Moving for Love: Interracial Marriage and Migration in Brazil

Abstract

The link between marriage and migration is usually considered in terms of international migration. However, domestic migration provides another lens in which to view this connection. In Brazil, despite the large migration from the northeast to the southern regions in the twentieth century, the role of domestic migration in race-mixing has been unacknowledged. Since race in Brazil is highly regionalized, with black and brown Brazilians comprising most northern regions and white Brazilians being in the majority of the southern areas of the country, migration can open possibilities for interracial marriage that are less likely to occur for non-migrants. At the same time, as gender plays an important role in opportunities for intermarriage, the effects of migration likely vary according to intersections of race and gender. An examination of the data on marital unions from the 2009 Brazilian National Household Survey, which includes large numbers of earlier cohorts of mass migration, demonstrates the influence of migration on interracial marriage. This study finds that the effect of migration on the odds of being interracially married (in comparison with being in a same-race marital union) vary according to the race and gender of the spouse. This study is one of the first to tie together two demographic phenomena—migration and interracial marriage—that have not previously been examined in the Brazilian context. It also provides a new lens through which to understand interracial marriage in Brazil and has implications for future studies of family formation in Latin America and the Caribbean.

1. Explaining Interracial Marriage

Brazil has race-mixing as a potent part of its national ideology, in which Brazilians are presumed to be racially mixed and interracial marriage is understood as widely prevalent (Guimarães 1999). Social scientists have used a variety of ways to explain these supposedly high amounts of race-mixing. One of the dominant ways of explaining intermarriage in Brazil has been the idea of racial democracy, popularized by the sociologist Gilberto Freyre, in which race did not prevent intermarriage (Freyre 1933). Another dominant perspective to explain interracial marriage has been whitening, in which Brazilians of color seek out relationships with white Brazilians to increase their social status as well as produce lighter offspring who will have more advantages in a society stratified by race and color (de Azevedo 1955; Schwartzman 2007). In addition, status exchange is yet another way to explain interracial marriage in Brazil, in which lower-educated white spouses and higher-educated black partners form relationships that allow them to match up due to their different social statuses (de Azevedo 1955; Guimarães 1999; Gullickson and Torche 2014; Telles 2004). More recently, appreciation for blackness and even the opposite of whitening–known as “darkening” (Osuji 2013)—have been offered as contemporary ways of viewing interracial marriage in Brazil in light of qualitative research on marital unions and family formation. In a racially unequal society like Brazil (Silva and Paixão 2014), understanding the dynamics behind interracial marriage reveals the ways that family formation challenges, maintains, or reconfigures racial inequality.

However, social scientists have overlooked one simple factor that may contribute to interracial marriage in Brazil: migration. Brazil experienced high levels of migration from the 1950s to the 1970s, in which people from the northeast migrated to places like São Paulo and other areas in the industrializing southeast (Gries et al. 2011; Skop et al. 2006; Yap 1976). For several decades, migration continued to the southeast (Fiess and Verner 2003). There has also been return migration to the northeast, as elderly migrants retire or return to their communities of origin (Barbosa et al. 2010). Scholars have often emphasized population size and physical proximity when explaining patterns of interracial marriage; when a community comprises a smaller proportion of the overall population, they should be more likely to intermarry than larger subpopulations (Blau et al. 1982; Lieberson and Waters 1988). Like most countries in Latin America, race in Brazil is highly regionalized, with the majority of white people living in the southern regions and most people of color in the northern regions (Telles 2004). In addition, people do not normally marry people at the national level; local marriage markets are very important when considering intermarriage (Tomás 2017). When taking local marriage markets into account, the odds of intermarriage drop from 15–43%, with some of the largest drops occurring between brown wives and white husbands.

Given the importance of population composition, moving to a place where fewer people share the same ethno-racial category can increase an individual’s likelihood of intermarriage. In this new area, they can be exposed to potential partners with whom they would have had less contact than before. On the other hand, those areas may have stronger social barriers to intermarriage in comparison with places where they are more numerous. Nevertheless, due to racialized and gendered notions of desirability in heterosexual marital unions, the effects of migration may differ by the race and gender of spouses (who would have different sets of potential partners). Overall, although migration increases the likelihood of non-traditional marital unions in the United States (Rosenfeld and Kim 2005), it is unclear whether Brazil—with its history of family formation, race relations, and migration through colonization and industrialization—would experience similar dynamics.

This study draws on data from the 2009 Brazilian National Household Survey (PNAD). Although newer data are available, the shorter life expectance of Brazilians who experienced the first large waves of domestic migration of the 1950s would not be included in newer surveys. Using a slightly older data set allows us to include older Brazilians alongside younger Brazilians who were not a part of the domestic migration boom. It also has the benefit of including race variables, which were not available during the migration of the industrialization period (Gries et al. 2011; Skop et al. 2006; Yap 1976).

Controlling for factors such as educational attainment and region of the country, this study reveals the role of migration on spouses’ likelihood of interracial marriage. In addition, this analysis reveals how the role of migration differs for husbands and wives based on the racial category of the partner vis-a-vis potential same-race mates. By examining migration and interracial marriage in a society that has been highly influenced by both, this study provides a more nuanced understanding of marriage and family formation and reveals new ways to understand the experiences of domestic migrants in host communities. This is despite most marriages occurring between people within the same racial category (Telles 2004). The dominant ways that scholars have understood race in Brazilian marriage markets have included racial democracy, whitening within and across generations, and status exchange (de Azevedo 1955; Gullickson and Torche 2014; Ribeiro and Silva 2009; Schwartzman 2007; Telles 2004). Qualitative research has shown that there are other ways to understand marital unions and family formation, including maintaining black pride (Hordge-Freeman 2013) as well as darkening (Osuji 2013).

However, migration, both international and domestic, has been an important feature of Brazilian society since the colonial period, in which largely Portuguese men and enslaved African and subjugated indigenous women created the first racially mixed populations (Morner 1967). With the industrialization of the country has come several waves of migrations, both domestic and international. Like most of Latin America (Wade 2010), race and ethnicity are regionalized in Brazil, with Afro-descendants present in larger numbers in the northeast, particularly along the coasts and on routes of the Atlantic trade in enslaved Africans. On the other hand, white Brazilians are more populous in the southern regions, which received large numbers of European immigrants in the 19th and early 20th centuries (Lesser 2013).

2. Migration and Race-Mixing

2.1. The Colonial Era

Migration was a central part of the initial conditions for race-mixing in Brazilian colonial history. When Pedro Álvares Cabral landed in Brazil in 1500, the goal of the Portuguese Crown was conquest and trade. This involved the decimation and enslavement of indigenous peoples and the subsequent forced migration of 4.9 million Africans, resulting in Brazil receiving the largest number of enslaved people in the Western hemisphere. However, unlike British North America, where entire families migrated from Europe to flee religious persecution, it was largely Portuguese men who made their way to colonial Brazil in search of fortune (Telles 2004). With the scarcity of Portuguese women, Portuguese migrants mated with indigenous and enslaved African women (Degler 1986), creating Brazil’s first racially mixed populations (Morner 1967). These sexual encounters often took place between unequals, such as between masters and their female slaves, and often involved sexual violence and coercion.

Brazil under Portuguese rule did not have anti-miscegenation laws similar to its British colonial counterparts, the Catholic Church’s Inquisition-related concerns over the purity of bloodlines led to the regulation of formal marriage between racial “unequals” (Nazzari 1996). This led to the use of dowries and other policies to ensure that religious marriage occurred mainly between social equals. Partially, as a result, concubinage proliferated in colonial and post-independence Brazil. However, migrating from major cities to the hinterlands, along with the shortage of white women in these areas, made the norms of the Portuguese Crown and the Catholic Church harder to put into practice. This meant that attempts to prevent and delegitimize even consensual, stable marital unions were largely futile.

2.2. European Immigration and Race-Mixing after Abolition

After abolition in 1888, Brazilian elites were horrified by the large numbers of former slaves and their descendants that formed the majority of the country (Viana 1952). They noted Afro-Brazilians’ higher rates of disease, malnutrition, and infant mortality that were all linked to their lower fertility rates in comparison with white Brazilians. Viewing Afro-descendants as dying off, European migration became a solution to the problems associated with transitioning the country from a slave- to a wage-based, industrial economy (Skidmore 1974; Stepan 1991). As a result, at the beginning of the 20th century, the Brazilian government sponsored thousands of European migrants as wage laborers to help modernize the country. They expected this influx of migrants to accelerate the whitening of the Afro-descendant population through race mixture and the already-existing demographic phenomena. This approach to solving the “race problem” through absorbing the Afro-Brazilian population was critiqued by many scholars in the African diaspora, from WEB DuBois (Hellwig 1992, p. 52; DuBois 1985, p. 181) in the United States to Brazilian Black movement leader and intellectual, Abdias do Nascimento (Nascimento 1989).

In the 1930s, Gilberto Freyre, a Brazilian sociologist and public intellectual, shifted understandings of Brazilians of color as a liability and hindrance to socioeconomic progress (Freyre 1933; Hasenbalg 1985). As social scientists realized they had underestimated the demise of the Afro-Brazilian population, Freyre praised the African cultural heritage of Brazil and popularized the idea of Brazil having harmonious race relations, the assimilation of European migrants into Brazilian culture and society, and high rates of race-mixing. Brazil as a “racial democracy” influenced future scholars and became part of the Brazilian national creed (Guimarães 2005). His bucolic vision of Brazil was similar to elite Latin Americans’ notions of mestizaje (race mixture) in other societies. This includes the formation of “the cosmic race” in Mexico (Vasconcelos 1976) and Venezuela as a “coffee with milk” society (Wright 1990). These perspectives all stood in stark contrast to the formal and informal Jim Crow of the United States.

However, the realities of race-mixing were very different from the ideals espoused in this notion of racial democracy. One critique of Freyre’s work is that it was unclear whether race mixture involved sexual liaisons, concubinage, rape, or marital unions (Telles 2004). For example, he referred to the common practice of white male elites, including himself, having sexual relations with mulata domestics, despite its intimations of rape and sexual coercion. In addition, in practice, interracial dating and marriage were stigmatized and outside of the norm. Freyre admitted that “although the whites procreate with the Negroes and mulattoes and they do not disdain to recognize publicly these children, they do not want their children to marry those races (Freyre 1980, pp. 395–96).” Many were also “social isolates,” living far from their friends and families of origin (Staley 1960).

On the rare occasions that interracial marriage occurred in the past, it involved Brazilian “elites of color,” often mulatos, having a higher socioeconomic status than their white partner (de Azevedo 1955; Ianni 1960). This phenomenon is known as “status exchange,” in which people compensate for their lower racial status categories by having a higher socioeconomic status than their other-race partners (Davis 1941; Merton 1941). In one study of elites of color, albeit without explicitly using the term “status exchange,” the Brazilian scholar Thales de Azevedo found that elites of color thought intermarriage would aid in their upward mobility (Azevedo 1955). However, he found that in many cases, the white spouses had a lower socioeconomic status than these elites of color due to greater opposition to intermarriage in the higher classes. Similar to previous attempts to infuse European blood into the nation, darker Brazilians marrying white or lighter partners to gain social status also became known as “whitening” (Skidmore 1974). This sentiment is conveyed in sayings that are still popular today that “money whitens”.

However, the meanings of race mixture have largely occurred along gendered lines. For example, the upward mobility of women of color has been historically tied to their fertility. This has its origins in slavery, in which, for centuries, masters were more likely to manumit enslaved women who were their sexual partners —albeit coercively—as well as the mixed-race offspring of those encounters (Freyre 1980). One famous case of upward mobility through intermarriage is Chica da Silva, who gained freedom for herself and her offspring by marrying her enslaver, a powerful white man in colonial Minas Gerais (Furtado 2009). Her story has been portrayed multiple times in Brazilian media. This link between fertility and upward mobility is encapsulated in the famous 1859 painting “Ham’s Redemption” by Modesto Broncos, featuring an elderly dark brown woman giving praise for her pale grandbaby featured at the center of the image with her mixed-race daughter sitting next to the child’s white father. These examples show how whitening across generations is understood as progress for women of color.

For men of color, race-mixing was not as linked to the color of their progeny. Instead, “socially white” or high-status Afro-Brazilian men recognized that marrying white women enabled them to gain legitimacy in elite circles, as well as status and wealth (de Azevedo 1955; Dantas 2016). Long after abolition, these men drew on their white wives’ family connections to cultivate their own elite status vis-à-vis other Afro-descendants (Daniel 2006). This was the basis for the notion of a “mulatto escape hatch” in which they would enjoy freedom from racial discrimination (Degler 1986), although this idea has largely been debunked (Telles 2004). On the other hand, Afro-Brazilian men were highly desired by white women, who stereotyped them as so “virile and strong that they did not like those with more delicate mannerisms (de Azevedo 1955, p. 84) [author’s translation]”.

3. Contemporary Interracial Marriage

According to Brazil’s 2022 Continuous National Household Survey (IBGE 2023b), close to 42.8% of the population identified as white, 45.3% as brown, and 10.6% as black, with the rest divided between Asian and Native American populations (IBGE 2023a). Despite nonwhite Brazilians comprising most of the population, marital unions across color categories comprise only 30% of all marriages, including both cohabitation and formal marriage (Beltrão et al. 2012). Thus, most marriages occur within racial categories—much lower than what would be predicted given Brazil’s large nonwhite population. It is also unexpected, given the popular “racial democracy” ideology that claims interracial marriage is prevalent in Brazil.

Brazilian interracial marriages are most common between proximate categories, such as between white and brown spouses and between brown and black spouses (Beltrão et al. 2012). The prevalence of interracial marriage in proximate racial categories may be due to several other factors. For example, although race and skin color are analytically different concepts around the world (Glenn 2009; Hunter 2005; Monk 2016), skin color largely drives racial categorization in Brazil (Telles and Paschel 2014). For this reason, although individuals in interracial marriages may be characterized differently by race, they may be similar in skin color. This physical similarity could facilitate relationships between ethno-racial others. For example, in one study, many black Brazilians who married white spouses recalled others questioning their blackness due to their more ambiguous features and lighter skin color (Osuji 2019). For this reason, they do not experience outsider hostility to their relationship the same way that couples of starkly different colors do.

Skin color may also explain why Brazilians recognize interracial marriage being more common than the statistical reality. Brazil is known for its ambiguity in racial classification. Distinctions between self-categorization and how outsiders racially categorize the same individuals become wider for darker individuals and are related to socioeconomic status (Telles 2002). This means that couples may identify themselves as members of the same racial category in surveys and census data, but Brazilians who see them may think they are an interracial couple. On the other hand, couples who identify as members of different racial categories may be seen by others as being of the same race due to similarities in color. Given the tendency of many demographers to collapse the brown (pardo) and black (preto) racial categories into a larger negro or Afro-Brazilian category due to socioeconomic similarities (Telles 2004), the extent of interracial marriage may be over- or under-estimated in the Brazilian context.

Yet, prior understandings of race-mixing remain pervasive. Contrary to the ideals of racial democracy, interracial couples continue to experience stigma (Moutinho 2004). In one qualitative study comparing black–white couples in Brazil and the United States, Rio de Janeiro white families often looked down on black spouses marrying into their families (Osuji 2019). Black spouses who married white partners often experienced overt opposition, the use of insults through humor, and an “irony of opposition” in which people of mixed-race parentage or who had dated or married interracially were opposed to the relationships. Couples who avoided white family opposition involved black women who were seen as racially ambiguous or white women who intermarried later in life once elder family members had passed on and they had more autonomy. This was different from US couples white in-laws and extended family largely employed “colorblind” racial discourse to express displeasure in socially desirable ways. In addition, when in public with strangers, Brazilian couples experienced hostility in predominantly white regions of the country and white, wealthy spaces nearby; US couples referred to black individuals—not spaces—as expressing hostility. In both sites, despite Brazil’s extensive history of race mixture over the last five centuries, parents often had their parenthood questioned by strangers.

Nationally representative studies show that status exchange continues to characterize contemporary interracial marriages, with nonwhite Brazilians often having higher levels of education than their white partners (Ribeiro and Silva 2009; Gullickson and Torche 2014). In addition, nonwhite spouses who interracially marry are more likely to identify their children as white, proving that the whitening ideology is still alive and well in Brazil (Schwartzman 2007). Nevertheless, Afro-Brazilian organizing over the last several decades has increased black consciousness (Paschel 2018; Paschel and Sawyer 2008). As a result, some white spouses have pursued relationships with Afro-Brazilians to attain the opposite of whitening: “darkening (Osuji 2013)”.

However, gender is an important aspect of interracial marriage in Brazil today. Intermarriage is more common between white women and brown and black men than involving white men and nonwhite women (Gullickson and Torche 2014; Ribeiro and Silva 2009; Telles 2004). In 1955, Azevedo explained that Afro-Brazilian women were seen as “the most ardent feminine type and thus more sexually available [my translation].” This idea of the hypersexuality of black women remains prominent throughout the Americas (see Hill Collins 2004 for a discussion of “The Jezebel”), with several qualitative studies in Bahia (Williams 2013) and Rio de Janeiro (Twine 1998; Osuji 2019) showing that Brazilians assume women of color are prostitutes when in the company of white husbands. Nevertheless, their fertility is still tied to interracial marriage in a way it is not for Afro-Brazilian men. Even today, Afro-Brazilian women who give birth to black and brown children are seen as having a “dirty womb (Hordge-Freeman 2015)”, providing an incentive for them to engage in race-mixing. In addition, those who marry white men are seen as whitening their descendants (Osuji 2013). This is different for contemporary white women who are sometimes understood as “darkening” their lineage. However, one recent qualitative study found that, even if outsiders understand black partners marrying white to increase their social status, it is seen more negatively than before. For example, Brazilians across color disparaged nonwhite soccer players for their preferences for white women in dating and marriage (Osuji 2019). In addition, in some black social movement circles, “marrying up” and interracial marriage more generally is seen as reproducing white supremacy.

Yet, research on race mixture and white women in Brazil shows that they repeat centuries-old stereotypes of black hypersexuality and their irresistibility (Moutinho 2004). They experience a “privilege of preferences” in which white women openly described desires to approximate Afro-Brazilian culture, religion, hobbies, and sexuality through marriage to “big black men” or negão (Osuji 2019). Several black men married to white women in the study described themselves as being “pursued by blondes” (with “blonde” acting as a euphemism for a white woman) and as ultimately giving in to the chase. This perspective is revealed in the contemporary adage that “every little white woman loves a big black man”.

4. Domestic Migration in Brazil

International migration has been important in Brazil’s history, especially involving enslaved Africans and European colonizers, yielding racially mixed populations. However, race in Latin America is highly regionalized, with people in these societies mapping ethno-racial and phenotypic attributes onto persons from particular areas of the country (Wade 2010). In Brazil, the country becomes progressively whiter from north to south (Telles 2004). For this reason, nonwhite spouses in the south and southeast regions, where they are smaller proportions of the population, marry white Brazilians at higher rates.

Beginning in the 1960s, industrialization, large-scale capital investment in agriculture, and declines in environmental and economic conditions prompted a “rural exodus” of millions of Brazilians to more urban centers (Gries et al. 2011; Skop et al. 2006; Yap 1976). Since then, rural-to-urban migration has continued with the majority coming from the northeast to the more industrial southeast region of the country, with São Paulo as the primary destination. While there has been some degree of “return migration” from the southeast to the northeast more recently, migration in the opposite direction remains dominant (Barbosa et al. 2010). These large fluctuations in the population may have influenced the potential for forming and staying in interracial marital unions. Specifically, people of color from the northeast, often racialized as nordestinos, as well as others who migrate to the south may have had more opportunities to intermarry given the differences in population characteristics. Yet, given an understanding of race and gender in Brazil, those social categories can affect opportunities for intermarriage for migrants. Given the racialization of region in the Brazilian context, examining the impact of migration on interracial marriage patterns can provide a new way of understanding family formation in Brazilian society.

5. Methodology

Data and Sample

This study draws on household data from the 2009 Brazilian National Household Survey (PNAD), a dataset that is collected by the Brazilian Institute of Geography and Statistics (IBGE) every year. The PNAD involves a geographically stratified sample of the entire country covering 153,837 households and 399,387 individuals. This analysis involves households in which the household head is in a heterosexual marital union with a cohabiting partner in which both partners were at least 18 years of age. One of the benefits of this survey is that it includes people who are alive yet old enough to have participated in the early waves of domestic migration during the 1960s (F). In Brazil, cohabiting relationships are known as “stable unions” and comprise close to a third of all marital unions in Brazil (Greene and Rao 1995). They are often long-term, involve high levels of commitment, and have community and property rights recognized by the state. For these reasons, stable unions are included alongside formal marital unions. In total, there are 122,656 individuals in 61,328 households with couples.

The IBGE collects data on race involving five different categories: black, brown, white, Asian, and indigenous. Due to the small representation of the latter two categories in the data, this study excludes households with marital unions involving indigenous individuals and “yellow” (Asian) individuals. It also excluded cases (0.2%, N = 119) in which color information on the respondent was missing.

To capture interstate migration, being a migrant is measured based on the survey question asking individuals if they had lived in another state or foreign country.1 To examine the effects of migration for only Brazilians, couples involving foreign-born partners were removed from the data set. Although the PNAD has questions about whether individuals have up to ten years of uninterrupted residence in their current state, the question asking whether they had ever migrated captures the experiences of older cohorts of Brazilians who were a part of earlier waves of migration mentioned earlier and settled away from the region of their birth.2 I examine whether at least one of the partners in the relationship is a migrant for partners within each model.

As Jessie Bernard famously put it, “…in every marriage, there are two marriages (Bernard 1982)”. Since most people in Brazil marry within color categories, migration may play one role for a white person who intermarries, for example, and a different one for their brown spouse. In addition, migration may play a completely different role in terms of whether a black person marries a white or brown person instead of another black person. Multinomial logistic regression models allow for these types of analyses in which they produce the odds of exogamy relative to endogamy while examining the effect of migration, controlling for other factors. In addition, since intermarriage most often occurs between proximate color categories, this method allows the effect of internal migration to be different when marrying people in the other two categories instead of one’s own. Furthermore, interracial marriage is an intersectional phenomenon in which people of the same race have different experiences in dating and marriage markets according to gender (Kao et al. 2019). An analysis of interracial marriage that merely controls for gender can miss out on important ways that intersections of race and ethnicity can yield different experiences. For example, the effect of migration for black women may be different for black men. For this reason, the analyses of the effect of migration on interracial marriage were done separately for each race-gender combination in heterosexual unions.

While logistic regression models predict the probabilities of two possible discrete outcomes, multinomial logistic regression allows for more than two discrete outcomes in a dependent variable that is categorically distributed. This model can be written as a series of models, as seen here.

$$\begin{array}{l}\ln \frac{\mathrm{P}\left(\mathrm{marriage}=\mathrm{other}\mathrm{race}1\right)}{\mathrm{P}\left(\mathrm{marriage}=\mathrm{same}\mathrm{race}0\right)}={\mathrm{b}}_{10}+{\mathrm{b}}_{11}+\dots +{\mathrm{b}}_{\mathrm{migrant}}\\ \ln \frac{\mathrm{P}\left(\mathrm{marriage}=\mathrm{other}\mathrm{race}2\right)}{\mathrm{P}\left(\mathrm{marriage}=\mathrm{same}\mathrm{race}0\right)}={\mathrm{b}}_{20}+{\mathrm{b}}_{21}+\dots +{\mathrm{b}}_{\mathrm{migrant}}\end{array}$$

In this model, b are the different independent variables and b_migrant shows the effect of one of the partners being a migrant on the likelihood of two different intermarriages in comparison to being in a same-race marital union.

Education levels, population composition of individuals within the different racial categories within a given region, and urban/rural residence all influence the likelihood of interracial marriage (Ribeiro and Silva 2009; Telles 2004). For these reasons, the models controlled for these factors in the analysis. It includes four educational categories measured as a series of dummy variables: having less than grammar school, completion of grammar school, high school, and college. As husbands on average tend to be older and more educated than wives on average in Brazil, as well as for the sake of similarity across models, the husband’s education is used in all of the models. The study also measured the five regions of residence in Brazil (the southeast, south, center-west, north, and the northeast) as a series of dummy variables. Each region is associated with varying white proportions of the population. Specifically, white Brazilians comprise a different proportion of the population in the south (78%), southeast (56%), center-west (42%), northeast (29%), and north (24%) (IBGE 2010). For this reason, the region variable is highly collinear with white/nonwhite proportions of each region.

Examining first marriages by only examining younger cohorts of individuals allows scholars to disentangle the associations between higher-order marriages and interracial marriages. However, data limitations prevent these types of analysis with the PNAD data in the Brazilian case; data for cohorts that were engaged in earlier migration do not have race information. In other developing countries, internal migration accelerates marriage for migrants once they return to the sending community (Jampaklay 2006). This challenges the usefulness of only solely examining the youngest age groups when understanding the effects of migration on interracial marriage. Due to these issues of data limitations and only looking at younger cohorts of migrants, this study examines the effect of migration across age cohorts. Intermarriage is far more common among recent cohorts of individuals than their older counterparts (Ribeiro and Silva 2009). For this reason, the model controls for the year of birth of respondents by creating a series of dummy variables for cohorts. These were: those born before 1950, in the 1950s, the 1960s, the 1970s, and the 1980s and so on, using the husband’s cohort since because they tend to be older than their wives on average.

A series of multinomialn logistic regression models examined the effect of migration on interracial marriage for men and women within each of the three (black, brown, white) racial categories. For the dependent variable, it was an endogamous couple if both individuals were members of the same racial category. Couples were coded as interracial couples if individuals were of different racial categories. The base comparison category would be the same race couple since this is most relationships in Brazil. However, the analyses are conducted by changing the race and gender of the spouses. For example, the analysis compares the odds of brown husbands marrying a white wife instead of a brown wife if either of the spouses in the couple was a migrant. There is a separate analysis comparing the odds of brown wives marrying a white husband instead of a brown husband. This allows for the effect of migration to change depending on both the race and gender composition of the couple.

One shortcoming of the PNAD survey is that it does not ask about when the marital union began, as it is impossible to know whether migration occurred before or after the marriage or whether these are first or later-term marriages. In addition, racial categories can be region-specific. For example, Bahian white (branco da Bahia) refers to a whiteness that would be recognized in the northeastern state of Bahia, but that would likely be classified as brown in the southeast. The reverse may also be true: a person who considers themselves brown in southern, predominantly white regions might be seen as white in the predominantly brown northeast region. While these are important considerations, the limits of the data cannot disentangle migration timing across age cohorts nor how people shift their classification due to migration. Still, it is useful for understanding whether there may be an overall relationship between interracial marriage and migration.

6. Findings

Table 1. Descriptive statistics.

Variable	Mean	Std. Dev.
Migrant in Couple	15.71%	0.36
Grammar School	9%	0.28
Grammar	50%	0.50
High school	27%	0.45
College	14%	0.35
Husband’s Cohort
Up to 1949	21%	0.41
1950 to 1959	20%	0.40
1960 to 1969	25%	0.50
1970 to 1979	23%	0.42
1980 and after	11%	0.31
South	17%	0.38
Center-west	11%	0.31
North	10%	0.30
Northeast	30%	0.46
Southeast	33%	0.47
N = 61,328

shows interracial couples by migration status in the sample. Close to 16% of all couples involve at least one person who is a migrant between states. Table 1 also shows the representation of couples across educational levels, with half of the sample completing grammar school, and close to another third having finished high school. Only 14% of respondents went to college and another 9% did not finish grammar school. Different age cohorts are also represented in the sample, with most of the sample being born between 1960 and 1969, comprising 25% of the sample. Those in the other cohorts comprise approximately 20% of respondents each, except for those born since 1980 who only make up 11% of the sample. All of the regions of the country are represented in the sample, with the greatest proportion living in the southeast of the country, where 33% of the sample lives. Another 30% is living in the northeast whereas the smallest proportion of respondents live in the northern region of the country, where only 10% of the sample lives.

The results from the multinomial logistic regression are shown in

Table 2. Brown wives’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a Brown Husband	White Husband		Black Husband
	Model 1	Model 2	Model 3	Model 4
Migrant spouse	____	1.07 *	____	1.01
		(0.03)		(0.05)
Husband’s education
Grammar	1.18 **	1.17 **	0.92	0.92
	(0.06)	(0.06)	(0.07)	(0.07)
High school	1.50 ***	1.50 ***	0.94	0.94
	(0.09)	(0.09)	(0.09)	(0.09)
College	2.33 ***	2.33 ***	1.14	1.14
	(0.16)	(0.16)	(0.14)	(0.09)
Husband’s cohort
Up to 1949	1.68 *	1.66 *	1.46	1.46
	(0.36)	(0.36)	(0.54)	(0.54)
1950 to 1959	1.58 *	1.57 *	1.53	1.53
	(0.34)	(0.34)	(0.58)	(0.57)
1960 to 1969	1.49	1.48 †	1.47	1.47
	(0.32)	(0.32)	(0.54)	(0.54)
1980 and after	1.48	1.47 †	1.50	1.51
	(0.32)	(0.32)	(0.56)	(0.56)
Urban area	1.09 *	1.08 †	1.51 ***	1.51 ***
	(0.04)	(0.04)	(0.10)	(0.11)
Region
South	2.13 ***	2.10 ***	0.90	0.90
	(0.13)	(0.13)	(0.11)	(0.11)
Center-west	0.91 †	0.92 †	0.83 *	0.83 *
	(0.04)	(0.04)	(0.07)	(0.07)
North	0.52 ***	0.52 ***	0.56 ***	0.56 ***
	(0.03)	(0.03)	(0.05)	(0.05)
Northeast	0.59 ***	0.59 ***	0.71 ***	0.71 ***
	(0.02)	(0.02)	(0.04)	(0.04)
Constant	0.24 ***	0.23 ***	0.07 ***	0.07 ***
	(0.05)	(0.05)	(0.03)	(0.03)
Likelihood ratio	1039.64 ***	1044.51 ***	1039.64 ***	1044.51 ***
Observations	26,111

* z < 0.05; ** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

,

Table 3. Brown husbands’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a Brown Wife	White Wife		Black Wife
	Model 1	Model 2	Model 3	Model 4
Migrant spouse	____	1.01	____	0.98
		(0.03)		(0.06)
Husband’s education
Grammar	1.22 ***	1.22 ***	0.75 **	0.75 **
	(0.06)	(0.06)	(0.07)	(0.07)
High school	1.67 ***	1.67 ***	0.67 ***	0.68 ***
	(0.09)	(0.09)	(0.07)	(0.07)
College	2.78 ***	2.78 ***	0.76 †	0.77 †
	(0.18)	(0.18)	(0.11)	(0.11)
Husband’s cohort
Up to 1949	1.01	1.00	1.22	1.22
	(0.17)	(0.17)	(0.52	(0.52)
1950 to 1959	0.98	0.98	1.44	1.44
	(0.17)	(0.17)	(0.61)	(0.61)
1960 to 1969	0.93	0.93	1.25	1.25
	(0.16)	(0.16)	(0.53)	(0.53)
1980 and after	0.87	0.87	1.39	1.39
	(0.15)	(0.15)	(0.60)	(0.58)
Urban area	1.19 ***	1.19 ***	1.50 ***	1.50 ***
	(0.05)	(0.05)	(0.12)	(0.13)
Region
South	2.21 ***	2.21 ***	0.79	0.79
	(0.13)	(0.13)	(0.13)	(0.13)
Center-west	0.86 ***	0.86 ***	0.75 **	0.75 **
	(0.04)	(0.04)	(0.08)	(0.08)
North	0.51 ***	0.51 ***	0.55 ***	0.55 ***
	(0.02)	(0.02)	(0.06)	(0.06)
Northeast	0.63 ***	0.63 ***	0.70 ***	0.70 ***
	(0.02)	(0.02)	(0.05)	(0.05)
Constant	0.39 ***	0.39 ***	0.06 ***	0.06 ***
	(0.07)	(0.05)	(0.03)	(0.03)
Likelihood ratio	1298.30 ***	1298.47 ***	1298.30 ***	1298.47 ***
Observations	26,671

** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

and

Table 4. Black wives’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a Black Husband	White Husbands		Brown Husbands
	Model 1	Model 2	Model 3	Model 4
Migrant spouse	____	1.44 ***	____	1.26 **
		(0.13)		(0.10)
Husband’s education
Grammar	1.12	1.09	1.01	0.99
	(0.16)	(0.16)	(0.12)	(0.12)
High school	1.21	1.19	0.77 †	0.76 †
	(0.20)	(0.20)	(0.11)	(0.11)
College	1.83 **	1.82 **	0.88	0.88
	(0.36)	(0.36)	(0.17)	(0.17)
Husband’s cohort
Up to 1949	1.20	1.12	1.36	1.30
	(0.65)	(0.61)	(0.68)	(0.65)
1950 to 1959	1.37	1.30	1.59	1.53
	(0.74)	(0.70)	(0.80)	(0.77)
1960 to 1969	1.59	1.51	1.59	1.45
	(0.85)	(0.80)	(0.74)	(0.72)
1980 and after	2.29	2.19	2.25 †	2.19 **
	(1.24)	(1.19)	(1.14)	(1.11)
Urban area	1.27 †	1.23 †	0.95	0.93
	(0.18)	(0.17)	(0.11)	(0.11)
Region
South	1.32 *	1.22	0.58 **	0.55 ***
	(0.19)	(0.18)	(0.10)	(0.10)
Center-west	1.90 ***	1.97 ***	1.80 ***	1.84 ***
	(0.26)	(0.27)	(0.25)	(0.26)
North	1.77 ***	1.79 ***	2.87 ***	2.89 ***
	(0.29)	(0.30)	(0.43)	(0.43)
Northeast	0.74 **	0.71 ***	1.40 ***	1.36 ***
	(0.08)	(0.07)	(0.13)	(0.12)
Constant	0.24 **	0.22 **	0.39 †	0.36 **
	(0.13)	(0.12)	(0.20)	(0.19)
Likelihood ratio	229.58 ***	249.25 ***	229.58 ***	249.25 ***
Observations	3794

* z < 0.05; ** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

. Table 2 shows the results for brown wives. The first equation in Model 1 shows that, as predicted, level of education has a statistically significant effect on whether brown partners marry white spouses instead of fellow brown spouses, controlling for the other variables in the model. It has a direct effect, with increasing levels of a husband’s education being associated with an increase in the probability of marrying a white husband instead of a brown one. In comparison with wives with husbands born in the 1970s, brown wives are more likely to be married to white husbands instead of another brown person in the earlier cohorts. This may reflect a change in racial identification over time; many who identify as brown today may have been white in an earlier period (Bailey 2009; Schwartzman 2007; Telles 2004).

Living in an urban area also increases the odds of marrying a white person by 9%. Region also has an impact on the marriage of white husbands to brown wives, with those in the predominantly white south being more than twice as likely to marry a white instead of a brown wife than if they lived in the southeast of the country. The reverse is true for the other regions of the country, where brown individuals are less likely to marry a white person, showing the importance of population composition on interracial marriage.

Model 2 shows how having a migrant spouse increases the odds of a brown wife being interracially married to a white husband by 7%, when controlling for the husband’s education, cohort, urban residence, and region. The inclusion of migration does not change the effect of the other variables in the equation. Model 3 shows the odds of brown wives marrying a black husband instead of another brown person. Level of education did not affect being married to a black husband and neither does the husband’s cohort. However, brown wives who live in the city have a 51% increase in the odds of being married to a black person when controlling for the other variables in the model. In addition, living in the center-west, north, or northeast regions all decrease the odds of being married to a black person instead of a brown person in comparison with living in the southeast. This reflects the population composition of those states, where brown people largely reside. Model 4 shows how, unlike for brown women who marry white husbands, migration does not affect being married to a black person.

The effect of migration for brown husbands is illustrated in Table 3, which shows the multinomial logistic regression results. Model 1 indicates that, similar to brown wives, higher levels of education increase the odds of having a white partner instead of a brown wife. For example, brown husbands who finished grammar school have a 22% increase in the odds of being married to a white wife. Brown husbands with a college degree have almost triple the odds of being married to a white woman in comparison with being married to a brown wife. This is to be expected, since higher levels of education are associated with an increase in the odds of marrying a brown partner. In addition, as with brown wives, living in an urban area as well as region all increase the odds of brown men being married to a white wife. Model 2 in Table 3 shows that, unlike for brown wives who marry white men, migration does not affect the odds of being married to a white woman when it comes to brown husbands. The same is true for marrying a black wife in comparison to a brown wife. Overall, migration does not increase the odds of interracial marriage for brown husbands. In fact, adding migration to the model barely changes the effects of the other variables in the model. This suggests that brown wives may have higher barriers to interracial marriage than their brown male counterparts.

Table 4 shows the results of the multinomial logistic regression for black wives interracially marrying. Unlike for brown husbands and wives, only the highest levels of education increase the odds of being interracially married to a white husband in comparison with marrying a brown one. In addition, cohort and urban residence were not found to be statistically significant in terms of affecting the odds of intermarriage. However, the direction of effects suggests a greater likelihood for those married to husbands in later cohorts. In comparison with living in the southeast, black wives who live in other regions experience greater likelihood of marriage to a white person. The only exception is for those who live in the northeast, who are less likely to marry white husbands than black wives who live in the southeast. Model 2 shows that migration increases the odds of black women intermarrying with white men by 44% in comparison with marrying black husbands.

However, examining Model 3 shows no statistically significant effect of education on the odds of having a brown husband in comparison with having a black one. In addition, cohort does not have an effect that is statistically significant. Only region has an effect that is statistically significant on the odds of being married to a brown man, in which case they increase for those living in the north, northeast, and center-west regions but decrease in the south in comparison with the southeast. However, Model 4 shows that when controlling for these other factors, migration increases the odds of black women marrying brown men by 26%. In addition, once migration is added to the model, having a husband in the youngest cohort has a statistically significant increase in the odds of marrying a brown man instead of a black one. Region also continues to have similar effects once migration is included in the model.

The results from the multinomial regression of black husbands who marry white and brown wives instead of black wives are found in

Table 5. Black husbands’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a Black Wife	White Wives		Brown Wives
	Model 1	Model 2	Model 3	Model 4
Migrant spouse	____	1.37 ***	____	1.32 ***
		(0.11)		(0.09)
Husband’s education
Grammar	1.16	1.15	1.18	1.18
	(0.16)	(0.16)	(0.13)	(0.13)
High school	1.46 **	1.45 **	1.06	1.05
	(0.22)	(0.22)	(0.14)	(0.13)
College	2.91 ***	2.88 ***	1.25	1.23
	(0.52)	(0.51)	(0.21)	(0.21)
Husband’s cohort
Up to 1949	1.07	1.04	1.80	1.75
	(0.51)	(0.50)	(0.82)	(0.80)
1950 to 1959	1.45	1.40	1.92	1.86
	(0.69)	(0.68)	(0.87)	(0.85)
1960 to 1969	1.44	1.42	1.95	1.92
	(0.68)	(0.68)	(0.88)	(0.87)
1980 and after	1.65	1.65 †	2.74 *	2.73 *
	(0.80)	(0.81)	(1.26)	(1.26)
Urban area	1.00	0.96	0.95	0.92
	(0.12)	(0.12)	(0.10)	(0.10)
Region
South	1.62 ***	1.52 **	0.65 **	0.61 ***
	(0.22)	(0.20)	(0.10)	(0.09)
Center-west	1.99 ***	2.07 ***	2.00 ***	2.06 ***
	(0.26)	(0.28)	(0.25)	(0.26)
North	1.96 ***	1.99 ***	2.92 ***	2.97 ***
	(0.26)	(0.31)	(0.40)	(0.41)
Northeast	0.81 *	0.79 *	1.40 ***	1.37 ***
	(0.08)	(0.08)	(0.11)	(0.11)
Constant	0.34 *	0.30 *	0.38 *	0.35 *
	(0.17)	(0.15)	(0.18)	(0.16)
Likelihood ratio	307.32 **	328.50 **	307.32 **	328.50 **
Observations	4687

* z < 0.05; ** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

. Unlike black wives who intermarry, when it comes to black men who marry white women, both high school and higher education increase the odds of interracial marriage, not just college. However, as with black wives, the region also matters for intermarrying with white spouses in comparison with marrying black ones. Once more, living in the northeast decreases the odds of intermarriage whereas it increases the odds in other areas in comparison with the southeast. As seen in Model 2, migration increases the likelihood of marrying a white wife by close to 40% after controlling for the other variables in the model. Both education and region are statistically significant in this model as well.

When comparing the dynamics of black husbands who marry brown wives, being born in the youngest cohort almost triples the odds of exogamy, as seen in Model 3. In addition, the region has a similar effect on black men marrying brown women instead of black ones; living in areas with smaller pockets of Afro-Brazilians, such as the south, decreased the likelihood of intermarriage but increased the odds where there was a larger brown population. Looking at Model 4, migration increases the odds of intermarrying with brown wives in comparison with marrying black women, controlling for other factors, including region and education. Being born in the youngest cohort, as well as region, continues to have a statistically significant effect on interracial marriage for black men who marry white wives.

Table 6. White wives’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a White Husband	Brown Spouses		Black Spouses
	Model 1	Model 2	Model 3	Model 4
Migrant Spouse	____	1.24 ***	____	1.29 ***
		(0.04)		(0.08)
Husband’s education
Grammar	0.72 ***	0.71 ***	0.52 ***	0.51 ***
	(0.04)	(0.04)	(0.61)	(0.06)
High school	0.53 ***	0.53 ***	0.39 ***	0.39 ***
	(0.03)	(0.03)	(0.05)	(0.05)
College	0.27 ***	0.27 ***	0.25 ***	0.25 ***
	(0.02)	(0.02)	(0.03)	(0.03)
Husband’s cohort
Up to 1949	0.28 ***	0.27 ***	0.25 ***	0.24 ***
	(0.06)	(0.06)	(0.11)	(0.10)
1950 to 1959	0.39 ***	0.37 ***	0.46 †	0.45 *
	(0.08)	(0.08)	(0.19)	(0.18)
1960 to 1969	0.52 **	0.50 ***	0.62	0.60
	(0.11)	(0.11)	(0.25)	(0.25)
1980 and after	0.62 *	0.61 *	0.67	0.66
	(0.13)	(0.13)	(0.28)	(0.28)
Urban area	1.11 **	1.08 †	1.51 ***	1.47 ***
	(0.05)	(0.05)	(0.15)	(0.15)
Region
South	0.40 ***	0.39 ***	0.42 ***	0.40 ***
	(0.02)	(0.02)	(0.04)	(0.04)
Center-west	2.20 ***	2.24 ***	2.18 ***	2.24 ***
	(0.10)	(0.10)	(0.20)	(0.21)
North	4.83 ***	4.92 ***	3.59 ***	3.67 ***
	(0.27)	(0.28)	(0.41)	(0.41)
Northeast	3.66 ***	3.61 ***	2.47 ***	2.43 ***
	(0.13)	(0.13)	(0.19)	(0.19)
Constant	1.00	0.93	0.15 ***	0.14 ***
	(0.22)	(0.21)	(0.06)	(0.06)
Likelihood ratio	5018.76 ***	5081.09 ***	5018.76 ***	5081.09 ***
Observations	31,423

* z < 0.05; ** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

and

Table 7. White husbands’ multinomial logistic relative risk ratios of interracial marriage.

Base = vs. a White Wife	Brown Spouses		Black Spouses
	Model 1	Model 2	Model 3	Model 4
Migrant spouse	____	1.30 ***	____	1.32 ***
		(0.04)		(0.09)
Husband’s education
Grammar	0.70 ***	0.69 ***	0.52 ***	0.52 ***
	(0.04)	(0.04)	(0.07)	(0.07)
High school	0.49 ***	0.49 ***	0.35 ***	0.34 ***
	(0.03)	(0.03)	(0.05)	(0.05)
College	0.24 ***	0.23 ***	0.17 ***	0.16 ***
	(0.02)	(0.02)	(0.28)	(0.03)
Husband’s cohort
Up to 1949	0.49 **	0.46 **	0.28 **	0.27 **
	(0.12)	(0.12)	(0.14)	(0.13)
1950 to 1959	0.65 †	0.62 †	0.45 †	0.43 †
	(0.17)	(0.16)	(0.22)	(0.21)
1960 to 1969	0.84	0.81	0.69	0.66
	(0.21)	(0.21)	(0.33)	(0.32)
1980 and after	1.06	1.03	0.94	0.92
	(0.27)	(0.26)	(0.46)	(0.44)
Urban area	0.98	0.95	1.89 ***	1.83 ***
	(0.04)	(0.04)	(0.22)	(0.22)
Region
South	0.38 ***	0.37 ***	0.35 ***	0.33 ***
	(0.02)	(0.02)	(0.04)	(0.04)
Center-west	2.33 ***	2.38 ***	2.03 ***	2.08 ***
	(0.11)	(0.12)	(0.21)	(0.22)
North	4.88 ***	5.00 ***	3.12 ***	3.18 ***
	(0.29)	(0.30)	(0.40)	(0.41)
Northeast	3.39 ***	3.32 ***	2.19 ***	2.14 ***
	(0.13)	(0.13)	(0.19)	(0.19)
Constant	0.61 †	0.57 **	0.10 ***	0.10 ***
	(0.16)	(0.15)	(0.05)	(0.05)
Likelihood ratio	4635.76 ***	4711.47 **	4635.76 ***	4711.47 ***
Observations	29,970

** z < 0.01; *** z < 0.001; ^† z < 0.1 Omitted categories are “having less than a grammar school education”, “being born between 1970 and 1979”, and “the southeast region”.

examine the multinomial logistic regression results for white wives and husbands in Brazil. The first table shows the odds of white women marrying brown and black husbands in comparison with white husbands. Model 1 in Table 6 reveals that education has a statistically significant effect on interracially marrying a brown husband that is inverse. Namely, the odds of a white woman marrying a brown husband in comparison with marrying a white man decrease by about 25% when he graduated elementary school in comparison with someone who did not. However, that decrease in the odds of marrying a brown person jumps to 73% when her partner has a college degree. In other words, education increases already high barriers to intermarriage for white spouses. Similarly, older cohorts of Brazilian husbands have lower odds of white wives marrying brown husbands instead of white husbands after controlling for the other variables in the model. This shows that, while the youngest cohort has a 38% decrease in the likelihood of marrying a brown husband, the oldest one has about a 70% decrease. This in turn shows a loosening in exogamy over time. On the other hand, living in an urban area as well as in regions with large numbers of people of color increases the odds of intermarriage whereas living in the predominantly white south decreases the odds of intermarriage to brown men for white wives.

However, Model 2 illustrates that migration increases the odds of exogamy to a brown husband instead of a white one after controlling for other variables in the model. Specifically, it increases the likelihood of intermarriage to a brown husband by 24%. Although living in an urban area is no longer statistically significant, the age cohort of the husband and the region still matter statistically when predicting the odds of intermarriage to a brown husband.

Similar dynamics play out in Models 3 and 4, with white wives marrying black men. Model 3 shows that husband’s higher education (75%) decreases the odds of intermarrying much more than if the husband had an elementary education (48%). Age cohort plays a similar role, in which out-marrying is less likely in the oldest husband’s cohorts; however, it is only statistically significant for husbands born in 1949 or earlier. Both those living in urban areas as well as those living in regions with larger Afro-Brazilian populations have an increase in the likelihood of white wives being married to a black man, controlling for other variables in the model. Model 4 shows that migration increases the likelihood of white women marrying black husbands instead of white husbands by 30%, controlling for education, region, urban residence, and age cohort. For white women, marrying brown and black men instead of white men show similar dynamics, including migration increasing their likelihood of marrying people of color.

Table 7 shows the results for white husbands’ likelihood of marrying brown and black wives in comparison with white wives. Model 1 shows that education, region, cohort, and urban dwelling are all statistically significant in terms of marrying brown women. Overall, in comparison with a white husband with no education, other white husbands are less likely to marry brown women. This gap increases with education, in which case higher-educated husbands experience a 76% decrease in the likelihood of marrying a brown woman in comparison with a 30% decrease for husbands who completed grammar school. Looking at age, intermarriage increases among the young, but is only significant for the oldest cohort, whose members have a 51% decrease in the likelihood of being married to a brown wife. Region also matters, with people from predominantly nonwhite regions being more likely to be married to a brown wife. As see in Model 2, migration increases the odds of exogamy controlling for the other variables in the model. Migration increases the odds of intermarriage to brown wives by 130%. The effects of these other variables largely remain the same, despite the addition of the migration variable.

Husbands’ predicted probabilities of interracial migrant marriages are shown in

Table 8. Predicted probabilities of husband’s interracial marriage.

Brown Husband’s Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
Brown wife	0.54	0.54
Black wife	0.42	0.42
White wife	0.04	0.03
Black Husband’s Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
Black wife	0.23	0.50
Brown wife	0.26	0.28
White wife	0.21	0.22
White Husband’s Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
White wife	0.73	0.71
Brown wife	0.22	0.24
Black wife	0.05	0.06

Predictions are for people with a high school education, born in the 1970s, and living in an urban area in the southeast.

. A brown husband’s probabilities of marrying a brown or black wife stays the same, regardless of whether there is a migration. While the multinomial logistic models show no effect of migration for intermarriage for brown husbands, the predicted probabilities show that migration decreases their probability of marrying a white spouse. For black husbands, having a migrant spouse increases the probability of marrying brown, and white wives. Strikingly, this also increases the chances of marrying a black wife. Examining

Table 9. Predicted probabilities of wives’ interracial marriage.

Brown Wives’ Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
Brown husband	0.67	0.66
White husband	0.26	0.27
Black husband	0.07	0.06
Black Wives’ Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
Black husband	0.61	0.56
Brown husband	0.22	0.27
White husband	0.17	0.18
White Wives’ Predicted Probabilities of Interracial Migrant Marriage
	Non-Migrant Spouse	Migrant Spouse
White husband	0.60	0.57
Brown husband	0.35	0.38
Black husband	0.05	0.06

Predictions are for people with a high school education, born in the 1970s, and living in an urban area in the southeast.

also shows the predicted probabilities of a white husband’s wife based on whether the relationship involves a migrant. This shows that migration increases the likelihood of white husbands marrying brown and black wives and decreases white husbands’ probability of marrying white wives.

7. Discussion

Migration was statistically significant in terms of the likelihood of exogamy. For black and brown spouses, moving to an area that is predominantly nonwhite would likely increase the odds of intermarriage. At the same time, their intermarriage to whites as well as brown or black spouses could be facilitated by their spouses moving to an area with more Brazilians of color.

However, the marriage market in which people engage and make decisions surrounding same- versus different-race partners depends on their intersection of social identities. As shown in research on intermarriage in the United States (Kao et al. 2019), important variations by race and gender can be ignored when we do not take their intersections seriously in understanding marriage and family formation. For brown wives, migration increased the odds of intermarriage, but only when they marry white husbands. This was different for brown husbands; migration did not have a statistically significant effect on marrying black and white wives. In addition, for brown husbands who marry black wives, education decreases the likelihood of intermarriage, whereas it has no effect for brown wives who marry black husbands or black spouses of either gender who intermarry with brown. This suggests a higher threshold to brown women intermarrying in comparison with their brown male counterparts. In addition, Brazilian racial categories exist in a greater continuum than many other societies, such as the United States, and is very relational (Harris 1964). In addition, demographers often collapse black and brown into one larger nonwhite or negro category (Telles 2004). However, the difference in the effect of migration by gender of the brown spouse shows that brown men may be driving understandings of a racial continuum, especially when it comes to family formation.

In terms of brown husbands, there are likely other dynamics at play that would prevent migration from influencing intermarriage. As mentioned before, their association of intermarriage to an experience of upward mobility in situ versus Brazilian women of color’s fertility may be a factor. While this is beyond the scope of this study, future studies should examine how they are different in comparison with brown women in dating markets. Migration to the southeast could “darken” them, leading to their further stigmatization there. On the other hand, staying in the northeast may allow them to live as brown people who are “socially white” as Azevedo and Freyre showed decades ago, even if they are not racially categorized as such. This can make them appealing to migrant spouses from other areas of Brazil.

For black wives and husbands, migration increased the odds of intermarriage to both brown and white spouses. However, black spouses comprise a much smaller proportion of the population than either brown or white partners, so it is important not to overstate the findings in their case. Nevertheless, their being at the bottom of the color hierarchy likely explains why migration plays a role in their intermarriage with white and brown Brazilians. For white spouses, migration increased the odds of intermarriage for both husbands and wives. Since there are so few marriages at the extremes, with white-black partnering being the least common of all interracial couplings, migration seems to facilitate these relationships.

One of the implications of the study is that scholars should not think that there is a single marriage market at the national level. Instead, scholars should examine how access to different types of partners influences marriage and family formation. For example, Brazilian white spouses have much higher levels of endogamy than nonwhite ones. From the perspective of a heterosexual white woman, there is likely little difference between a white man and a black man with an elementary school degree. However, that difference increases drastically when comparing white college educated men with brown or black educated men. A white woman with the option of a black or white man with a college degree, all else being equal, is even less likely to choose a white spouse. Access to highly educated white husbands makes highly educated brown and black men pale in comparison. This may be why status exchange is so important in the marriage market, particularly for white women. These white women may not have access to highly educated white husbands—the elites of their country. However, they may have access to Afro-Brazilian men due to the moderate levels of residential segregation (Telles 2004) alongside institutional racism that put these men at a disadvantage in Brazil vis-à-vis white men. This is likely the reason for “blondes” chasing brown men: they have access to better quality black men but not white men. This may explain why there is a lot of race mixture in poorer areas of Brazil, even if not in formal marital unions.

8. Conclusions

There have been several ways of understanding interracial marriage in Brazil: whitening, racial democracy, status exchange, and—more recently—darkening. This study provides evidence that migration should also be a factor in understanding the dynamics of interracial marriages. It also suggests that the classic history of nordestinos from predominantly brown and black areas of the country migrating to the whiter regions seems to play a role in rates of intermarriage in Brazil.

Migration and interracial marriage are social phenomena that both occur among significant segments of the Brazilian population. However, analysis of them together has been surprisingly absent or taken for granted in the understanding of race mixture in Brazil. This is despite the important role that migration has played in intermarriage since the colonial period. Whether it took place by force, as with the Portuguese enslaving and conquering Africans and people indigenous to South America or through nineteenth century policies of whitening the nation, migration has been central to Brazil’s race-mixing. Brazil’s large-scale internal migration over the last several decades punctuates this link as not solely an international but also a domestic phenomenon.

Migration often forces people to create new social ties, providing opportunity for more racially heterogeneous networks. Furthermore, distance from a community of origin can aid in the maintenance of an interracial marriage by providing freedom from the social norms of influential peer groups, including families of origin. Although both dynamics have been established as linking migration and intermarriage in the United States context (Rosenfeld and Kim 2005), further research is necessary to definitively unpack the causal direction in the Brazilian case. Moreover, there may be underlying personal characteristics, such as a sense of adventure, that may link migration and interracial marriage. For example, white Brazilian women who intermarried described themselves as negra frustradas who pursued relationships with black men to satiate their desires for blackness (Osuji 2019). These sentiments may lead them to migrate in order to achieve those goals in a way that may or may not be true for their spouses or opposite race/gender pairings.

The effect of migration may occur differently at international, regional, state, metropolitan, and even neighborhood levels. Data with more measures of migration is needed to disentangle these potential differences as well as examine them for different racial categories. For example, William’s study of sex tourism in Brazil shows how many foreigners travel there in search of interactions with Afro-Brazilian women (Williams 2013). However, these interactions can go beyond “ambiguous entanglements” and turn into marital unions, involving international migration. While this is beyond the scope of this paper, future studies should examine the relationship between levels of migration, including international, domestic, and regional factors as well as the intersections of spouses’ race and gender combinations. This will reveal the extent to which types of migration produce forms of marital unions, including other nontraditional unions, such as same-sex couples.

There is a significant discrepancy between the analytical concepts of race and skin color in Brazil (Monk 2016). In addition, an increase in wealth whitens Brazilians from brown to white while education seems to darken them from brown to black (Telles and Paschel 2014). For these reasons, people of vastly different skin colors may share a racial category, leading everyday Brazilians to see intermarriage where it would not be captured by official data. More research, both quantitative and qualitative is necessary to understand whether census understandings of what constitutes an interracial couple align with how Brazilians understand them. There is already qualitative evidence of the reverse, in which couples who are similar in skin tone do not share a racial category (and are seen by outsiders as an interracial couple) (Osuji 2019). Future studies should detangle the role of skin color versus racial categories in understanding intermarriage writ large as well as the role of migration in the process. A dyadic approach would be necessary to understand these subtleties of inter-color marital unions.

Furthermore, in a context of global white supremacy, migration may provide opportunities for other types of interracial marital unions. Many places in the Global South have experienced great deals of displacement, whether due to climate change, war, or neoliberal capitalism (Oliveira and Pereda 2020). In these contexts, in which people are having increased contact across ethnic and racial lines, we should expect more intermarriage. However, the United States is experiencing a “Reverse Great Migration” of African Americans, in which many northerners are migrating to the south, home of many of their ancestors. This may provide opportunities for more intermarriage. On the other hand, since the destination is largely the south, where interracial marriage has met the most resistance, the rates of intermarriage may slow for Black Americans. Future research should untangle the extent to which internal migration affects intermarriage across the world.