Older immigrants are more likely to share residence with their adult children and other family members than are U.S.-born older adults. Because socioeconomic factors only partially explain these differences and direct measures of cultural preferences are seldom available, the persistently high rates of intergenerational coresidence among the older foreign-born are often interpreted as driven by cultural preferences and/or a lack of assimilation. To challenge this interpretation, this study investigates the extent to which older immigrants’ living arrangements deviate from those of older adults in their home countries. The analysis combines data on immigrants from the 2008–2012 American Community Survey (ACS) with census data from three major immigrant-sending countries: Mexico, the Dominican Republic, and Vietnam. Despite persistent differences from U.S.-born whites, coresidence in later life is significantly less common than in the sending countries among the older foreign-born who migrated as young adults, and especially among those who migrated as children. The older foreign-born who migrated after age 50, however, are more likely to coreside and less likely to live independently than the older adults in their home countries. The similarity of these patterns across the three immigrant subgroups suggests that the unusually high coresidence among late-life immigrants is driven by U.S. family reunification policy and not simply by cultural influences.
Compared with U.S.-born older adults, older immigrants are substantially more likely to live in complex households with grown children and other relatives and nonrelatives (Angel et al. 1996; Glick and Van Hook 2002; Kaida and Boyd 2011; Wilmoth 2001; Wilmoth et al. 1997). Two major explanations for these coresidential arrangements have been cited: (1) immigrants’ cultural preferences for extended family living; and (2) their socioeconomic disadvantage, which requires pooling resources and precludes independent living. Socioeconomic disadvantage of older foreign-born individuals in terms of education, income, and especially wealth are well documented (Batalova 2012; Treas and Batalova 2009), but direct measures of cultural preferences are usually not available in census data and large-scale surveys often used to analyze older immigrants living arrangements. In the absence of such measures, the persistent differences in living arrangements between older foreign-born and U.S.-born Americans that remain after accounting for socioeconomic factors have been interpreted as driven by cultural preferences and/or a lack of assimilation to U.S. family norms. Although researchers recognize that international immigration may affect kin availability and influence household formation, none of the previous studies considered how older immigrants’ living arrangements may be affected by availability of adult children, which is closely linked to the U.S. immigration policy regulating admission of family members.
Understanding the within-group variations and the relative role of cultural and structural factors in determining living arrangements of older immigrants has important implications for social theory and public policy. Living arrangements are consequential for older adults’ health and well-being (e.g., Glick 1999; Keene and Batson 2010; Lawton et al. 1984; Magaziner et al. 1988; Speare and Avery 1993). Viewing coresidence as driven primarily by cultural preferences underestimates older immigrants’ need for community programs and services. Treating all foreign-born as a relatively homogeneous group in terms of socioeconomic resources and degree of assimilation is less cost-effective than providing targeted services to those older immigrants who need them most. Given that one in every five older adults in the United States soon will be foreign-born (Treas and Batalova 2009), understanding the patterns of living arrangements of older immigrants is crucial to “improve the health, function, and quality of life of older adults,” one of the new objectives for health promotion set by the U.S. Department of Health and Human Services (Office of Disease Prevention and Health Promotion n.d.).
This article makes three unique contributions. First, using a multinational approach, we compare the living arrangements of older immigrants in the United States with those of older adults in their country of origin. The extent to which living arrangements of older immigrants deviate from those of older adults in their home countries provides a better benchmark for evaluating preferences for living arrangements among older immigrants and thus allows for a better estimate of immigrant assimilation or lack thereof. Second, recognizing that immigrant assimilation into American society depends not only on the time spent in the United States but also on the opportunities for acculturation and socioeconomic incorporation (Treas 2014), we highlight the substantial differences in living arrangements of older immigrants by the age at migration. Finally, we show that the patterns of living arrangements of the older foreign-born who migrated in later life are unique because they are likely to be influenced by the family reunification policy of U.S. immigration laws.
This article combines data on older immigrants from the 2008–2012 American Community Survey (ACS) (Ruggles et al. 2015) with census data from three major immigrant-sending countries: Mexico, the Dominican Republic, and Vietnam (Minnesota Population Center 2015). The results show a clear pattern of assimilation within the first generation: rates of coresidence in later life are considerably lower among the older foreign-born who migrated as young adults (or especially as children) compared with older immigrants who migrated after age 50. Most importantly, the foreign-born who migrated after age 50 are more likely to coreside and less likely to live independently than older adults in their home countries. The similarity of the patterns across older immigrants from three different countries suggests that the unusually high coresidence among late-life migrants is likely to be driven by the unique family reunification policies in the United States.
Some benefits of coresidence for the well-being of older adults are well known. Living alone, with a spouse/partner, or with an adult child is indicative of the availability of instrumental help and psychological support (Hays 2002; Teerawichitchainan et al. 2015; Wilmoth and Chen 2003). Living with others is conducive to early diagnosis and management of health conditions because family members notice symptoms of disease and help with the health care arrangements (Han et al. 2013). Living with a spouse is linked to better dietary quality, especially for men (Davis et al. 2000). Family members are usually the primary source of long-term care (Brandt et al. 2009; McGarry 1998). However, residence in large multigenerational households is also associated with a number of negative outcomes, such as poor mental health, residential crowding, and intergenerational conflicts (Burr et al. 2010; White and Rogers 1997). It may also indicate economic or social vulnerability of family members who cannot live independently.
Previous research has found that the patterns of living arrangements among older adults are often determined by cultural and economic factors (e.g., Angel and Tienda 1982; Kaida et al. 2009). Financial difficulties of either adult children or older parents make the intergenerational coresidence economically advantageous because of income sharing and the exchange of unpaid domestic services, such as cooking, cleaning, and caregiving. Individuals with low personal income are more likely to coreside, and coresidence usually spikes during economic recessions (Bitler and Hoynes 2013; Mykyta and Macartney 2011). Beliefs that family members are responsible for helping each other also encourage intergenerational coresidence in later life, but most U.S. adults prefer to live close by rather than share a household with their parents or adult children (Compton and Pollak 2009; Ruggles 2007; Seltzer et al. 2012). Coresidence is more likely when a family member is poor or needs instrumental help due to declining health or presence of young children, but it is also conditional on other factors, such as gender, availability of other potential helpers (spouse, siblings), and geographical separation (Aquilino 1990; Choi 2003; Coleman and Ganong 2008; Keene and Batson 2010; Silverstein et al. 2006; Ward et al. 1992).
Immigrants in the United States in general, and older immigrants in particular, are more likely to share residence with their extended family members than are U.S.-born Americans (Angel et al. 1996; Glick and Van Hook 2002; Kaida and Boyd 2011; Wilmoth 2001; Wilmoth et al. 1997). Similar patterns have been found in other immigrant-receiving countries, such as Canada and Israel (Boyd 1991; Burr et al. 2012; Kaida et al. 2009). The nativity difference becomes smaller after socioeconomic factors, such as education and income, are taken into account. This finding is consistent with the argument that intergenerational coresidence among immigrants is driven by economic necessity. But because coresidence rates are also higher in China, India, Mexico, Latin America, and other immigrant-sending countries and regions than in the United States or Canada (Bongaarts and Zimmer 2002), the tendency of the foreign-born to share a household with extended kin is also attributed to familistic values and a desire to follow the cultural norms of their home countries (Diwan et al. 2011; Gonzales 2007). Within the foreign-born population, those who spent more years in the United States tend to be more similar to the U.S.-born in terms of their living arrangements, but persistent differences have led researchers to conclude that the foreign-born stick with their cultural norms of old-age support even after spending many years in the United States (Burr and Mutchler 1993; Burr et al. 2010; Glick 2000; Gurak and Kritz 2010).
The explanations for older immigrant living arrangements based primarily on socioeconomic factors and the differences from the U.S.-born Americans may be incomplete. First, assimilation refers not only to the degree of similarity to the U.S.-born but also to the degree of difference from the population in their home countries. Most previous studies, however, compared the living arrangements of the foreign-born only with those of U.S.-born Americans (but see Van Hook and Glick 2007). Even if the foreign-born do not quite resemble the U.S.-born, it does not follow that they are necessarily similar to people in their home countries. Second, although time in the United States is a popular measure of assimilation, it may not adequately capture older immigrants’ experiences. Immigrants face different sets of opportunities and constraints depending on their age at migration, which is especially important for understanding the variations in cultural preferences and socioeconomic outcomes of older foreign-born individuals. Finally, previous research typically neglects the role of U.S. immigration policy in shaping living arrangements of older immigrants.
Higher intergenerational coresidence rates in most immigrant-sending countries (Bongaarts and Zimmer 2002; De Vos et al. 2004; Guilmoto and de Loenzien 2015; Sereny 2011) seem to indicate that the persistent differences in living arrangements between older foreign-born and U.S.-born Americans are driven by cultural preferences. However, because of methodological differences between country-specific studies and to the scarcity of cross-national studies, the extent to which older immigrants’ living arrangements in the United States are similar or different from those in their home countries is largely unknown. This study improves on previous research by providing such estimates for older immigrants from three major immigrant-sending countries: Mexico, the Dominican Republic, and Vietnam.
Although the choice of the immigrant subgroups was affected by data availability,1 important differences in the history of migration and patterns of living arrangement in Mexico, the Dominican Republic, and Vietnam allow for strategic comparisons. Older Mexicans make up the largest and the most studied immigrant subgroup (Angel et al. 2000; Glick 1999; Gonzales 2007; Wilmoth 2001). As of 2014, Mexicans accounted for approximately 15 % of the older immigrant population in the United States. Approximately one-quarter of Mexicans who are now aged 60 and older came to the United States between the late 1940s and mid-1960s, and another one-third migrated during the late 1960s and 1970s (authors’ calculations based on the 2014 ACS; also see Table 5 in the appendix). Most of these migrants had low levels of education and sought employment in agriculture and other low-skilled occupations (Hanson and McIntosh 2010). A large population and a relatively long history of migration allow for multiple comparisons within the older Mexican-born population.
Dominicans and Vietnamese account for 2.8 % and 1.2 %, respectively, of the foreign-born aged 60 and older (authors’ calculations based on the 2014 ACS). Close to 40 % of older Dominicans also migrated to the United States in the 1960s and 1970s (authors’ calculations based on the 2014 ACS), some to escape political turmoil and some attracted by employment in manufacturing (Grasmuck and Pessar 1991). Like Mexicans, most Dominican immigrants had low levels of education (see Table 5 in the appendix). A distinct characteristic of Dominican immigration is a relatively large share of female migrants and high prevalence of female-headed households (Gurak and Kritz 1996; Soy and Bosworth 2008). Immigration from Vietnam is relatively recent: one-half of the older Vietnamese in the United States were admitted as refugees in the late 1970s and 1980s, after the Vietnam War (Zhou and Bankston 1998). Although many Vietnamese immigrants are poorly educated, approximately 15 % hold college and professional degrees (see Table 5 in the appendix). Despite their much smaller population size and shorter history of migration, they still rank among the largest origin groups of older immigrants.
Despite inevitable family separation by war and refugee experience, Vietnamese immigrants in the United States faced relatively few barriers to family reunification because of their refugee status and high naturalization rates (Woodrow-Lafield et al. 2004). In contrast, many Mexicans and a nontrivial number of Dominicans were undocumented. Although a substantial number of those who migrated in the 1960s and 1970s were able to obtain a lawful permanent residency status through the Immigration Reform and Control Act (IRCA) of 1986, Mexicans and Dominicans are the least likely to naturalize (Woodrow-Lafield et al. 2004), which limits their ability to bring relatives (such as adult children or aging parents) to the United States (Enchuaegui 2013).
Old-age coresidence rates are generally higher in South Asian countries than in Latin America and the Caribbean (Bongaarts and Zimmer 2002; Guilmoto and de Loenzien 2015). Among the latter, coresidence is more common in the Dominican Republic than in Mexico (De Vos 1990). Female-headed households, however, are more prevalent in the Dominican Republic (Rogers 1995). Similar differences have been found among older immigrants from these countries living in the United States (Angel et al. 1996; Burr and Mutchler 1993; Gurak and Kritz 2010; Wilmoth 2001). Examining how living arrangements of older immigrants from Mexico, the Dominican Republic, and Vietnam differ from those of older adults in these three home countries would provide insights into shared and country-specific factors shaping the living arrangements of older immigrants in the United States.
Age at Migration
The degrees of acculturation, economic integration, and kin availability are among the most important factors determining how similar or different older immigrants’ living arrangements are from those of their counterparts in home countries. Because these factors are often difficult or impossible to measure directly from available data, a fruitful approach is to identify subgroups of the older foreign-born who are apt to be differentially affected by them. Age at migration is a productive measure given its implications for economic integration, cultural preferences, and kin availability.
Previous research on older immigrants has emphasized the importance of age at migration for understanding immigrant outcomes in later life (Angel et al. 1999, 2000; Blank and Torrecilha 1998; Treas and Mazumdar 2002). In addition to the length of exposure to U.S. society, age at migration captures opportunities for acculturation and socioeconomic incorporation, which generally decline with age (Treas 2014; Treas and Gubernskaya 2016). Because the speed or probability of assimilation depends on the age at migration, a given number of years in the country may not translate into equivalent acculturation or socioeconomic incorporation for someone who migrated as a young person versus an older adult. For example, younger immigrants are likely to benefit from U.S.-based education, have more opportunities to learn English, develop more diverse social networks, and work long enough in the United States to be eligible for Social Security when they reach the retirement age. In contrast, later-life migrants are often not eligible for Social Security because they lack necessary 10 years of eligible employment. They also face substantial barriers to employment because of the combination of advanced age, poor English language proficiency, foreign degrees and credentials, and declining health. Research has consistently confirmed that older age at migration is related to lower income (O’Neil and Tienda 2014), poor and intermittent health insurance coverage (Choi 2006; Reyes and Hardy 2015), and steeper health decline in later life (Gubernskaya 2015).
Although age at migration is a continuous variable, theoretical considerations and previous research suggest that it is meaningful to think of it as a set of broader categories (Myers et al. 2009; Treas 2014). For the purposes of this study, we distinguish the foreign-born who (1) migrated as children (most often brought by their adult migrant parents), (2) migrated independently as adults, or (3) migrated in later adulthood (most often invited by an adult child migrant). Those older foreign-born who migrated as children are likely to be the most assimilated not only because they spent more years in the United States but also because they had more opportunities for linguistic, cultural, and structural assimilation via U.S.-based schooling and employment. Child immigrants have been called the “1.5 generation,” whose assimilation falls between the first generation immigrating as adults and second generation of U.S.-born children of immigrants (Rumbaut 2004). Although immigrant children’s integration benefits from U.S. schools, adult immigrants have workplace opportunities for assimilation. In contrast, lacking U.S. schooling and having limited ties to the American workforce, late-life migrants are at a disadvantage. For adults who migrate in later life, decades spent in the United States may never translate into the same level of acculturation, especially structural assimilation, as enjoyed by peers who migrated when they were younger.
Family Reunification Policy
According to the family reunification provision of the 1965 Immigration and Nationality Act (Pub.L. 89–236), a naturalized U.S. citizen aged 21 and older may sponsor the immigration of parent(s) with only a signed affidavit of support. As immediate relatives, parents of U.S. citizens are admitted outside numerical quotas that limit migration of siblings and adult children and immediate relatives of lawful permanent residents. To file an affidavit of support, one’s household income has to be at or above 125 % of the poverty level for a given household size (U.S. Citizenship and Immigration Services 2016). One can add a cosponsor to the affidavit if the income requirements are not met. Each year, 80,000 to 120,000 middle-aged or older parents become lawful permanent residents through this route (U.S. Department of Homeland Security 2016).2
The implications of this policy for the living arrangements of immigrants cannot be overstated. Most foreign-born individuals who immigrate as parents of U.S. citizens come to the United States at an advanced age, when it is difficult to adjust to a new culture and to learn a new language (Kim et al. 2011; Stevens 1999). Coming from less-developed countries with weak pension systems, older foreign-born migrants typically have inadequate retirement savings. Unsurprisingly, they become dependent on their adult children for both financial and social support (Angel et al. 1999; Treas and Mazumdar 2002). Depending on geographic location, older immigrant parents may find friends and sometimes employment in immigrant communities. However, even relatively educated later-life migrants are unlikely to participate in the mainstream formal economy because of advanced age, linguistic barriers, and lack of cultural competence. Older immigrants are, nonetheless, important members of their families, if only because they perform essential unpaid tasks such as cooking, housekeeping, and childcare (Treas and Mazumdar 2004). They serve vital social roles in kin-keeping, providing cultural linkages across generations, and supporting bilingualism of U.S.-born grandchildren. Lacking financial independence, responsive to household needs, and often supported by familistic norms (Treas and Mazumdar 2002), older newcomers are likely to join their adult children’s households for extended periods. This dynamic should be reflected in high rates of coresidence and low rates of independent living among those immigrating in later life.
Family reunification policy also relates to the availability of adult children with whom older immigrants can potentially coreside. The policy is selective of older immigrant parents with adult child(ren) in the United States who are willing to sponsor their immigration. U.S.-born Americans and those foreign-born who immigrated at younger ages and grew old in the United States may or may not have surviving adult children, or their children may live outside the United States. Furthermore, their adult children may or may not be willing to coreside with parents. However, the immigration of parents is conditional on the availability of adult children willing to support them in the United States. This aspect of the law works to increase coresidence rates among late-life immigrants compared with other older immigrants and also compared with nonimmigrants in their home countries who face similar restrictions on availability of adult children.
Older foreign-born who migrated as children or young adults have lived most of their lives in the United States. Given their opportunities for acculturation and socioeconomic incorporation through schooling and employment, as well as longer exposure to the U.S. society, older foreign-born individuals who migrated as children and adults will be the more assimilated and thus more likely to have living arrangements that resemble those of native-born Americans. Indeed, most of these older immigrants have lived independently throughout their adulthood. Similar to native-born individuals and nonmigrants in their home countries, after they reach old age and experience health declines, they continue to live independently, move in with adult children or other relatives, or transition to an institution. Whether because of the familistic cultural orientation and/or the willingness of their adult children, older immigrants are more likely to coreside than are older U.S.-born Americans. However, because of acculturation and socioeconomic incorporation, we expect the following:
Hypothesis 1: Older foreign-born individuals who migrated as children and young adults will be less likely to coreside (that is, more likely to live independently) than older adults in their home country.
Older foreign-born adults who migrated after age 50 come primarily through the family reunification provision of U.S. immigration policy (Carr and Tienda 2013). Besides being less acculturated, they have likely never lived independently in the United States. However, unlike those who migrated at younger ages or nonmigrants in their home countries, late-life migrants almost certainly have an adult child to provide assistance for at least some time after migration. Thus, we expect the following:
Hypothesis 2: Older foreign-born who migrated after age 50 will be more likely to live with their adult children (that is, less likely to live independently) compared with older adults in their home countries.
Although the differences in the patterns of living arrangements in the sending countries, the history of migration, and barriers to family reunification may explain the variations in living arrangements among the older foreign-born by country of origin, age at migration has similar implications for assimilation and availability of adult children in the United States, regardless of origin. Older Dominicans are more likely to be unpartnered (Gurak and Kritz 2010), but it is unclear whether this would influence living arrangements beyond lowering the share of married-couple households. Undocumented status, deportations, long waits related to adjustment of immigration status, and low naturalization rates coupled with limits on family reunification for noncitizens are likely to contribute to family separation and household structure of Mexican immigrants in the United States. These factors will work to increase the rates of independent living or the probability of residence in complex households for the adult and child migrants (e.g., if the Mexico-born adult children live in Mexico because they were deported or because adult undocumented migrants were not able to bring their parents to the United States). Although barriers to family reunification would affect the ability of adult Mexican migrants to bring their parents to the United States and the overall rates of later-life migration from Mexico, they are unlikely to differentially affect the living arrangements of those older immigrants who were able to reunite with their adult children. We expect that the differences in living arrangements by age at migration and country of residence will be similar for the older foreign-born from Mexico, the Dominican Republic, and Vietnam.
Data and Method
We combine data from the 2008–2012 American Community Survey (ACS) available through IPUMS-USA (Ruggles et al. 2015) with data from the 2010 Mexico census, the 2010 Dominican Republic census, and the 2009 Vietnam census available through IPMUS-International (Minnesota Population Center 2015) to compare the patterns of living arrangements of older Mexicans, Dominican, and Vietnamese immigrants in the United States with those of older U.S.-born non-Hispanic whites and those of older adults in their sending countries. We chose these countries based on the size of the older immigrant population in the United States and the availability of data collected at approximately the same time, between 2008 and 2012. The analytical sample consists of adults aged 60 and older. Because the sample size for U.S.-born non-Hispanic whites and the national census subsamples are very large compared with the immigrant subsamples, we drew a random sample of U.S.-born older adults and older adults in the sending countries to achieve a sample size of approximately one-half the sample size of older immigrants from each country.
Because living arrangements are partially determined by one’s marital status (Treas and Sanabria 2016; Wilmoth 2001), we analyze currently partnered (married or cohabiting) and unpartnered (never married, divorced, separated, or widowed) older adults separately. The living arrangements are classified into three categories: living alone (= 0, baseline comparison group for unpartnered respondents) or living with a spouse/partner only (= 0, baseline comparison group for partnered respondents); living with adult children/children-in-law (= 1); and “other” living arrangements (= 2), which includes all other household types. The variables are constructed from the household size and household rosters variables. All adults aged 60 and older in household size of 1 are coded as living alone. In two-person households, those who were household heads living with the spouse or unmarried partner, and those who were a spouse or a partner of the household head, are coded as “living with spouse only.” We define adult children as individuals aged 25 and older3 who share a household with at least one parent or parent-in-law. Thus, the “live with adult children” category includes the parents/parents-in-law of the household heads aged 25 and older and the heads or spouses of the heads of households that include at least one child of the householder aged 25 and older. Older persons classified into this category have at least one child (in-law) aged 25 or older in the household, but these households can be of any size (2 or more) and may also include a spouse, other relatives, and nonrelatives.4 Finally, all older adults who are not in one of the first three living arrangements fall into the residual “other” category.5
We begin by estimating weighted percentages of adults aged 60 and older in different types of living arrangements among the U.S.-born whites: older immigrants from Mexico, the Dominican Republic, and Vietnam residing in the United States; and older Mexicans, Dominicans, and Vietnamese residing in those countries. We use the place of birth variable to identify the immigrants from Mexico, the Dominican Republic, and Vietnam in the U.S. data. The year of migration to the United States was used to classify the older foreign-born in the United States into three groups: those who migrated as children (aged 0–17), those who migrated as adults (aged 18–49), and those who migrated in old age (aged 50 and older). We also include the U.S.-born Mexicans as an additional category for the Mexican subsample. The birthplace, race and Hispanic origin variables were used to identify U.S.-born non-Hispanic whites and U.S.-born Mexicans. Unfortunately, we could not include the U.S.-born older Dominicans and Vietnamese because of the very small size of these subpopulations and, consequently, inadequate sample size in the data. For the same reason, a very small number of older Dominicans and Vietnamese who migrated as children were excluded from the analysis.
We construct multinomial logistic regression models on the pooled data from the United States and the national censuses to test our hypotheses about the relationship between the age at migration and living arrangements in older age. Besides the nativity/age at migration categories, we include several variables to account for the sociodemographic differences between the subgroups. Age is a continuous variable centered on 65. Females are coded as 1, with males being the reference category. We use a general educational attainment variable to construct a categorical variable capturing level of education. Although it does not reflect the various definitions of levels of schooling or each country’s system of educational tracks, this variable is largely comparable across countries (Minnesota Population Center 2015). After checking the country-specific education classifications and the source variables for Mexico, the Dominican Republic, and Vietnam, we chose the following three categories: incomplete secondary or less than high school (reference), high school, and college. In addition, because we pool the U.S. data across multiple years, we control for the year of the survey with the census year of the baseline comparison group as the reference category (e.g., 2010 for Mexicans in Mexico because the Mexican census data is from 2010). We run similar models for the constructed subsamples of older Mexican, Dominican, and Vietnamese immigrants and nonimmigrants.
Table 1 presents the weighted percentage distributions of living arrangements by partnership status, nativity/age at migration, and country of residence. Among currently partnered U.S.-born white adults aged 60 and older, approximately 84 % live with a spouse only, compared with 32.7 % of older adults in Mexico, 22.8 % in the Dominican Republic, and 30.7 % in Vietnam. Similarly, coresidence rates are relatively low among U.S.-born whites compared with older adults in Mexico, the Dominican Republic, and Vietnam (9.7 % vs. 48.9 %, 44.9 %, and 54.1 %, respectively). “Other” living arrangements are also more common in the immigrant-sending countries. Approximately 18.3 % of Mexicans, 32.3 % of Dominicans, and 15.2 % of Vietnamese aged 60 and older live neither with the spouse only nor with the adult children, compared with only 6.3 % of U.S.-born whites. Among older adults without a partner, approximately 69 % of U.S.-born whites live alone, compared with only 27.4 % of Mexicans, 23.8 % of Dominicans, and 23.4 % of Vietnamese in their respective countries. Coresidence rates are higher among older U.S.-born whites without a partner (17.6 %), but they do not match still higher rates in Mexico (51.6 %), the Dominican Republic (45.7 %), and Vietnam (65.9 %). There are also interesting differences in the proportion of older unpartnered adults who live in “other” household types. The lowest proportion is in Vietnam (10.7 %), followed by U.S.-born whites (13.2 %) and Mexicans (21 %), with the highest rates among Dominicans (30.5 %).
Despite some convergence, the living arrangements of older immigrants from Mexico, the Dominican Republic, and Vietnam more closely resemble those of older adults in their home countries than those of older U.S.-born non-Hispanic whites. Compared with Mexicans in Mexico, partnered Mexican-born in the United States are more likely to live with a spouse only (38.4 % vs. 32.7 %), less likely to coreside with adult children (45.5 % vs. 48.9 %), and less likely to live in “other” types of households (16.1 % vs. 18.3 %). The differences are somewhat larger among older Dominican and Vietnamese immigrants, with a few exceptions. There is nearly no difference in the proportion of living in “other” households between Vietnamese immigrants and nonimmigrants. Older Dominican immigrants in the United States are less likely to live in these types of households than are their counterparts in the Dominican Republic (21.5 % vs. 32.3 %). Older Mexico- and Vietnam-born individuals in the United States who do not have a spouse are even less likely to live alone or coreside but are more likely to live in “other” households compared with their counterparts in Mexico and Vietnam. The pattern is very similar for Dominicans, although nonpartnered Dominicans in the United States are more likely to live alone (30.6 % vs. 23.8 %).
Table 1 also shows the substantial differences in the patterns of living arrangements among the older foreign-born by nativity and age at migration. The differences are especially pronounced when it comes to coresidence. Those currently partnered Mexicans who migrated to the United States as children are less likely to share residence with their adult children compared with those who migrated as young adults, and especially compared with those who migrated after age 50 (36.3 % vs. 44.3 % vs. 59.5 %). Late-life migrants stand out because of their very high coresidence rates compared not only with other immigrants but also with older adults in their home countries. Although 59.5 % of partnered late-life migrants from Mexico live with their adult children, only 48.9 % of their counterparts in Mexico do. Approximately 53.4 % of partnered late-life migrants from the Dominican Republic coreside with their adult children, compared with only 44.9 % of Dominicans in the Dominican Republic. Approximately 59.5 % of older partnered Vietnamese late-life migrants in the United States coreside, compared with 54.1 % of older Vietnamese in Vietnam. The differences are even larger among older unpartnered adults. Only 40.1 % of child migrants from Mexico coreside, compared with 47.4 % of adult migrants and 66.4 % of late life migrants. The patterns are similar among older Dominicans and Vietnamese.
The differences in the descriptive percentages presented in Table 1 may be affected by the compositional differences related to age, sex, and education. As Table 5 in the appendix shows, late-life migrants tend to be older, are more likely to be female, and have lower levels of education compared with those who migrated as children or adults. Tables 2, 3, and 4 present the odds ratios from the multinomial logistic regression models by partnership status for Mexican, Dominican, and Vietnamese subsamples, respectively. For partnered, the reference category in the dependent variable is “living with the spouse only.” For persons who are not married or cohabiting, the reference category in the dependent variable is “living alone.” Older adults in the immigrant-sending country are the reference category of the main independent variable—the combined measure of nativity, country of residence, and the age at migration.
Table 2 presents the results for the Mexican subsample. Controlling for age, sex, and education, partnered older U.S.-born whites have 87.4 % lower odds of living with adult children relative to living with the spouse only compared with partnered older Mexicans in Mexico. Partnered U.S.-born Mexicans have 65 % lower odds of coresiding with adult children relative to living with spouse only than married Mexicans in Mexico. Those who migrated before age 18 have 41 % lower odds, and those who migrated as adults have 17.4 % lower odds of coresiding with adult children relative to living with a spouse only than older adults in Mexico. But those who migrated after age 50 have 69.6 % higher odds of living with adult children relative to living with a spouse only than Mexicans in Mexico. Compared with their counterparts in Mexico, unmarried non-Hispanic whites in the United States and U.S.-born Mexicans are 81.8 % and 52.8 % less likely to coreside than to live alone, respectively. For those who migrated before age 18, the odds of coresidence are 31.2 % lower. Those who migrated at age 18–49 are as likely to live with adult children, but later-life migrants are more than three times as likely to coreside as Mexicans without a partner in Mexico.
Females are less likely to live with adult children or in “other” households than to live with a spouse if they have a partner, but they are more likely to coreside rather than live alone if they do not have a partner. The better-educated, with or without a partner, are more likely to live independently than to share a household with their children or other relatives. Older age is associated with lower odds of living with children or in “other” types of households relative to living with a spouse only among partnered older adults. Among those who do not have a spouse or a partner, older age is associated with higher odds of coresiding and lower odds of living in “other” households compared with living alone.
Table 3 presents the results for the Dominican subsample. Net of the other predictors, Dominican-born individuals who migrated as adults have 31 % lower odds of living with adult children and 63.2 % lower odds of living in “other” households relative to living with a spouse only in comparison with older adults in the Dominican Republic. In contrast, partnered older Dominicans who migrated after age 50 are 42.1 % more likely than older adults in their home country to live with adult children. Among the unpartnered, the odds of living with adult children relative to living alone are 63.1 % higher for the later-life Dominican immigrants compared with their counterparts in the Dominican Republic.
Table 4 presents the results for the Vietnamese subsample. Controlling for age, sex, and education, older Vietnamese immigrants who are married and migrated to the United States as adults have 42.5 % lower odds of coresiding than living with a spouse only compared with their peers in Vietnam. However, those Vietnamese immigrants who came to the United States after age 50 have 25 % higher odds of living with their adult children compared with married older adults in Vietnam. Unmarried adult migrants are as likely to coreside as their counterparts in Vietnam compared with living alone. However, later-life immigrants from Vietnam who do not have a partner have 92.4 % higher odds of coresiding and 2.7 times higher odds of living in “other” households relative to living alone compared with their counterparts in Vietnam.
Figures 1, 2, and 3 translate the results from Tables 2, 3, and 4, respectively, into predicted probabilities of living in specific type of households adjusted for age, sex, and education. Supporting Hypothesis 1, Fig. 1 shows clearly the gradient in the patterns of living arrangements among older Mexican immigrants. The living arrangements of older adults who were born in Mexico and migrated as adults, those who migrated as children, and U.S.-born Mexicans are increasingly different from those of older Mexicans in Mexico. However, Fig. 1 also supports Hypothesis 2 by showing that even after we adjust for age, sex, and education, late-life migrants from Mexico have much higher coresidence rates and much lower rates of independent living than their counterparts in Mexico.
Figures 2 and 3 show predicted probabilities of living in specific household types for Dominicans and Vietnamese, respectively. Although the differences in the predicted probabilities are not always statistically significant (which could be due to much smaller sample sizes), the general patterns are similar, providing additional support for Hypotheses 1 and 2. Late-life migrants have higher coresidence rates and lower rates of independent living than do adult immigrants and to nonmigrants in their home countries. There are also some interesting differences by country of origin. Both partnered and unpartnered late-life migrants from the Dominican Republic are less likely to live in “other” households than their nonimmigrant counterparts. In contrast, late-life migrants from Mexico and Vietnam are slightly less likely to live in “other” households only if they have a partner; if not partnered, they are more likely to live in “other” households than their counterparts in Mexico and Vietnam. Admittedly, “other” households are much more prevalent in the Dominican Republic compared with Mexico or Vietnam.
Discussion and Conclusion
Previous research typically explained high coresidence among older immigrants in terms of socioeconomic disadvantage and cultural preferences (e.g., Angel and Tienda 1982; Kaida et al. 2009). The cultural preference explanation was sustained by references to the unexplained residual after socioeconomic characteristics were taken into account and/or proxies (such as time spent in the United States) for assimilation to American residential norms. But more recent binational studies questioned this approach (Van Hook and Glick 2007; Riosmena et al. 2013). By considering the residential patterns of older adults in several sending countries and emphasizing the differences in living arrangements among older immigrants in the United States by age at migration, we provide a better benchmark for evaluating preferences for living arrangements among older immigrants. This approach not only improves estimates of immigrant assimilation (or lack of thereof) but also shows the implications of immigration policies and kin availability for understanding older immigrants’ residential patterns.
Combining ACS data with censuses from Mexico, the Dominican Republic, and Vietnam, this article demonstrates that despite persistent difference from the U.S.-born whites, older immigrants who arrived at younger ages are considerably less likely to coreside, and are more likely to live independently, than older adults in their countries of origin. These patterns are consistent with the immigrant assimilation explanation predicting gradual convergence of the immigrants’ living arrangements toward those of the U.S.-born Americans. However, older adults who migrated after age 50 are significantly more likely to coreside, and less likely to live independently, than older adults in their home countries—a pattern that neither cultural preferences nor immigrant incorporation theories are sufficiently able to explain.
Remarkable similarities for three national origin groups in living arrangements patterns by age at migration suggest that the timing of migration in the life course is one of the strongest predictors of coresidence in older age. It also means that the context of the receiving country that immigrant groups share may be more important than cultural preferences. Older adults who migrate to the United States in later life usually do so through the family reunification provision of U.S. immigration law (Carr and Tienda 2013). Signing an affidavit of support, an adult citizen child can sponsor the immigration of a parent who, if only by virtue of age, is less likely to incorporate readily into U.S. society without family assistance. Age-related barriers to acculturation and socioeconomic assimilation hinder late-life migrants from establishing residential independence (Angel et al. 1999; O’Neil and Tienda 2014; Stevens 1999; Treas 2014). Being ineligible for public assistance programs during their first five years after migration only increases late life migrants’ dependence on support from their adult children, regardless of their country of origin. At the same time, family reunification assures the selection of older newcomers who have at least one adult child with whom they can live. Thus, high coresidence among later-life migrants is driven not only by socioeconomic and cultural factors but also by the kin availability built into the immigration policy.
Estimating the degree to which older immigrants’ living arrangements diverge from the patterns in their home countries is an important step toward understanding the relative importance of the socioeconomic, cultural, and other factors in determining older immigrants’ living arrangements. If the U.S.-born are the only comparison group, the persistent large differences in old-age coresidence rates, even after sociodemographic characteristics are controlled for, may be misinterpreted as driven primarily by cultural preferences and/or lack of assimilation. If age at migration is not taken into account, the overall difference from the U.S.-born for those older foreign-born individuals who migrated as children and young adults will be overestimated because of the unusually high coresidence rates among late-life migrants. Simultaneously considering the residential patterns of older adults in the sending countries and the differences in living arrangements among older immigrants in the United States by age at migration reveals a clear pattern of assimilation among children and young adult immigrants and unusually high coresidence rates among later-life migrants, pointing to a previously neglected factor: immigration policies. Thus, overemphasizing cultural factors behind old-age coresidence rates may contribute to stigmatization of the foreign-born for their lack of assimilation, but it may also lead to inaccurate assessment of older immigrants’ needs and preferences for nonfamily programs and services.
The study is not free of limitations. Although the results are suggestive, we are unable to test directly the impact of family reunification policies on older immigrants’ living arrangements because neither the mode of migration nor the presence of noncoresident adult children is available in the data. We assume that older adults who live alone or only with their spouse maintain a certain degree of economic independence, but some of them may still be heavily dependent on their adult children who live nearby. However, whether living in close proximity is more common in the immigrant-sending countries than among older immigrants in the United States remains an open question. Inconsistent measurement of income across the countries precluded us from including it in the models. Although education is a good proxy for socioeconomic resources, adjusting for income differences might further reduce the differences in living arrangements by age at migration. Despite remarkable similarities in patterns across the three immigrant subgroups, there are also interesting differences. For example, not-partnered Dominican immigrants in the United States are less likely to live in “other” households than their counterparts in the Dominican Republic; the opposite is true for Mexicans and Vietnamese. These findings may be related to a rather simplistic classification of living arrangements with the diverse residual “other” category used in this research. As with most previous studies, our data lack direct measures of attitudes or values, which does not allow a direct test of the cultural preferences behind the living arrangement patterns of older immigrants. Future research needs to address these limitations when better data become available.
The results have important policy implications. Differences in the rates of coresidence and independent living among older foreign-born individuals imply that there will also be differences in access to family support in old age. Those who migrated as children and young adults may not rely, or even want to rely, on their adult children to the same degree as those who migrated in later life (Diwan et al. 2011). This may be a problem if the immigrants were not able to accumulate enough resources to maintain independent living at older ages. Late-life migrants, on the other hand, may require substantial financial, social, and emotional assistance (Treas and Gubernskaya 2015). Even though adult children do not necessarily see coresidence with parents or parents-in-law as a burden, providing this kind of support may necessitate redistribution of family resources away from other members (e.g., young children) and likely increase risk of interpersonal tensions and intergenerational conflict (Burr et al. 2010; White and Rogers 1997). To be sure, older immigrants play important roles in their families by providing childcare, helping with housework, and maintaining kinship ties with extended family members (Treas and Mazumdar 2002, 2004). Still, later-life migrants often report poor physical and mental health (e.g., Diwan 2008; Gubernskaya 2015; Takeuchi et al. 2007), and many resent limited autonomy and dependence on their adult children’s families (Treas and Mazumdar 2002). These findings suggest that family support has its limits. Complex needs of older immigrants and their families may be better met through a combination of state programs, local community services, and family support. Better understanding of factors behind the patterns of living arrangements of older foreign-born individuals is critical for developing cost-effective, high-quality support programs for older immigrants and their families.
This research was partially supported by a seed grant from the Center for Social and Demographic Analysis (CSDA) at the University at Albany, SUNY, funded by the National Institute of Child Health and Human Development (R24-HD044943). We thank Judy Treas, Sonya Grover, the editors, and the three anonymous reviewers of Demography for their insightful comments. An earlier version of this research was presented as a poster at the 2016 Population Association of America meeting in Washington, DC.
For accurate comparison, the data sources have to be roughly similar. Unfortunately, we did not have access to census data collected around 2010 for other large, older immigrant subgroups.
This is approximately 12 % of the total foreign-born population obtaining lawful permanent resident (LPR) status; the number is comparable to the number of refugees obtaining LPR status each year and only slightly less than the number of employment-based immigrants.
We chose age 25 as a cutoff point to better capture the old age coresidence rather than the coresidence due to adult children’s delayed home-leaving. Alternative cutoff points (e.g., age 18) resulted in slightly different estimates of the overall coresidence rates, but the patterns by nativity and age at migration were similar.
The data coding was consistent across the data sets with a few exceptions. “Child” includes adopted children, stepchildren, and foster children as well as biological children in the U.S. and Mexico data; adopted children are not included in the Dominican data; and only biological children are included in the Vietnamese census data. Children-in-law could not be identified in the Vietnamese census.
This category includes households of different types. Partnered older adults in Mexico, the Dominican Republic, and Vietnam, as well as U.S. whites in this category, predominantly live with children under age 25. Among unpartnered older adults, the percentage of those who live with other relatives and nonrelatives of the householder is almost as high as the percentage living with younger children. The patterns are fairly similar across the three subgroups.