The Uphill Climb: A Transnational Perspective on Wealth Accumulation among Latino Immigrants in Durham, NC

Abstract

Wealth accumulation is a key dimension of ethno-racial stratification, and, among immigrants, an important indicator of incorporation. Dramatically low assets among immigrant Latinos is thus a pressing concern, necessitating a better understanding of the social forces that shape wealth assimilation. Drawing on a survey of Latino immigrants in Durham, NC, I argue for the importance of a transnational perspective on wealth for immigrant populations. Nationally representative surveys designed to assess inequality among the general population generally lack information on wealth held abroad, which accounts for the lion’s share of assets held by immigrants in our sample. Likewise, these data sources rarely have information on factors salient to immigrants, particularly legal status and informal employment. Finally, I show that the socio-demographic characteristics central to life-cycle wealth models operate in different ways for U.S. and foreign assets, and for men and women. For instance, while household earnings and duration of Durham residence are associated with greater U.S. assets among Durham’s Latino migrants, they fail to predict wealth held abroad. Likewise, low educational attainment and informal employment are associated with lower U.S., but not foreign, wealth. Instead, the key predictors of wealth abroad relate to family structure. I further document structural barriers to immigrant Latino wealth accumulation, such as employment marginality and lack of access to mainstream financial institutions.

Introduction

Immigrant Latinos numbered nearly 20 million in 2015, or six percent of the U.S. population. As such, the degree to which this population is able to successfully integrate into the U.S economy and society will have a profound impact on both ethno-racial stratification and the socio-economic landscape of the United States itself. It is thus not surprising that copious research addresses immigrant Latinos’ incorporation into U.S. labor and housing markets. However, research on financial incorporation and the process of immigrant asset accumulation remains limited.

Wealth has long been recognized as central to economic security. More stable than income, wealth can be used to meet both short- and long-term needs, and, along with income, is considered one of the “twin pillars” of middle class status (Oliver and Shapiro 1995). Wealth’s transferability, potential for return and use as collateral, and tax-favored status also contribute to its importance. Wealth is often described as transformative, giving families access to higher quality neighborhoods and education than would be attainable based on income alone. Indeed, the impact of parental wealth during childhood on later-life outcomes is substantial (Keister, Agius Vallejo and Borelli 2013), which along with massive disparities in inheritance, is a key mechanism through which inequality is reproduced across generations. As such, wealth is central to understanding systems of stratification (Killewald, Pfeffer, and Schachner 2017; Agius Vallejo and Keister 2019).

While much attention has focused on the concentration of wealth among elites, it is also critical to understanding economic security among low-skill and –income populations; even modest wealth contributes to living standards, and can help buffer against employment and health shocks (Valdez 2019). For immigrant Latinos, wealth is arguably even more important. Concentrated in low-skill industries, they simultaneously suffer from high rates of working poverty and employment insecurity and limited access to employer-sponsored health and retirement benefits (De Jong and Madamba, 2001; Flippen 2012, 2013). Their reliance on private savings is further heightened by their lack of eligibility for most public safety net provisions, such as unemployment compensation, the Earned Income Tax Credit (EITC), and federal retirement programs, rendering the ability to accumulate assets a key dimension of incorporation.

Unfortunately, information on immigrant wealth remains extremely limited. While new data sources have improved our ability to assess patterns of wealth inequality overall, traditional sources tend to capture immigrant wealth poorly. Much of the previous work describing Latino and immigrant wealth is drawn from large, nationally representative data sources not specifically formulated to understand migrant populations. Data is often lacking on assets held abroad, and when it is available, most analyses tend to add foreign wealth to total assets, rather than analyzing it separately. However, ample research documents that the social forces structuring wealth accumulation differ for different types of assets (i.e., housing versus financial wealth). It is also likely that for migrants, assets held in the United States and abroad operate in different ways as well. Because immigrant men and women differ in their migration histories, family structures, and labor market positions, wealth accumulation likely also varies by gender.

To address these issues, I draw on an original survey of immigrant Latinos in Durham, NC to examine the socio-demographic predictors of asset accumulation among recent migrants. The survey was specifically designed to capture both assets and debt held in receiving and sending societies, and to collect other characteristics specific to immigrant Latinos. I argue for the need for more transnational approaches to analyzing immigrant wealth, that not only include foreign assets, but also model wealth held abroad and in the United States separately. I also argue for the need to examine assets, debt, and net worth separately. Immigrants are limited in their ability to access the mainstream financial institutions that facilitate saving (Rhine and Greene 2006). But they also face significant barriers to access to conventional credit. Assessing these processes separately offers additional insight into Latino financial incorporation.

Theoretical Background

Ethno-racial gaps in wealth are staggering - far larger than comparable gaps in income and other labor market outcomes (Wolff 2002). Immigrant Latinos, in particular, stand out for remarkably low asset ownership relative to other groups (Hao 2007). These low wealth levels are especially troubling given the labor market vulnerability of this population (Grogger and Trejo 2002). Concentration in occupations and industries with chronic employment instability, low wages, and low benefit coverage, coupled with extremely limited access to social safety net provisions, heightens the importance of private savings for immigrant Latinos. Low levels of wealth thus represent a threat to economic security for low-skill and working-class Latinos, and even jeopardize the stability of the middle class (Wheary, Shapiro, and Meschede 2008).

Much of the immigrant Latino wealth disadvantage relative to native, non-Hispanic whites stems from life-cycle and socioeconomic differences across groups. Wealth is higher among those with more education, employment stability, occupational prestige, and incomes; among those who are married; and at older ages (Oliver and Shapiro 1995). Latinos in general, and immigrant Latinos in particular, average lower educational attainment and higher concentration in low-wage sectors of the economy, contributing to their lower wealth levels than whites. Likewise, they are younger and average more children, further undermining their relative wealth position (Cobb-Clark and Hildebrand 2006).

Among immigrant populations, ability to speak English and longer durations of U.S. residence are also associated with higher wealth levels (Akresh 2011; Krivo 1995), as is legal status (Diaz McConnell 2015). These factors, along with differences in human capital, help explain lower levels of assets among immigrant Latinos relative to immigrants from other regions (Hao 2007).

However, even after taking differences in the socio-demographic characteristics into account, disparities in wealth remain large; blacks and Latinos continue to average substantially lower wealth levels than whites, and immigrants average far fewer assets than their native-born counterparts (Campbell and Kaufman 2006; Cobb-Clark and Hildebrand 2006). The source of the remaining residual is often attributed to either behavioral differences across groups (i.e., differences in saving vs. consumption behavior, differential asset allocations, and so on) or discrimination, particularly in housing and rental markets.

While there is no consistent evidence that differential rates of saving versus consumption explain the low asset levels of immigrant Latinos (Cobb-Clark and Hildebrand 2006b), they do accumulate wealth more slowly than natives (Amuedo-DOrantes and Pozo 2002). In addition, Latinos are less likely than whites to invest in financial assets and higher-performing (though also higher risk) financial instruments. While it is unclear how much savings decisions affect the wealth holdings of low-income populations, it is nevertheless important to consider the influence of attitudinal attributes when possible.

The previous literature has also pointed to the importance of structural impediments to asset accumulation among Latinos and immigrants in explaining residual wealth inequality. A common finding is that the wealth pay-off to life-cycle and human capital characteristics is often lower for Latinos (along with blacks) than for native whites (Campbell and Kaufman 2006). For immigrant Latinos, in particular, traditional predictors often explain little of the variation in wealth. For example, while immigration and young age explain much of Asians’ initial wealth disadvantage relative to whites, the same is not true for Latinos, who seem more vulnerable to structural forces (Krivo and Kaufman 2004), including the negative effect of darker skin tone (Painter, Holmes and Bateman 2016; Painter and Qian 2016) and limited access to mainstream financial institutions (Rhine and Greene 2006). In fact, the socio-demographic correlates of assets are so different between U.S. Latinos and others that Fisher and Hsu went so far as to conclude that, “the model of saving applied to U.S. households may not be appropriate in studies focusing on the savings behaviors of Hispanics” (2012: 137).

While our understanding of the social forces shaping Latino and immigrant wealth accumulation is improving, significant gaps remain, owing largely to data limitations. The Survey of Consumer Finance (SCF) is one of the most commonly used instruments for assessing wealth among the U.S. population, but does not identify the foreign born. Both the Survey of Income and Program Participation and Panel Study of Income Dynamics have modules on wealth, but they do not query specifically about wealth held abroad. The New Immigrant Survey (NIS) does contain information on U.S. and foreign assets, as well as detailed immigration histories, but the sample contains primarily legal residents, and thus undocumented and recently arrived immigrants tend to be under-represented. Moreover, most previous analyses of immigrant wealth using the NIS have included wealth held abroad into their measures of household assets, but have not analyzed wealth separately by the location of the assets.

What is needed is additional research that examines the processes of financial incorporation and wealth accumulation from a transnational perspective, taking into account the unique forms of saving and debt relevant to immigrant populations.

That is, while most analyses of wealth look at either total net worth (all assets minus all debts) in all locations, or specific asset types (such as financial wealth or housing equity) net of debt, I argue that for low-skill Latino immigrants it is instructive to both distinguish between wealth held abroad and in the United States, and also to evaluate assets, debt, and net worth separately.

First, immigrant populations have an added level of decision making when it comes to assets; not only do they decide how much to save and in what form (i.e., in property or financial instruments), they also decide where to save (Keister and Vallejo 2019). Describing migrants’ asset accumulation transnationally is important both to assess how much wealth is missed by surveys that focus on assets on the U.S. side of the border and because the location of assets has implications for immigrant incorporation. That immigrants are disadvantaged in their ability to access formal modes of savings is well known. This is problematic because access to formal banking protects against theft and loss, and helps accommodate other financial transactions such as bill-paying and purchases. It helps to establish credit, encourages wealth accumulation, and also sets in motion a number of consumer protections and financial regulations (Rhine and Greene 2006). Foreign banks can serve some of these functions, but are less easily accessed by migrants living abroad, and lack the credit-building function.

A transnational wealth perspective would also enhance our understanding of the social forces shaping asset accumulation. In particular, the poor applicability of traditional life-cycle models to immigrant Latino earnings could be a result of both missing assets held abroad and differences in the impact of socio-demographic forces on assets by location. This could be particularly true for family structure. An important segment of the migrant Latino population consists of target earners, who do not intend to remain in the United States. While migrant trips have a tendency to drift towards permanent settlement, there is still tremendous variability in family structure and living arrangements within the migrant community. Experiences of family separation could shape not only where migrants prefer to save, but also the cost of living and supporting families, potentially altering the link between family structure and asset accumulation.

Second, the ability to save and the ability to borrow are related but distinct phenomenon. An important vein of research into wealth inequality focuses on the explosion of debt, particularly consumer debt, in recent decades and its negative implications for financial security (Wolff 2011). Yet among immigrant populations, the ability to borrow is often essential for income smoothing and acquiring essential consumer goods after arrival. And yet, access to conventional forms of credit, even consumer credit, is limited for immigrants, who instead often rely on more informal sources such as family members. Assessing debt separately from assets, and distinguishing between institutional and informal sources, is thus important in its own right. Moreover, if a large share of debt is informal in nature, total net worth does not accurately reflect a household’s ability to absorb income shocks; repayment to family and friends is easier to postpone in an emergency than repayment to formal lenders. And finally, if debt is more evenly distributed than assets then variability in saving will be masked by measures that look only at total net worth.

Finally, we also need more research that examines saving, borrowing, and socio-demographic characteristics that are unique to Latino and other immigrants. For example, while the SCF asks respondents about an impressive array of assets (from checking accounts to mutual funds, stocks, and even assets held in trusts) and debt (from credit cards to home equity lines of credit), they do not ask about saving in cash or money borrowed from family and friends. This is problematic because previous research has highlighted that many immigrant Latinos are unbanked, and by definition do much of their saving in cash. In fact, this is one of the principal reasons that Latino immigrants are so often the victims of crime. Likewise, for some migrants, inability to repay debt incurred abroad could in fact prompt migration, and many more carry debt associated with crossing into the United States, or paying for the crossing of family members. And finally, traditional surveys often lack information on legal status and other immigrant characteristics essential to financial incorporation among immigrants. For all of these reasons, more exploratory analyses of local immigrant populations are needed.

Data nad Methods

To address these issues, I draw on data from the Gender, Migration, and Health among Hispanics Study, a project that collected original and locally representative data among Latino immigrants during 2006 and early 2007 in the Durham, Carrboro, Chapel Hill, NC metropolitan area (hereafter referred to as “Durham,” where the vast majority of respondents lived). Durham represents a valuable vantage point to study Hispanic immigrant incorporation for a number of reasons. The overall area has been growing rapidly, as part of the national shift in population from Rustbelt to Sunbelt states. The influx of highly educated workers attracted to growing job opportunities in the nearby Research Triangle Park, universities, and other large employers generated an intense demand for low-skill service and construction labor. Some employers responded by recruiting Latino immigrant laborers from more traditional receiving areas or even directly from Mexico (Flippen and Parrado, 2012; Johnson-Webb, 2003; Parrado, Flippen, and Uribe, 2010). Once early migrants became established, a cycle of chain migration began that fueled growth in the Latino population from a mere 1 percent of Durham’s population in 1990 to 11.9 percent by 2007. While our data are only locally representative, a similar scenario played out in other “new destinations” throughout the American southeast (Zuniga and Hernandez-Leon 2006).

The precarious position of Latino immigrants in Durham presented unique challenges for approximating a locally representative sample. Community involvement and original sampling techniques were necessary to enhance coverage and reduce under representation, especially of the undocumented. Our study relied heavily on Community Based Participatory Research (CBPR), a participatory approach to research that incorporates members of the target community in all phases of the research process (Israel et al., 2005). In our case, a group of 14 community members assisted in the planning phase of the study, survey construction and revision, and devising strategies to boost response rates and data quality. In addition, CBPR members were trained in research methods and conducted all surveys. Finally, through ongoing collaborative meetings, they were also influential in the interpretation of survey results. It is difficult to overstate the wealth of culturally grounded understanding that they brought to project findings.

At the same time, the relatively recent nature of the Latino community in Durham rendered simple random sampling prohibitively expensive. We therefore employed targeted random sampling techniques (Waters and Bernacki, 1989). Based on our knowledge of the community, we identified 49 apartment complexes and blocks that house large numbers of immigrant Latinos. We then collected a census of all the apartments in these areas and randomly selected individual units to be visited by interviewers. Men and women’s random samples were independently drawn, and surveys were collected by same-sex interviewers. Using community members as interviewers helped achieve a refusal rate of only 9 percent, and a response rate, which also discounted randomly selected units in which contact was not made after numerous attempts, of over 72 percent. While the larger project included a sample of 2,800 immigrant Latino/a men and women between the ages of 18 and 49, this analysis is based on a wealth module administered to 353 and 339 men and women, respectively. All interviews were conducted in Spanish, usually in the homes of respondents, with interviewers filling out paper surveys that included a mix of close-ended and open-ended questions.

To evaluate potential biases arising from targeted random sampling, we compared our sample with data on Latino immigrant men from the 2000 Census. The results show that the vast majority of Durham’s Latinos, close to 80 percent, live in areas similar to those in which our targeted samples are located, i.e. in blocks that are between 25 and 60 percent Latino. This figure would likely be even higher if block-level data identifying the foreign born were available. Moreover, there were no statistically significant differences between our data and census data on a number of socio-demographic characteristics such as age, employment status, hourly wages, marital status, and year of arrival in the United States. Our respondents were slightly less educated than Latino immigrants enumerated in the Census, reflecting differences in question wording and the tendency for less well-educated individuals to be under-enumerated in the Census (Parrado, McQuiston, and Flippen 2005). However, even though we included a wide diversity of neighborhoods in our sampling strategy we cannot rule out the possibility that more established immigrants may be under-represented in the study. More importantly, our sampling strategy by definition excludes homeowners, and thus the analysis should be taken as an examination of the early stages of wealth accumulation among this population. While it is important to consider this limitation, the targeted approach is better suited to the study of immigrant populations than other methods of recruitment such as snowball or convenience samples.

A main advantage of the original data collected as well as the collaboration with community members was the ability to development a questionnaire specifically tailored to assess the financial incorporation of low-skill immigrant Latinos. In additional to traditional questions on assets and debt, for example, we also probed about saving in cash, which is common among immigrant Latinos in Durham, as well as formal and informal types of debt. This included not only credit related to automobiles, educational expenditures, credit cards, and property ownership, but also store credit, debt associated with crossing the border, money borrowed from family and friends, and other debt held abroad. We also collected detailed information on the human capital, life-cycle, labor market, and immigration characteristics found by previous studies to influence wealth accumulation. In addition, we also collected data on both family structure and living arrangements, to capture the influence of family separation on assets, as well as legal status.

Analytic strategy and model specification

Our overall objective is to assess the extent of asset ownership in the United States, abroad, and overall; to assess the extent of institutional and informal debt; and to evaluate wealth net of debt. Our first set of dependent variables thus relate to U.S. and foreign assets. This includes a dummy variable indicating whether or not respondents report owning a U.S. savings or checking account, as well as a dummy indicator of reporting savings in cash. These variables are not mutually exclusive. We also queried about property ownership, employer-sponsored savings accounts, and other forms of wealth in the United States. However, because our sample targeted apartment complexes in Durham, very few respondents reported owning property in the United States (other than consumer durables such as cars and electronics equipment), and virtually none reported stock ownership. I therefore do not consider these types of asset ownership separately. The measure of total U.S. assets thus sums the savings reported in either cash or U.S. checking/savings accounts. I also create a dummy indicator of foreign banking, which includes checking and savings accounts. I also include a dummy indicator of foreign property ownership, and a measure of equity held in foreign property).

The next set of dependent variables relate to debt. Here the main distinction is between formal, institutional borrowing and informal credit. The survey included a series of questions about debt incurred crossing the U.S. border, car loans, credit cards, store credit, educational debt, borrowing from family and friends, and other debt. From this information, I create a dummy indicator of formal, institutional credit, which includes car loans, store credit, property loans, and educational loans. An additional dummy variable indicates informal debt, related to border crossing or other borrowing from family and friends. Additional variables were then created indicating the total amounts of institutional and informal lending, respectively. Finally, I also calculate total net worth, defined as all assets (U.S. and abroad) minus total debt, both institutional and informal.

The model includes socio-demographic and human capital predictors important to theories of wealth accumulation: age, educational attainment, and family structure. I include both age and a squared term to capture non-linear effects. Educational attainment is measured with a dummy variable indicating relatively low levels of completed schooling, i.e., those who did not advance beyond primary school. Marital status is captured differently for men and women, reflecting the gendered dynamics of Latin American migration to the U.S. Specifically, among men I distinguish between those who are unmarried (single, separated/divorced, or widowed); married and accompanied/co-resident with their wives; and married but unaccompanied by their wives, who continue to reside in their country of origin. Among women living in Durham, it is extremely rare for married women to be unaccompanied. I therefore distinguish only between married and unmarried women.

Immigration-related characteristics include indicators of English ability, duration of residence in the Durham area, internal migration, and legal status. English ability is measured by a dummy variable indicating whether the respondent reported being able to speak English well or very well (as opposed to more or less or not at all). I also include a continuous measure of self-reported number of years of residence in the Durham area, and a dummy indicator of whether respondents came to Durham via another U.S. location, as opposed to directly from abroad. Results indicate that this specification more often predicts assets and debt than either total U.S. or Durham residence alone. Finally, a dummy variable for undocumented status reflects the response from a direct question on legal status.

To capture economic resources, I include a measure of total weekly household income (logged), as well as whether or not respondents report being paid in cash. Off-the-books employment in this sample is associated with employment instability and lack of benefit provision (Flippen 2012), and can be considered an overall indicator of employment marginality that may impact wealth over and above its association with income. And finally, I also include a dummy variable indicating a long planning horizon (namely, reporting that when planning family finances, respondents plan five or more years ahead, as opposed to several months, the next year, or the next two to three years).

After presenting descriptive statistics for all dependent and independent variables, I next discuss findings from logistic regression models predicting the likelihood of reporting different types of assets: U.S. banking, saving in cash, foreign banking, and foreign property ownership. I then estimate ordinary least squares (OLS) models of the logged total amount of assets held in the United States, abroad, and overall.¹ The next step is to model whether respondents report institutional and informal debt, again using logistic regression, and then the logged amount of debt held in these forms, using OLS. Finally, I use OLS to predict the log of TNW.² For all outcomes, I run the models both separately for men and women, to assess whether the predictors of wealth vary by sex, and also for the pooled sample, to assess the relative wealth position of women to men.

Results

Table 1 reports descriptive statistics for the assets, debt, and net worth reported by respondents, separately by the location of the assets and sex. Roughly one-third of migrant Latinos in Durham reported having a checking account in the United States (33.9 percent among women and 31.7 percent among men), and a somewhat smaller share reported a U.S. savings account (27.1 and 29.2 percent for women and men, respectively). Taken together, 45 percent of men and women in the sample reported U.S. bank account ownership. The average amounts held in these accounts is relatively low. Among those who report U.S. banking, the mean held is a meager $1,339 for men and $1,498 for women. A relatively large share of area migrants also reports saving in cash, particularly among men. Specifically, the 27.5 percent of men who report cash savings average nearly $750 in this asset, while the 17.1 percent of women reporting saving in cash average a similar $780. Combining formal banking and cash savings, 62 percent of men and 53.7 percent of women report saving in the United States, and average nearly $1,300 and $1,500 in U.S. assets, respectively.³

Table 1.

Descriptive statistics of assets and debt held by immigrant Latinos in Durham, NC by sex

	MEN			WOMEN
	% Reporting	Mean for all	Mean for asset holders	% Reporting	Mean for all	Mean for asset holders
Assets
U.S.
Bank (checking/saving)	44.8	$ 600	$ 1,339	44.5	$ 133	$ 1,498
Cash	27.5	$ 204	$ 741	17.1	$ 134	$ 780
Total	62.0	$ 803	$ 1,294	53.7	$ 801	$ 1,491
Abroad
Bank (checking/saving)	30.6	$ 1,945	$ 6,356	18.0	$ 870	$ 4,837
Property	43.1	$ 6,654	$ 15,453	28.6	$ 3,938	$ 13,761
Total	55.2	$ 8,599	$ 15,566	36.9	$ 4,808	$ 13,039
Total - U.S. and abroad	81.3	$ 9,402	$ 11,546	66.4	$ 5,609	$ 11,790
Debt
Institutional
Automobile	9.4	$ 685	$ 7,327	17.7	$ 958	$ 5,416
School	1.1	$ 216	$ 19,087	1.5	$ 18	$ 1,240
Store	2.6	$ 13	$ 506	6.2	$ 49	$ 785
Property abroad	4.3	$ 263	$ 6,172	3.0	$ 123	$ 735
Total	16.4	$ 833	$ 5,546	24.4	$ 1,129	$ 4,910
Informal
Family	34.6	$ 304	$ 878	15.0	$ 98	$ 654
Crossing	5.7	$ 93	$ 1,643	6.5	$ 28	$ 735
Total informal	39.4	$ 456	$ 1,157	23.0	$ 1,130	$ 4,911
Total debt	$ 47.0	$ 1,370	$ 2,913	$ 37.8	$ 1,275	$ 3,375
Total net worth		$ 8,375			$ 4,352
N		353			339

A substantial share of Durham’s Latino migrants also reported saving in their countries of origin, though the phenomenon was highly gendered. Thus, while 30.6 percent of men reported checking or savings accounts abroad, only 18 percent of women did so. Men also averaged higher savings in those accounts, $6,356 relative to $4,837 among women. Men were also far more likely than women to report property ownership in Mexico, 43.1 percent relative to 28.6 percent. Among owners they also averaged more equity in property, nearly $15,600 relative to $13,800 among women. Taken together, the 55.2 percent of men reporting foreign assets averaged nearly $15,600 in wealth abroad, while the 36.9 percent of women with these assets averaged roughly $13,000.

Additional tabulations reveal that relatively similar shares of women and men (29.5 and 26.1 percent, respectively) report only U.S. savings. Men, however, are far more likely than women to report only saving abroad (19.3 vs. 12.7 percent), or saving in both the United States and abroad (36.0 vs. 24.2 percent). Women, in contrast, are far more likely than men to report no savings (33.6 vs. 18.7 percent).

Table 1 also reports descriptive statistics on the debts reported by Durham migrants. The most common form of institutional, formal debt held by Latino migrants in the area relates to automobile financing. Nine percent of men and fully 17.7 percent of women report outstanding automobile debt, and the average amounts are substantial: over $7,300 for men and $5,400 for women. Credit card debt, common among the American mainstream, is relatively rare among this population. While 9.8 percent reported having a credit card (not shown), only a small handful (less than one percent) reported ever carrying a balance on their cards, and the number with a balance at interview was even smaller. It was more common to carry store debt, usually for furniture or appliances. At 6.2 percent and nearly $800, on average, women were also more likely to report this kind of debt than men (2.6 percent and roughly $500). Just over one percent of men and women reported debt related to schooling, but when they did the amounts were substantial, especially among men. And finally, 4.3 and 3.0 percent of men and women, respectively, reported debt related to property held abroad. Among those reporting this type of debt, men averaged far higher amounts (nearly $6,200) than women ($735). Overall, 23 percent of women averaged $4,910 in institutional debt, relative to 15 percent and $5,546 among men.

The most common form of informal debt related to the costs of crossing into the United States. Roughly four percent of men and women reported crossing-related debt at the time of survey, and the average amounts owed were sizeable: $1,500 for men and $1,900 for women. Moreover, eighty seven percent of men and 77 percent of women reported paying for their crossing into the United States, with costs averaging $2,420 and $3,089 among men and women, respectively. Thus, the cost of crossing to the United States is a significant drain on the meager financial resources of recent migrants.

Debt to family members for other reasons was also quite common – 34.6 percent among men and 15 percent among women. This type of debt most commonly stemmed from assistance getting established in Durham, but also included money borrowed to cover medical emergencies and other types of expenses. Taken together, roughly 39.4 percent of men and 23 percent of women held at least some informal debt, and among those who did the amount owed was over $1,157 for men and $4,911 for women. Overall, 47 percent of men and 37.8 percent of women held at least some debt, formal or informal. Additional tabulations reveal that only 6 percent of men and women hold both institutional and informal debt. Women are more likely than men to hold only institutional debt (16.8 vs. 9.1 percent), while men are more likely than women to hold only informal debt (33.4 vs. 16.2 percent). Once debt levels are taken into account, the average TNW for migrant men in the sample was $8,375, compared to a substantially lower $4,352 among women.

The overall low levels of wealth reflect the disadvantaged socio-demographic characteristics of Durham’s immigrant Latinos. Table 2 presents independent variables separately by sex. There are sharp differences by sex in the family structure and life-cycle characteristics predictive of wealth. Latino immigrant men in Durham are slightly less likely to be married than their female counterparts, 73.1 percent and 82.3 percent, respectively. But a full 20.1 percent of men in the sample are married but unaccompanied, with their wives continuing to reside abroad, a phenomenon that is all but non-existent among women. Women are more than twice as likely to be co-residing with minor children (80 percent relative to 37.1 percent), and somewhat less likely to have minor children abroad (30 percent relative to 38.8 percent). Men and women in the sample average roughly 31 years of age, and relatively low levels of education; a full 43.6 and 39.4 percent of women and men, respectively, report not advancing beyond primary education. The population in our sample is also recently arrived to the area, averaging a scant five years for men and women, though 54.4 and 34.2 percent of men and women, respectively, came to Durham via another U.S. location. Only 8.2 percent of men and 5.3 percent of women report speaking English well or very well, and a full 90 percent are undocumented. Average weekly household earnings are roughly $510 for men and $539 for women, and a sizeable fraction – 17.9 percent of men and 14.2 percent of women – are paid in cash, indicating informal, off-the-books employment. Finally, at 33.1 percent, men are nearly than twice as likely as women (16.8 percent) to report a long time-horizon when planning family finances.

Table 2.

Descriptive statistics of independent variables, by sex

	Men	Women
Marital Status (%)
Married accompanied	53.2	82.3
Married unaccompanied	20.1	-
Not married	26.7	17.7
Children (%)
Minor at home	37.1	80.0
Minor abroad	38.8	30.0
Age (mean)	31.0	30.5
(S.D.)	(8.9)	(7.7)
Low Education (%)	39.4	43.6
Years in NC	4.8	5.0
(S.D.)	(3.8)	(3.6)
Internal Migrant (%)	54.4	34.2
Speaks English well	8.2	5.3
Undocumented	90.4	90.3
Monthly household earnings (mean)	$ 510.90	$ 539.00
(S.D.)	(238.92)	(247.12)
Paid in cash	17.9	14.2
Long time horizon	33.1	16.8
N	353	339

Asset ownership

The next step in the analysis is to examine the socio-demographic correlates of asset accumulation in the United States, abroad, and overall. Table 3 presents coefficients from binomial logistic regression models of asset ownership, as well as those from ordinary least squares models of logged asset values, separately by sex and asset type/location. Important differences are observed in the predictors of U.S., foreign, and total assets, and also between men and women.

Table 3.

Binomial logit models of asset ownership and OLS models of asset values, by type of asset and sex (S.E. in parentheses)

	MEN
	Logit Models of Ownership				OLS Models of Asset Amounts
	U.S. Bank	Cash saving	Forgn. Bank	Forgn. Prop.	U.S. Assets	Forgn. Assets	Total Assets
Marital Status (ref=accompanied married)
Unmarried	−1.00 *** (0.40)	0.48 (0.40)	−0.08 (0.38)	0.61 * (0.37)	−1.61 *** (0.49)	0.85 (0.77)	0.43 (0.60)
Unacc. Married	−0.36 (0.37)	0.80 ** (0.39)	0.29 (0.34)	1.17 *** (0.35)	−0.97 ** (0.46)	1.85 *** (0.72)	0.92 * (0.56)
Children
Minor at home	0.36 (0.33)	−0.57 (0.36)	−0.77 ** (0.33)	−0.27 (0.32)	0.31 (0.42)	−0.71 (0.66)	−0.13 (0.52)
Minor abroad	−0.74 ** (0.33)	−0.69 * (0.36)	0.38 (0.32)	−0.21 (0.31)	−1.02 ** (0.41)	0.08 (0.65)	−0.36 (0.51)
Age	0.20 * (0.12)	−0.18 (0.11)	−0.05 (0.11)	0.28 *** (0.11)	0.27 ** (0.14)	0.31 (0.21)	0.26 (0.17)
Age squared	0.00 * (0.00)	0.00 * (0.00)	0.00 (0.00)	0.00 ** (0.00)	0.00 * (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.28 (0.28)	−0.57 ** (0.28)	0.05 (0.26)	0.00 (0.25)	−0.25 (0.33)	−0.04 (0.51)	−0.37 (0.40)
Years in NC	0.08 ** (0.04)	0.02 (0.04)	0.00 (0.04)	0.02 (0.04)	0.06 (0.05)	−0.03 (0.08)	0.02 (0.06)
Internal migrant	0.15 (0.28)	0.00 (0.27)	0.25 (0.26)	0.41 (0.25)	0.17 (0.33)	1.05 ** (0.52)	0.66 (0.41)
Speaks English well	0.35 (0.50)	−1.17 * (0.62)	−0.26 (0.50)	−0.14 (0.46)	0.73 (0.60)	−0.32 (0.93)	0.13 (0.73)
Undocumented	−1.16 ** (0.52)	0.52 (0.53)	0.51 (0.49)	−0.03 (0.42)	−1.23 ** (0.56)	0.29 (0.87)	−0.12 (0.69)
HH earnings (log)	0.44 ** (0.19)	0.65 ** (0.29)	0.24 * (0.15)	0.08 (0.11)	0.43 *** (0.13)	0.25 (0.21)	0.54 *** (0.16)
Paid in cash	−1.73 *** (0.42)	0.29 (0.33)	−0.14 (0.33)	−0.46 (0.32)	−1.89 *** (0.41)	−0.61 (0.64)	−0.34 (0.51)
Long time horizon	0.11 (0.27)	−0.25 (0.28)	0.34 (0.25)	0.00 (0.25)	0.39 (0.33)	0.28 (0.51)	0.38 (0.41)
Intercept	−5.09 ** (2.20)	−2.34 (2.45)	−2.02 (1.99)	−6.32 *** (1.90)	−2.50 (2.38)	−3.49 (3.70)	−1.81 (2.92)
Chi/R Sq	112.12	39.37	24.21	42.45	0.28	0.09	0.08
	WOMEN
	Logit Models of Ownership				OLS Models of Asset Amounts
	U.S. Bank	Cash saving	Forgn. Bank	Forgn. Prop.	U.S. Assets	Forgn. Assets	Total Assets
Marital Status (ref=unmarried)
Married	−1.85 *** (0.50)	0.11 (0.52)	0.63 (0.66)	1.22 *** (0.50)	−0.82 (0.53)	2.18 *** (0.77)	0.54 (0.70)
Children
Minor at home	−0.41 (0.39)	−0.06 (0.40)	0.43 (0.46)	0.30 (0.38)	−0.56 (0.44)	0.60 (0.64)	−0.32 (0.58)
Minor abroad	−0.61 * (0.32)	−0.03 (0.36)	0.38 (0.35)	−0.15 (0.30)	−0.78 ** (0.37)	0.10 (0.54)	−0.53 (0.49)
Age	0.05 (0.14)	−0.21 (0.13)	0.07 (0.15)	0.16 (0.13)	0.04 (0.15)	0.19 (0.22)	−0.01 (0.20)
Age squared	0.00 (0.00)	0.00 * (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.77 *** (0.29)	−0.44 (0.33)	−0.20 (0.34)	−0.16 (0.28)	−1.12 *** (0.34)	−0.19 (0.49)	−0.92 ** (0.45)
Years in NC	0.10 ** (0.05)	−0.08 (0.05)	−0.08 (0.05)	−0.02 (0.04)	0.11 ** (0.05)	−0.08 (0.07)	0.10 (0.07)
Internal migrant	0.24 (0.31)	−0.40 (0.34)	−0.33 (0.34)	0.01 (0.28)	0.02 (0.35)	−0.10 (0.51)	0.06 (0.46)
Speaks English well	1.42 * (0.77)	−0.89 (0.82)	- -	−0.38 (0.72)	1.51 ** (0.76)	−0.95 (1.10)	0.74 (1.00)
Undocumented	0.62 (0.51)	−0.93 * (0.51)	0.76 (0.65)	−0.27 (0.47)	0.67 (0.59)	−0.33 (0.85)	0.67 (0.77)
HH earnings (log)	2.24 *** (0.42)	0.08 (0.17)	1.15 *** (0.45)	−0.07 (0.16)	0.71 *** (0.17)	−0.05 (0.24)	0.39 * (0.22)
Paid in cash	−3.20 *** (0.77)	0.53 (0.39)	−1.93 *** (0.76)	−0.16 (0.38)	−2.50 *** (0.47)	−0.74 (0.67)	−1.59 *** (0.61)
Long time horizon	0.69 * (0.38)	0.06 (0.41)	1.18 *** (0.35)	0.78 *** (0.32)	1.16 *** (0.43)	2.21 *** (0.62)	1.81 *** (0.57)
Intercept	−14.30 *** (3.19)	2.46 (2.01)	−12.04 *** (3.55)	−5.06 ** (2.12)	−2.40 (2.38)	−3.16 (3.43)	0.54 (3.11)
Chi/R Sq	136.62	15.00	52.85	33.76	0.27	0.13	0.15
	POOLED MEN AND WOMEN
	Logit Models of Ownership				OLS Models of Asset Amounts
	U.S. Bank	Cash saving	Forgn. Bank	Forgn. Prop.	U.S. Assets	Forgn. Assets	Total Assets
Marital Status (ref=accompanied married man)
Unmarried man	−0.97 *** (0.35)	0.57 * (0.34)	0.64 * (0.34)	0.77 ** (0.32)	−1.58 *** (0.41)	1.55 *** (0.61)	1.00 * (0.52)
Unacc. marr. man	−0.60 (0.38)	0.76 ** (0.38)	0.63 * (0.36)	1.37 *** (0.35)	−1.40 *** ######	2.30 *** (0.69)	1.32 ** (0.58)
Married woman	−0.66 *** (0.25)	−0.21 (0.27)	−0.03 (0.26)	0.05 (0.23)	−0.90 *** (0.31)	−0.09 (0.46)	−0.65 * (0.39)
Unmarr. woman	0.39 (0.43)	0.13 (0.46)	−1.02 * (0.58)	−0.98 ** (0.43)	−0.31 (0.48)	−1.95 *** (0.73)	−0.84 (0.61)
Children
Minor at home	0.00 (0.25)	−0.31 (0.26)	−0.25 (0.26)	0.02 (0.24)	−0.13 (0.31)	0.06 (0.46)	−0.02 (0.39)
Minor abroad	−0.56 *** (0.22)	−0.45 * (0.25)	0.32 (0.22)	−0.28 (0.21)	−0.78 *** (0.27)	−0.07 (0.40)	−0.57 * (0.34)
Age	0.12 (0.08)	−0.20 ** (0.08)	0.03 (0.09)	0.24 *** (0.08)	0.19 ** (0.10)	0.30 ** (0.15)	0.16 (0.13)
Age squared	0.00 (0.00)	0.00 ** (0.00)	0.00 (0.00)	0.00 ** (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.50 *** (0.20)	−0.53 *** (0.21)	−0.12 (0.20)	−0.08 (0.18)	−0.71 *** (0.24)	−0.16 (0.35)	−0.68 * (0.30)
Years in NC	0.08 *** (0.03)	−0.01 (0.03)	−0.02 (0.03)	0.00 (0.03)	0.07 ** (0.04)	−0.05 (0.05)	0.06 (0.05)
Internal Migrant	0.17 (0.20)	−0.13 (0.21)	0.06 (0.20)	0.22 (0.18)	0.07 (0.24)	0.48 (0.35)	0.43 (0.30)
Speaks English well	0.72 * (0.41)	−1.07 ** (0.48)	−0.54 (0.46)	−0.18 (0.38)	1.12 ** (0.47)	−0.58 (0.70)	0.26 (0.59)
Undocumented	−0.35 (0.35)	−0.07 (0.35)	0.61 (0.38)	−0.06 (0.31)	−0.27 (0.40)	0.20 (0.60)	0.38 (0.51)
HH earnings (log)	0.95 *** (0.25)	0.35 *** (0.14)	0.37 ** (0.16)	0.06 (0.09)	0.52 *** (0.10)	0.18 (0.15)	0.52 *** (13.00)
Paid in cash	−2.08 *** (0.35)	0.45 * (0.25)	−0.58 ** (0.28)	−0.32 (0.24)	−2.14 *** (0.31)	−0.63 (0.46)	−0.88 ** (0.39)
Long time horizon	0.30 (0.21)	−0.12 (0.22)	0.71 *** (0.20)	0.34 * (0.19)	0.74 ** (0.26)	1.14 *** (0.39)	1.03 *** (0.33)
Intercept	−7.74 *** (2.01)	0.21 (1.49)	−4.85 ** (1.71)	−5.86 *** (1.43)	−2.78 * (1.70)	−3.70 (2.51)	−1.39 (2.12)
Chi/R sq.	217.12	54.31	71.29	85.07	0.25	0.12	0.14

^***p<.01

^**p<.05

^*p<.10

Beginning with women, U.S. based assets vary considerably according to human capital, labor market, family structure, and immigration characteristics. When it comes to whether or not immigrant women have accessed U.S. banking institutions, those with higher household earnings and education are more likely, and those who are paid in cash less likely, to have either a checking or savings account in the United States. They also average higher asset amounts in the United States than their less educated and lower-earning peers. Those with better English skills and longer durations in Durham are similarly more likely to be banked than other women, and likewise average higher U.S. assets than other women. Those with children abroad, in contrast, are both less likely to be banked and average lower overall U.S. assets than those without minor children in sending communities. Married women, somewhat surprisingly, are less likely than single women to report ties to mainstream U.S. financial institutions, though this could reflect the higher independence of unmarried women rather than superior economic position. The fact that single women do not report higher U.S. asset amounts supports this view. Finally, those with longer planning horizons are also somewhat more likely to bank and average higher asset levels in the United States. Interestingly, none of the socio-demographic variables significantly predicts cash savings among women.

Men exhibit a somewhat different pattern with respect to U.S. assets. Specifically, unlike women neither educational attainment nor English ability is related to U.S. assets among men - either the probability of being banked or the amount of U.S. assets reported. Legal status, however, is associated with lower U.S. banking rates and average asset amounts among men, but not among women.⁴ Likewise, single men are less likely to be banked and average lower U.S. assets than married men. While unaccompanied married men do not have lower rates of U.S. banking than accompanied married men, they do average lower U.S. assets. Like women, men with minor children abroad save in the U.S. less, both in terms of lower odds of U.S. banking and also lower average U.S. assets, while formal workers and those with higher household earnings are more likely to be banked and average higher U.S. assets.

Also unlike women, the model predicting whether or not migrants save in cash produced interesting results for men. One might expect, given the impediments to mainstream financial institutions evident in Table 2, that the least well off would be most likely to save in cash, but this is not the case. While unaccompanied married men, those with children abroad, and those who do not speak English well are more likely than other men to save in cash, so too are higher earning and better educated men. Thus, while saving in cash is common among less settled men, it is a generalized strategy that more established men also engage in.

The socio-demographic predictors of savings abroad, also presented in Table 2, differ in important ways from U.S. savings. First, foreign savings are less responsive to life-cycle, human capital, and other characteristics; fewer independent variables rise to the level of statistical significance, and less of the variation in foreign savings is explained in the models (for instance, the R squared for U.S. asset values among men is .28, relative to only .08 for foreign asset amounts). Among women, only labor market indicators and planning horizon predict foreign banking; those with higher household earnings and longer time horizons are more likely to report foreign bank accounts, while those who are paid in case are less likely. Only married women and those with a long time-horizon are more likely to own property abroad, and to average significantly higher foreign assets than other women.

For men, foreign assets, especially equity in property, is more responsive to socio-demographic conditions. Unaccompanied married and single men are more likely to own property abroad. The former also average higher foreign asset amounts than accompanied married men, as do men who are older. For both men and women, the models are far less predictive of total assets than for country-specific assets.

The final set of models, that pool men and women, are instructive as well. As in the previous models, relative to accompanied married men, single and unaccompanied married men tend to save less in the United States and more abroad. Married women also report lower rates of U.S. banking and lower U.S. asset amounts than married accompanied men, though they do not differ in terms of foreign assets. It is unclear whether this represents a true difference or whether married women are just in the dark about some of their household assets. Single women, in contrast, do not differ significantly from accompanied married men with respect to U.S. assets, but are significantly less likely to report either banking or property ownership abroad, and also average substantially lower asset values abroad.

Debt

Table 4 presents the same series of binomial logistic and ordinary least squares regressions for institutional and informal debt holding and logged amounts, again separately by sex. Results confirm the importance of distinguishing the two types of debt, and identify important barriers to conventional credit among immigrant Latinos.

Table 4.

Binomial logit models of debt and OLS models of debt levels, by type of debt and sex

	MEN
	Logit Models of Debt		OLS Models of Debt Amounts
	Inst.	Inform.	Inst.	Inform.	Total Debt
Marital Status (ref=accompanied married)
Unmarried	−0.41 (0.55)	0.29 (0.37)	−0.61 (0.63)	0.61 (0.72)	0.44 (0.79)
Unacc. Married	0.34 (0.47)	0.02 (0.35)	0.11 (0.59)	0.03 (0.67)	−0.29 (0.74)
Children
Minor at home	0.11 (0.42)	0.09 (0.32)	0.20 (0.54)	0.36 (0.61)	0.59 (0.68)
Minor abroad	−0.09 (0.41)	0.45 (0.31)	−0.17 (0.53)	0.93 (0.61)	0.86 (0.67)
Age	0.12 (0.15)	−0.01 (0.10)	0.21 (0.18)	0.02 (0.20)	0.00 (0.22)
Age squared	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.11 (0.34)	−0.09 (0.25)	−0.13 (0.42)	−0.20 (0.48)	−0.06 (0.53)
Years in NC	0.02 (0.05)	−0.05 (0.04)	0.03 (0.06)	−0.10 (0.07)	−0.08 (0.08)
Internal Migrant	0.24 (0.34)	0.52 ** (0.25)	0.26 (0.43)	0.95 ** (0.49)	1.14 ** (0.54)
Speaks English well	−0.27 (0.60)	−0.16 (0.48)	−0.04 (0.77)	−0.22 (0.87)	−0.16 (0.96)
Undocumented	−0.85 ** (0.49)	0.97 ** (0.51)	−1.62 ** (0.72)	1.36 * (0.82)	−0.91 (0.90)
HH earnings (log)	0.93 ** (0.47)	−0.20 ** (0.10)	0.29 * (0.17)	−0.35 * (0.19)	−0.20 (0.21)
Paid in cash	−0.58 (0.52)	−0.75 *** (0.30)	−0.79 (0.53)	1.53 *** (0.60)	1.04 (*) (0.67)
Long time horizon	0.00 (0.33)	−0.41 (0.25)	0.17 (0.43)	−0.58 (0.48)	−0.14 (0.53)
Intercept	−8.33 ** (3.77)	−0.13 (1.80)	−3.82 (3.06)	1.20 (3.47)	3.39 (3.83)
Chi/R Sq	24.76	37.75	0.06	0.09	0.04
	WOMEN
	Logit Models of Debt		OLS Models of Debt Amounts
	Inst.	Inform.	Inst.	Inform.	Total
Marital Status (ref=unmarried)
Married	−0.27 (0.46)	−0.09 (0.44)	−0.26 (0.80)	−0.19 (0.68)	−0.78 (0.86)
Children
Minor at home	0.50 (0.39)	0.22 (0.38)	1.01 (0.66)	0.31 (0.56)	1.34 (0.71)
Minor abroad	0.09 (0.31)	0.07 (0.31)	0.13 (0.56)	0.20 (0.47)	0.35 * (0.60)
Age	−0.11 (0.13)	0.00 (0.13)	−0.09 (0.22)	−0.02 (0.19)	−0.08 (0.24)
Age squared	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.25 (0.29)	0.23 (0.29)	−0.41 (0.51)	0.27 (0.43)	−0.72 (0.55)
Years in NC	−0.02 (0.04)	−0.04 (0.05)	−0.04 (0.08)	−0.06 (0.07)	−0.09 (0.08)
Internal Migrant	0.25 (0.29)	0.16 (0.29)	0.51 (0.53)	0.12 (0.45)	0.13 (0.57)
Speaks English well	0.16 (0.61)	0.75 (0.61)	0.25 (1.14)	1.02 (0.97)	0.74 (1.23)
Undocumented	0.43 (0.50)	1.00 * (0.60)	0.38 (0.88)	1.25 * (0.75)	0.96 (0.94)
HH earnings (log)	0.46 * (0.27)	0.08 (0.15)	0.40 (*) (0.25)	0.15 (0.21)	0.50 * (0.27)
Paid in cash	−0.98 ** (0.51)	−0.30 (0.41)	−1.31 * (0.70)	−0.39 (0.59)	−1.33 * (0.75)
Long time horizon	0.23 (0.34)	0.33 (0.34)	0.53 (0.65)	0.60 (0.55)	0.91 (0.70)
Intercept	−2.93 (2.41)	−3.05 (2.09)	−1.90 (3.55)	−2.53 (3.02)	−1.84 (3.83)
Chi/R Sq	16.78	8.38	0.04	0.03	0.06
	POOLED MEN AND WOMEN
	Logit Models of Debt		OLS Models of Debt Amounts
	Inst.	Inform.	Inst.	Inform.	Total Debt
Marital Status (ref=accompanied married man)
Unmarried man	−0.35 (0.44)	0.91 *** (0.31)	−0.29 (0.57)	1.51 *** (0.56)	1.30 ** (0.66)
Unacc. marr. man	0.18 (0.44)	0.46 (0.35)	0.09 (0.64)	0.68 (0.63)	0.30 (0.74)
Married woman	0.35 (0.27)	−0.44 * (0.24)	0.52 (0.42)	−0.84 (0.42)	−0.32 (0.49)
Unmarr. woman	0.69 (0.46)	−0.50 (0.40)	0.59 (0.67)	−0.86 (0.66)	−0.19 (0.78)
Children
Minor at home	0.30 (0.28)	0.34 (0.25)	0.59 (0.42)	0.56 (0.42)	1.09 ** (0.49)
Minor abroad	0.01 (0.24)	0.23 (0.21)	0.01 (0.37)	0.48 (0.37)	0.48 (0.43)
Age	0.01 (0.09)	0.02 (0.08)	0.11 (0.14)	0.06 (0.14)	0.05 (0.16)
Age squared	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Low Education	−0.20 (0.22)	−0.01 (0.19)	−0.29 (0.33)	−0.04 (0.32)	−0.45 (0.38)
Years in NC	0.00 (0.03)	−0.04 (0.03)	0.00 (0.05)	−0.07 (0.05)	−0.08 (0.06)
Internal Migrant	0.21 (0.22)	0.37 ** (0.19)	0.33 (0.33)	0.58 * (0.33)	0.63 * (0.39)
Speaks English well	−0.02 (0.42)	0.12 (0.38)	0.16 (0.65)	0.23 (0.64)	0.27 (0.76)
Undocumented	−0.13 (0.34)	1.00 *** (0.38)	−0.57 (0.56)	1.37 ** (0.55)	0.10 (0.65)
HH earnings (log)	0.57 *** (0.24)	−0.07 (0.08)	0.31 ** (0.14)	−0.09 (0.14)	0.12 (0.17)
Paid in cash	−0.82 ** (0.36)	0.33 (0.23)	−1.04 ** (0.43)	0.68 * (0 42)	−0.01 (0.50)
Long time horizon	0.20 (0.23)	−0.11 (0.20)	0.40 (0.36)	−0.05 (0.36)	0.44 (0.42)
Intercept	−5.38 *** (2.09)	−2.09 (1.33)	−3.74 * (2.32)	−1.75 (2.29)	−0.92 (2.70)
Chi/R Sq	39.38	58.77	0.05	0.08	0.03

^***p<.0l

^**p<.05

^*p<.10

Among men and women, but particularly the former, the holding of formal, institutional debt is clearly higher among more advantaged migrants. Among men, the undocumented are significantly less likely to hold institutional debt, and average lower amounts of this debt, than their peers with legal residence. Among men and women higher earnings are positively associated with institutional debt, and among women (and in the full sample) being paid in cash is negatively associated with formal debt. Thus, holding institutional debt is a marker of financial incorporation, rather than a sign of economic distress. The pooled model suggests few differences across men and women in different family arrangements in access to formal credit, which instead appears to be largely a function of earnings.

Informal debt, in contrast, tends to be higher among those with fewer resources. Among men (and the total population), informal debt is both less common and lower among those with longer durations in Durham, indicating that over time migrants are able to repay loans from family and friends. Economic marginality is also associated with higher informal debt, at least among men; men with higher incomes are less likely and those paid in cash are more likely to report informal debt, and average higher amounts. Among both women and men, the undocumented are also more likely to hold informal debt than their legal resident counterparts. The pooled model is also instructive. Accompanied married men report less informal debt than their accompanied peers, likely reflected the added cost of arranging the crossing of wives and children into the United States. Unmarried women average higher informal debt than accompanied married men.

Total net worth (TNW)

The most common barometer with which to assess wealth levels is TNW. However, I argue that among low-skill immigrant Latinos, a focus on TNW obscures important variation in wealth. Essentially, because debt is relatively evenly distributed across the immigrant population, when subtracted from assets the overall variation in net wealth is also obscured. And, because much of this debt is non-institutional and thus has more flexible repayment schedules, this obfuscation masks important differences across individuals in their ability to absorb income shocks.

Indeed, when comparing models of TNW with those of assets, we see far fewer significant predictors. Among men, those who are married and unaccompanied have higher levels of foreign, and total, TNW than their accompanied peers, suggesting the cost-savings associated with supporting families abroad relative to the United States. While family separation requires maintaining two separate households, most unaccompanied migrant men in Durham co-reside with other migrants to keep costs to a minimum. Thus the lower cost of living abroad seem to facilitate asset accumulation. The only other significant predictor of TNW among men is household wages.

Among women, those who are married average higher levels of TNW than their single counterparts, as do those with longer time horizons. The expense of childrearing in the United States is again suggested here, as women living with minor children average significantly lower TNW than those without. Once debt levels are taken into account, the human capital and immigration characteristics that predicted asset values in Table 2 are no longer significant.

Finally, the TNW advantage of unaccompanied married men relative to accompanied married men remains in the pooled sample. Single men average more foreign wealth net of debt than accompanied married men. In the pooled sample both co-resident children and children residing abroad are associated with lower TNW. Once again, the negative association between being paid in cash and assets is evident, suggesting the serious disadvantage faced by those working in the informal sector. Other sociodemographic differences are relatively slight, especially compared the models of asset levels (Table 2).

Conclusions

The overwhelming majority of work on U.S. stratification in general, and the incorporation of immigrants in particular, has focused on labor market processes and rewards, to the relative neglect of assets and wealth. At the same time, research on wealth stratification tends to incorporate immigrants poorly. Immigration accounts for sixty percent of U.S. population growth. The impending retirement of the baby boom generation, coupled with the enduring impact of parental wealth during childhood on adult status attainment, make the asset incorporation of immigrants a pressing concern. And, while wealth levels among this population are extremely low, even modest assets can improve one’s standard of living and buffer against adverse labor market and health events. While data on immigrant wealth has improved in recent years, significant gaps remain. This paper draws on original survey data collected among low-skill Latino migrants in Durham, NC to address that gap.

Results show important variation among Durham’s immigrant Latinos in the ability to accumulate assets. One of the most basic conclusions from this project is the need to take a transnational approach to analyzing immigrant wealth. Nearly as many migrants in our sample held wealth abroad as had assets in the United States, particularly among men. And, for those who did, average asset amounts were far higher abroad as well. Property located abroad stands out as a particularly common and valuable asset among this population.

A transnational approach to immigrant wealth is important not only because foreign assets are a key component of savings, however, but also because the life-cycle and labor market predictors of wealth vary according to the location of the asset. In general, traditional life-cycle and human capital theories of wealth explain variation in assets held in the United States better than variation in wealth held abroad. For U.S. versus foreign assets one of the key determinants relates to family separation; men and women with children abroad, and men with wives abroad, are less likely to save in the United States and more likely to save in sending countries. Single men also tend to save less in the United States, and are more likely to invest in foreign property than married men who have reconstructed their families in Durham. Single women, in contrast, are more likely than their married counterparts to save in U.S. banks, and less likely to invest in foreign property. Not surprisingly, immigration characteristics such as time in Durham, English ability, and legal status predict U.S. but not foreign savings. The only characteristic found to consistently predict both U.S. and foreign assets was household earnings.

Results also show that a large share of migrants – roughly one-quarter of men and 17 percent of women – save in cash. When they do, average amounts are substantial; including cash savings in estimates of asset levels increases the average U.S. asset holding by one-third among the total population, and more than 50 percent for those reporting any type of U.S. asset. Clearly, among this low-income and recently arrived population, cash savings are an important component of wealth. While part of the impetus behind saving in cash stems from restricted access to mainstream financial institutions, among men even higher earning and better-established men are likely to save in cash.

In contrast to assets, debt is relatively evenly distributed across the immigrant Latino population. Family structure has little relation to either formal or informal types of borrowing, which is largely determined by labor market factors and duration of Durham residence. Those with higher earnings are better able to access institutional credit, and tend to owe less to informal lenders. The opposite is true for those who are paid in cash. The undocumented are also constrained in their access to institutional credit, and more reliant on informal sources.

The models of formal banking and debt, taken together, suggest important structural impediments to access to mainstream financial institutions among Durham’s migrant Latino community. Those working in the informal sector (as indicated by being paid in cash) average consistently lower wealth levels and less access to institutional forms of debt, even net of household earnings. Undocumented status is also associated with lower access to U.S. banks and institutional credit, particularly among men. These disparities add to the growing empirical support for the challenges to wealth posed by undocumented status (Rugh 2019).

Finally, Total Net Worth shows far less variation than assets, and is less well explained by socio-demographic characteristics. Married men and women average higher TNW than their single counterparts, though among men unaccompanied married men also average higher wealth than their accompanied peers. Those with minor children, either at home or abroad, average lower TWN. As was the case with previous studies, other significant predictors of TNW are surprisingly scant.

However, comparing the models of TNW with those of assets and debt suggest some explanations as to why life-cycle and human capital theories seem to explain immigrant Latino wealth so poorly. Human capital, labor market, and immigration characteristics exert a strong influence on asset levels, in keeping with prevailing wealth theories. However, debt among immigrants is more evenly distributed. All immigrants face substantial costs crossing the border and establishing households in the United States. At the same time, immigrant access to more variable types of debt such as consumer credit is relatively limited. Because overall asset levels are low, the evenness of debt levels reduces variation in net worth, making it appear that previous theories do not fit immigrant Latinos. Moreover, it is also important to consider the implications of different types of debt when considering TNW. Because all debt must ultimately be repaid, TNW provides an important assessment of long-term asset and financial incorporation. However, because much of the debt held by immigrant Latinos is informal in nature, which tends to me more flexible in terms of repayment, assets alone may be a better indicator of short-term economic security and the ability to absorb income shocks.

While this study adds to our understanding of the process of wealth incorporation among Latino immigrants, a number of caveats are in order. The most serious limitation of the study is the targeted nature of the sample, which under-represents migrants who have made the transition into U.S. homeownership. Some of the lower TNW among accompanied married men relative to their unaccompanied counterparts, for instance, could reflect the fact that accompanied married men are more likely to purchase property in the Durham. Thus, results should be taken as providing insight into the early phase of asset accumulation among immigrant Latinos, and subsequent research should investigate transnational wealth accumulation among more settled populations.

Table 5.

Ordinary least squares coefficients of models predicting Total Net Worth (assets minus debt), by sex

	Men		Women		All
Marital Status (ref=accompanied married)		Marital Status (ref=single)		Marital Status (ref=accompanied married man)
Unmarried	0.01 (0.06)	Marr.	0.07 (0.07)	Unacc. marr. man	0.14 ** (0.05)
Unacc. married	0.13 ** (0.05)			Marr. woman	−0.02 (0.04)
				Single woman	−0.05 (0.06)
				Single man	0.08* (0.05)
Children
Minor at home	0.00 (0.05)		−0.05 (0.05)		0.00 (0.04)
Minor abroad	−0.05 (0.05)		0.01 (0.05)		0.20 (0.30)
Age	0.02 (0.02)		0.00 (0.02)		0.20 (0.10)
Age squared	0.00 (0.00)		0.00 (0.00)		0.00 (0.00)
Low Education	−0.01 (0.04)		−0.02 (0.04)		−0.02 (0.03)
Years in NC	0.00 (0.01)		0.00 (0.01)		0.00 (0.00)
Internal Migrant	0.05 (0.04)		−0.02 (0.04)		0.02 (0.03)
Speaks English well	0.07 (0.07)		−0.01 (0.09)		0.03 (0.06)
Undocumented	0.06 (0.07)		0.10 (0.07)		0.09 * (0.05)
HH earnings (log)	0.01 (0.02)		0.00 (0.02)		0.01 (0.01)
Paid in cash	−0.02 (0.05)		−0.04 (0.06)		−0.03 (0.04)
Long time horizon	0.04 (0.04)		0.15 *** (0.05)		0.09 *** (0.03)
Intercept	9.90 *** (0.28)		10.17 *** (0.29)		−3.10 (3.70)
R square	0.05		0.06		0.06

^***p<.01

^**p<.05

^*p<.10

Biography

Chenoa Flippen is an Associate Professor at the University of Pennsylvania’s Department of Sociology and Population Research Center. Her research addresses the connection between racial and ethnic inequality and contextual forces at the neighborhood, metropolitan, and national level. She has published on diverse topics in stratification, including minority aging and retirement security, the impact of residential segregation on minority homeownership and housing wealth, and Hispanic immigrant adaptation to the United States.