the method is understood and used here.

I seek to identify distinct configurations of personal, family, and community attributes that characterize different migrant types in Mexico-U.S. flows. This approach is often considered to be "data-driven" or "inductive" because it employs the data to identify a categorization scheme — a model through which we can understand those data. Like models in general, I do not evaluate a categorization scheme as "true" or "false," but instead ask: Is it useful? Does it allow us to describe the migrant population parsimoniously, while recognizing its heterogeneity? Does it reveal patterns that can be validated externally? These questions shape the analytic strategy that follows.

WHEN DO PEOPLE MIGRATE?

Cluster analysis, in this study, classifies migrants into distinct groups on the basis of their personal, family, and community characteristics. The method allows me to discover different configurations of migrants, and potentially, the different logics guiding their migration decisions.

I will explain this with an example. Say I use cluster analysis and discover two migrant groups in the data: One group includes mostly young uneducated men who are farmworkers in rural areas, and the other includes young professionals with advanced degrees in cities. This dichotomy reveals different configurations among migrants, but does it tell us anything about the logic underlying their migration? To see that, I can look at the conditions under which each group proliferates. Now, say, I observe the first group when there is a drought that reduces farm production, and the second when there is an economic crisis that leads to higher unemployment among professionals. This observation suggests that two groups may be reacting to different conditions when they migrate. In this case, the specific categorization scheme not only shows the different groups among migrants, but also suggests the different reasons that motivate migration behavior of each group.