Temporary Labor Migration and Spouses’ Marital Quality: Evidence from Nepal

Abstract

Objective:

This study investigated the effect of temporary labor migration on spouses’ marital quality.

Background:

How temporary international labor migration affects the marital relationship remains unclear. Research shows migration increases couples’ risk of dissolution, whereas studies of spouses’ marital quality—much of which is cross-sectional and/or limited to either internal or joint migration—is more mixed. This lack of consensus masks the possibility that, under certain conditions, migration may improve spouses’ marital quality.

Method:

This study uses data from the Chitwan Valley Family Study, a panel study set in Nepal, and primary data collected among a subsample of migrant husbands (in East Asia, Middle East) and their wives and nonmigrant couples (in Nepal).

Results:

Findings from linear regression models show that, relative to non-migrant spouses, spouses engaged in temporary international labor migration report significantly higher marital quality—less conflict and more love—net of marital quality assessed 6 years earlier. However, these benefits are not enjoyed equally between spouses: husbands’ marital quality improves, whereas changes in their wives’ are less conclusive.

Conclusion:

The fact that these benefits (a) diverge from previous understandings and (b) vary by spouse’s gender extends current understandings of the conditions shaping this association: social and structural forces supporting men as breadwinners, a strong husband–wife bond facilitating husbands’ migration, and marriage-protective social environments at both ends of migration.

Introduction

One stream of research aiming to understand marital processes across contexts has focused on the conceptualization and measurement of the marital relationship, with particular interest in marital quality. Marital quality can provide a lens through which to study relationship change, as well as spouses’ distinct experiences within the same marriage. Important factors shown to shape marital quality across settings include gender, age, spouses’ workforce participation, and whether (or not) the latter is congruent with gender ideologies (Allendorf & Ghimire, 2013; Karney & Bradbury, 1995; VanLaningham et al., 2001).

Despite these literatures, few studies have examined changes in marital quality with regard to an increasingly common circumstance linked with these factors: temporary international labor migration. Existing studies—much of which investigate family transitions or marital quality cross-sectionally—show that transnational marriages experience higher rates of infidelity and dissolution than couples at home (Coe, 2011; Frank & Wildsmith, 2005). A more general literature around how married people experience temporary labor migration paints a more complicated story (e.g., Lu 2012; McEvoy et al., 2012; Menjívar & Agadjanian, 2007). Across all studies, a primary concern is the selection into migration and our ability to observe the effect of migration at later time points. Yet the selection process is not simply an empirical concern; it should also shape hypotheses regarding the effect of migration on marital quality over time. In other words, these same factors likely continue to shape how spouses experience their marriage while the spouse is working far away from home.

This study argues that a lack of consensus in the literature obscures the possibility that, under certain conditions, temporary international labor migration may actually improve spouses’ marital quality. It tests this possibility in the rural, agricultural setting of Nepal, where marriage remains nearly universal, individuals marry in late adolescence or early adulthood, and rarely divorce (Jennings, 2016; Yabiku, 2005). At the same time, a culture of labor migration has emerged (Adhikari & Hobley, 2015; Kandel & Massey, 2002): Due to persistent poverty associated with a continued reliance on subsistence agriculture (Maharjan et al., 2012), the number of individuals formally migrating abroad—those for whom the Nepali government has issued a labor permit for foreign employment—has increased annually from negligible to over 500,000 since the 1990s. In recent years, an estimated 33% of all households had a member working abroad at a given time (Sharma et al., 2014), many of whom are married men signing multi-year labor contracts in the Middle East or Malaysia (Ministry of Labor and Employment, 2015; Sharma et al., 2014) due to the substantially higher wages abroad. Importantly, although marriage is common, variation in marital experiences has grown as the nature of the marital bond has evolved in recent decades (Allendorf, 2017; Axinn et al., 2017).

The study uses the Chitwan Valley Family Study (CVFS) (collected 1996 to the present in Chitwan, Nepal; https://www.icpsr.umich.edu/icpsrweb/DSDR/studies/4538) and primary data collected among a subsample of migrant and nonmigrant couples to examine this association. The CVFS tracks international migrants over time and collected complete work, travel, and family histories for all respondents. The study also collected repeated measures of multiple dimensions of marital quality among the same married individuals: prior to husbands’ migration and 6 years later. Furthermore, the study provides the ability to analytically link husbands’ and their wives’ separate reports of marital quality. These tools offer a unique window into this association, evaluating in new depth spouses’ marital experiences while the husband is living and working away, simultaneously accounting for factors shaping husbands’ propensity to migrate as well as considering these factors in new light.

Theoretical Framework

Existing studies generally find that temporary labor migration threatens marriage. Across settings, transnational marriages are more likely to experience dissolution, higher rates of infidelity, and promiscuous behaviors than couples at home (Coe, 2011; Frank & Wildsmith, 2005; Landale & Ogena, 1995; Luke, 2010; Tamang, 2001; Yang, 2004). Studies of marital quality are more mixed. In their study of internal migrants in China, for example, Zhuang et al. (2014) found migrant couples experience increased marital conflict, whereas Jacka (2012) found women left behind reported less intra-household conflict. Other studies show marital satisfaction does not decline in China (Abbott & Meredith, 1994) and varies among migrants’ wives in Mexico (Wood et al., 2015).

The association between temporary international labor migration and changes in spouses’ marital quality remains unclear. However, existing research informs hypotheses regarding this association. One is the highly selective nature of marriage and migration. Previous studies discuss this as a limitation and estimate many supplemental models in attempts to show their results are not spurious. Two is theorized pathways through which migration may influence marital instability. This study discusses key factors shaping a spouse’s propensity to migrate, but argues these same factors likely influence how a couple experiences migration. In blending these literatures, there is reason to believe that, under specific conditions, migration does not threaten marriage, and may in fact improve couples’ marital quality, although these benefits may not be enjoyed equally between husbands and their wives. The goal is to identify setting-specific hypotheses and move to test them empirically.

Before Migration: Gender, Social Roles, and Migration

The most influential theory linking migrants with their families back home is the household strategy approach: households send a member away to diversify income and fill gaps in access to capital (Stark & Bloom, 1985). As such, decades of research have examined economic predictors of labor migration. Yet engagement in this strategy is often more complicated due to its embeddedness in social contexts in which norms and expectations shape individuals’ preferences regarding who should migrate for work. Notable is gender, a powerful social category shaping men’s and women’s experiences and expectations of others. In many regions of the world, men’s roles have historically centered on financial provision and women’s roles as caretakers. These preferences toward productive labor remain evident in many low-income settings in which labor migration is common (De Jong, 2000). In fact, men still comprise the majority of labor migrants worldwide (International Labor Organization, 2015). Despite local preferences, however, structural forces may not align with them. Scholars have documented a “feminization” of migration in recent decades (Hoang, 2011; Mahler & Pessar, 2006), due in part to rising demand in female-dominated occupations. Research has examined distinct causes and consequences associated with female migration, with particular interest in whether it is congruent with or challenging historical gender ideologies (Paul, 2015).

In Nepal, gender remains strongly associated with productive labor. Hindu beliefs dictate that women move to their husband’s village and remain close to home to care for children, whereas men should be employed and provide financially for their families. Financial opportunities have shifted greatly in recent decades in response to global markets: The wage differential in the Middle East and Malaysia—the most common migrant destinations—is significant: an estimated 54% of Nepalese live below $60 per month, whereas intergovernmental agreements recently set migrant wages in these destinations at $200 per month, making such international destinations especially lucrative for labor migrants. The enduring expectation of husbands as breadwinners (Adhikari & Hobley, 2015; Sharma, 2008) is reflected in the gender imbalance among labor migrants: Despite foreign demand for both typically female (e.g., domestic) and male (e.g., construction) labor, men comprise the majority (96%) of formal international migrants (Ministry of Labour and Employment, 2015). This preference is reinforced at the institutional level with intermittent government bans on female migration (Graner, 2001). So, unlike other settings in which migration is common, preferences that it is men who should move remain relatively unchallenged. And, due to Nepal’s early ages at marriage and low divorce rates, the majority of Nepalese labor migrants are married men (Sharma et al., 2014).

Before Migration: The Marital Relationship as Facilitating a Desirable Livelihood Strategy

Even if it is a desirable and viable livelihood strategy, not all couples experience a spouse’s migration or, in Nepal’s case, send a husband away for work. Rather, spouses’ marital quality (e.g., less conflict and greater satisfaction) (Bradbury et al., 2000; Moore et al., 2004) may select couples into migration. For example, when spouses do not agree, spousal conflict increases (Salazar, 2015). This negatively affects the quality of communication (Alberts & Driscoll, 1992; Link, 2011), which then decreases the likelihood of desired behaviors being carried out. Migration as a behavior is a high-risk, high-reward strategy (Stark 1991; Stark & Fan, 2007). In many low-income settings, the economic rewards are clear, but they must be weighed against the myriad costs for spouses, which can be severe. Higher marital quality has been linked to respect, commitment, and the ability to compromise, all of which provide a foundation for couples’ risk-taking behaviors (Kurdick, 1995; Moore et al., 2004). Regarding a spouse’s labor migration, this foundation is likely crucial (Sana & Massey, 2005; Seshan & Yang, 2014; Shaw & Charsley, 2006; Stark & Lucas, 1988).

Although marriage remains near universal in Nepal—more than 98% of men and women in the larger panel study examined here had married by 1996, with only 6% experiencing marital dissolution by 2008 (Jennings, 2016; Yabiku, 2005)—variation in the nature of the husband–wife bond is increasing, due in part to ideational factors. “Traditional” marriages were historically seen as bonds between families, with people having little say in who they marry and women deferring to their husbands or in-laws (Fricke 1986). In Nepal, exposure to new ideas has shifted attitudes and behaviors toward love marriages emphasizing an emotional connection between spouses (Allendorf, 2017; Allendorf & Ghimire, 2013; Axinn et al., 2017). As the husband-wife bond grows more central to family life—the fundamental social unit in this setting—marital quality likely plays a key role in crucial decisions, such as a spouse’s migration.

While Away: Pathways Through Which Migration May Alter Marital Stability

Keeping these key selection factors—gender ideologies shaping work preferences and marital quality as facilitating migration—in mind, what happens once a spouse is away? Scholars often cite a decline in social support, which may cause spouses to grow apart, as a pathway through which migration threatens marriage. Prolonged geographic separation adversely affects the well-being of migrants and their spouses at home (Farooq & Javed, 2009; Lu 2012; Wilkerson et al., 2009), spouses worry about misused earnings or remittances (Lama et al., 2017), and wives in Mexico fear abandonment and struggle with increased responsibilities in the migrant’s absence (Nobles & McKelvey, 2015). A related pathway is exposure to new networks at the point of destination, which may exhibit different social acceptability regarding separation and divorce (Frank & Wildsmith, 2005). These networks may also exert less social control, allowing migrants to explore new potential partners. They can impact the nature of migrants’ commitment to those back home as well. For example, different normative values in the U.S. expose Mexican migrants to new ideals based on bonds of desire, rather than obligation (Hirsch, 1999), which may influence remittance behaviors (Vanwey, 2004).

Importantly, conditions in sending and receiving countries may lessen these commonly cited threats. For spouses at home, challenges may be less problematic in settings in which divided households are the norm (Gupta, 2002; Kandel & Massey, 2002), as local networks and extended family provide social support in the migrants’ absence. This is likely particularly true for settings such as Nepal where, as descriptive statistics of this study sample show, 77% and 62% of respondents have had a family member and friend migrate, respectively. As for migrants, the threat of new networks hinges on receiving countries’ social contexts (Conway, 2007). The most common Nepalese migrant destination (Gulf Cooperation Council (GCC) member states) maintains strict migrant monitoring systems, which, coupled with historically patriarchal and marriage-protective norms, work to minimize migrants’ integration into local social life, particularly the pursuit of romantic relationships (Al-Nasr, 2011; De Bel-Air, 2014; Gardner, 2012; Gardner et al., 2013). East Asia is similarly challenging given ethnic and cultural differences and language barriers, although these are not as restrictive given a history of Nepalese Gurkha soldiers serving in Malaysia and Singapore as part of the British Army (see National Army Museum, n.d.). So, while Nepalese migrants work in settings far from home, their opportunities to freely engage with these networks and explore potential partners are limited, which may lead migrants to place more value on their marriage.

Separate studies show temporary labor migration may strengthen transnational ties. This is because successfully having a spouse working abroad—global householding (Douglass, 2006)—demands cooperation on both ends of migration. Earning money abroad and sending it home has been found to maintain familial bonds in Latin America (Sana & Massey, 2005). And scholars identify “wife-work” back home as a crucial component to South Asian men’s migration decades ago (Charsley & Shaw, 2006) and in Mexico more recently (Kanaiaupuni, 2002). Also embedded in this strategy is that it is temporary. Chinese couples who chose to live apart (e.g., job opportunities) did not experience declines in marital satisfaction (Abbot & Meredith 1994), but acted together to pursue what they deemed a short-term situation for a long-term solution. Couples with an already strong foundation likely understand this strategy as temporary and cope accordingly as aspirations evolve (Toyota et al., 2007). In Nepal, common migration streams are temporary by design: GCC member states and Malaysia limit foreign workers to 2- to 3-year labor contracts, providing no paths for migrants to obtain citizen status, own property, or sponsor their families (Kaur, 2014; Khadria, 2008; Naithani, 2010). Relatedly, spouses spending much less time together are at lower risk for conflict, which is distinct from a strengthening connection; this possibility is discussed in more detail later.

Of course, given how stratified social life is by gender, it is likely that spouses do not experience migration in the same ways when it comes to their marriage. This is true in high-income settings and in Mexico, where family relocation and/or separation tends to benefit or adversely affect one spouse more than the other (Boyle et al., 2008; Gutierrez-Vazquez et al., 2018). And in China, internal migration influences husbands’ and wives’ marital satisfaction differently (Zhuang et al., 2014). A key factor here is whether and how labor migration aligns with spouses’ gender ideologies. For example, female migrants often challenge historical expectations that women prioritize family emotional care and that men serve as the primary financial providers. Scholars have explored the myriad consequences that female migration can have for married persons (e.g., Acedera & Yeoh, 2019; Hoang, 2011; Hoang & Yeoh, 2011; Paul, 2015). In Nepal, however, the men-as-breadwinner role persists, and even places pressure on men to migrate (Adhikari & Hobley, 2015). Husbands earning higher wages abroad are fulfilling expectations, and this is powerful: studies in Western settings have linked financial stress to marital satisfaction, particularly for husbands (Amato et al., 2003; Conger et al., 1990). Magnifying the financial dimension of migration is, as noted later, the inability to integrate socially in the destination country.

As for migrants’ wives, many studies show that women involved in migration can experience more say in household decisions and access to new income-generating opportunities, including in Nepal (Bhadra, 2007; Gartaula & Vissar, 2010; Kasper, 2005; Maharjan et al., 2012; Yabiku et al., 2010), which may improve how they regard their marriage. However, benefits to migration remain tied to gender systems and this is true across settings (Adhikari & Hobley, 2015; McEvoy et al., 2012). In Armenia and Guatemala, Menjívar and Agadjanian (2007) found that husbands’ migration maintains a division of labor that further reinforces gender inequality: men remain earners, with women in subordinate positions in the household. Similarly, migrants’ wives’ marital satisfaction depends on their gender-type attributes in Mexico (Wood et al., 2015) and their adherence to modern versus traditional gender ideologies in China (Zhuang et al., 2014). So, while migration may alleviate women from financial hardship, their lived experiences often remain constrained by patriarchal power structures—what Choi et al. (2018) refer to as a stalled gender revolution in global migration.

Hypotheses

The theoretical framework discussed earlier helps identify a number of hypotheses regarding the association between male labor migration and spouses’ marital quality in Nepal. This study examines how migration changes couple-level as well as individual-level (husbands’ and wives’) valuations of marital quality, following husband’s migration.

Hypothesis 1: As Nepalese spouses work toward a difficult, but shared goal, migration likely improves spouses’ marital quality (e.g., decline in conflict and more love).

Hypothesis 2: As Nepalese migrant husbands pursue a strategy that fits with local gender norms and involves destinations in which opportunities to explore new networks, partners, and freedoms are extremely limited, migration likely improves their marital quality.

Hypothesis 3: Given the potential range of gains and associated constraints, migrants’ wives may experience improvements (Hypothesis 3a), declines (Hypothesis 3b), or neither (e.g., mutually canceling positive and negative effects) (Hypothesis 3c) in their marital quality.

Data and Methods

Sample

This study uses data from the Chitwan Valley Family Study (CVFS), a panel study in South Central Nepal, to test these setting-specific hypotheses. The study collected data from all households in 151 neighborhoods using a clustered sampling design. Baseline face-to-face individual and household interviews were conducted in 1996 with all household members aged 15–59 years and their spouses, regardless of age or place of residence. Individual interviews in 2008 again assessed marriage-related experiences, including reports of marital quality, similarly followed by family and travel updates for everyone in the household. Following both individual interviews, regular household interviews (HHR) tracked every respondent measuring monthly updates of family transitions and travel events. Individual life history calendars (LHC) in 2008 captured annual retrospective measures of travel, marital status, schooling, and employment.

From the larger panel study sample, husband–wife couples were identified that fit the following criteria: (a) they were married in 2008; (b) both spouses were living in Chitwan and participated in the individual interview in 2008; and (c) were still married in 2014. Because multiple marriages occur, although increasingly less common, only most recent spouses were included in this list (4% of the sample has been married more than once). Using the HHR data, couples were then characterized based on husbands’ and wives’ migration experience between 2008 and 2014, comprising two lists: one consisted of all couples in which neither husband nor wife migrated outside Nepal or India at any point between 2008 and 2014, and both spouses were currently living in Chitwan in 2014 (“nonmigrant” couples). A second consisted of couples in which the husband was currently living and working outside Nepal and India in 2014 (“migrant” couples). International migration destinations are conceptualized as different from movement to India. This is for two reasons. First, the very nature of the movement between Chitwan and these destinations differs: one can take a bus into India, with no paperwork, due to geographic proximity and an open border. In contrast, migrating to Qatar requires a visa, formal labor contract, and significant recruiter fees. Second, reasons for migration between Nepal and India are diverse: individuals study in Delhi, visit their spouse’s natal home, or migrate seasonally. These trips may last anywhere from 1 day to a few months. Travel to Malaysia or the Gulf, however, are rarely for purposes other than work and typically last 2–3 years. The present study aims to understand temporary labor migration arrangements that are inherently longer term, require substantial investment, and with little option of easy return, thus motivating the focus on non-India destinations. Location-specific data in the HHR account for these differences. Previous migration studies in Nepal make a similar distinction (Adhikari & Hobley, 2015).

Next, 130 “migrant” couples and 150 “non-migrant” couples were randomly selected for participation in a second survey questionnaire. So, “migrant” husbands were living and working outside Nepal/India and “nonmigrant” husbands and all wives, regardless of their husbands’ migration status, were living in Chitwan at the time of the 2014 survey. Fifty-eight percent of the sample (31% of husbands and 85% of wives) were surveyed in-person, with the remaining participants being surveyed telephonically (41% of nonmigrant men; 17% and 13% of nonmigrant and migrant wives, respectively). Additional models show that survey mode does not significantly influence martial quality reports.

Migration, marital transitions, and marital quality are tightly linked with age. This study thus focuses on the age range in which the majority of individuals are married, few are divorced, widowhood is rare, and temporary labor migration is most common: 17–40 years at baseline (2008). This results in an analytic sample of 267 couples (138 nonmigrant couples and 129 migrant couples). Additional analyses based on the full sample yield similar results.

Measures

Dimensions of Marital Quality (2008 and 2014).

Dimensions of marital quality assessed in the 2008 interview were assessed again in 2014 with all 534 individuals. Scholars have used these measures to examine marital quality in Nepal previously (Allendorf & Ghimire, 2013; Axinn et al., 2017; Jennings, 2016). Two measures assess marital conflict. Although closely related, these two components reveal distinct characteristics of the spousal relationship. Disagreements indicate interactions in which spouses may not agree, but they experience any negativity equally. Criticism, in contrast, reflects interactions in which one spouse, not both, is the recipient of negativity; any criticism between spouses indicates some level of disrespect. These measures ask how often the respondent has disagreements with and is criticized by his/her most recent spouse, respectively. The measures are coded 1–4, with 1 being frequently and 4 being never. A third measure assesses the strength of their emotional bond, asking how much a respondent loves his/her most recent husband/wife, with 1 being very much and 4 being not at all. This measure is reverse coded so that larger numbers indicate more positive marital quality on all measures. Although conflict and satisfaction are linked, they represent different dimensions of the relationship. In fact, studies show that conflict is relatively stable over time—conflict at one time point often predicts conflict later on—whereas satisfaction is less so (Moore et al., 2004). Additional models including both conflict and satisfaction are robust to those presented here.

Migration (2014).

One dichotomous measure indicates whether the respondent is a migrant husband or has a migrant husband at the time of the 2014 survey. So, “1” indicates that the respondent is either currently living and working away from his wife outside Nepal/India (if a male respondent) or that the respondent has a husband who is currently living and working outside Nepal/India (if a female respondent). Conversely, “0” indicates that the respondent (either male or female) is part of a couple in which both the husband and wife are currently living in Chitwan at the time of the 2014 survey, and that neither has never lived or worked outside Nepal/India since 2008. Couples in which the husband migrated between 2008 and 2014, but had since returned and were living in Chitwan at the time of the 2014 survey, were not included in analyses.

Controls (2008).

Marital characteristics. Analyses include important marital characteristics previously documented to influence migration and/or marital quality. Each of these measures was assessed in 2008. Spouse choice is included as a categorical measure, with one group indicating that the respondent participated in selecting his/her spouse along with relatives and another group indicating the respondent selected his/her spouse independently (reference: no participation). Higher quality marriages are likely to be positively correlated with greater say in spouse choice, as these unions are based more on individuals’ emotional connection than on parents’ arrangement; due to historically patriarchal norms, husbands tend to have greater say in this decision (Allendorf & Ghimire, 2013). A continuous measure reflects respondents’ age at first marriage. Individuals who enter marriage later are likely to hold different views and attitudes toward marriage, specifically that marriage is a union based on love and personal fulfillment. Models also include a dichotomous variable indicating whether the couple has any children, which has been linked to migration and marital quality (Allendorf & Ghimire, 2013; Lu & Treiman, 2011). Last, duration of marriage has been documented to influence marital quality (VanLaningham et al., 2001).

Other important factors. Human capital influences an individual’s income opportunities and expected earnings, thereby affecting one’s motivation to seek work abroad (Stark & Bloom, 1985), as well as the timing and characteristics of marriage. As a key source of human capital, models include a continuous measure indicating an individual’s educational attainment in 2008 (Massey et al., 2010). Models also include a dichotomous measure of whether the respondent works for wages in 2008, which represents a key occupational divide in this setting. Models include travel history prior to 2008, with “1” indicating any experience outside Nepal/India before 2008. Migrants transmit information about the process and experience, thereby helping to alleviate concerns associated with moving abroad and increasing the probability of doing so (Massey et al. 1998). Key sources of information are family (Massey et al., 2010) and friends (Bohra & Massey, 2009, p. 632). Models include a dichotomous measure indicating whether the respondent’s family or friends have traveled outside Nepal/India by 2008.

Demographic characteristics. Significant ethnic/caste differences have been evidenced to influence social life in Nepal (Ghimire et al., 2006). Models include a set of dichotomous measures corresponding to five broad ethnicity/caste categories reflecting meaningful distinctions in Nepalese society. Descriptive statistics for control variables are presented in the bottom rows of Table 1. Respondent age in 2008 is highly correlated with the duration of their marriage (r = 0.69 for males and r = 0.81 for females). Due to issues with multicollinearity, models do not include both measures. Models accounting for respondents’ age in 2008 yield similar results.

Table 1.

Descriptive Statistics for Marital Quality and Key Covariates Used in Analyses

	Husbands				Wives
	Nonmigrant (N = 138)		Migrant (N = 129)		Nonmigrant (N = 138)		Migrant (N = 129)
Measure	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Marital quality, Time 1 (2008)
Disagree (1–4)b	3.17	0.55	3.16	0.69	2.98	0.71	3.18	0.62
Spouse criticize (1–4)	3.28	0.65	3.47	0.64	3.44	0.67	3.50	0.71
Love spouse (1–4)a	3.18	0.69	3.30	0.58	3.36	0.66	3.43	0.67
Marital quality, Time 2 (2014)
Disagree (1–4)a	3.06	0.56	3.20	0.56	3.05	0.56	3.09	0.63
Spouse criticize (1–4)a	3.25	0.68	3.47	0.53	3.34	0.57	3.47	0.60
Love spouse (2–4)a	3.27	0.53	3.45	0.51	3.36	0.64	3.50	0.55
Marital characteristics (2008)
No choice in recent spouse	0.18		0.23		0.46		0.57
Joint choice in recent spouse	0.29		0.27		0.17		0.14
Full choice in recent spouse	0.53		0.50		0.36		0.29
Age at first marriage (10–32)a	22.99	3.51	21.67	3.38	18.64	2.83	18.06	2.51
Any children	0.93		0.88		0.93		0.86
Duration of marriage (0–19)	7.64	4.08	7.41	4.68	7.69	4.27	7.44	4.48
Other important factors (2008)
Educational attainment (0–16)	8.43	4.07	7.74	3.12	7.00	3.71	6.68	3.43
Wage worka	0.23		0.12		0.38		0.37
Travel outside Nepal	0.34		0.29		0.04		0.08
Family or friend travel experienceb	0.91		0.94		0.72		0.87
Caste/ethnicity
Brahmin/Chhetri	0.44		0.45		0.43		0.45
Dalit	0.10		0.14		0.09		0.14
Newar	0.19		0.22		0.19		0.22
Terai Janajati	0.07		0.02		0.07		0.02
Hill Janajati	0.20		0.16		0.22		0.16

Note: Dimensions of marital quality measures are coded from 1 to 4, with higher values indicating higher quality marriage. Ranges in parentheses.

^aIndicates statistically significant difference between migrant husbands and nonmigrant husbands

^bindicates statistically significant (p < .05) difference between migrants’ wives and nonmigrants’ wives; two-sample t-tests.

Analytic Strategy

This study uses multivariate linear regression to test the effects of husbands’ international migration on 2014 dimensions of husbands’ and wives’ marital quality, controlling for the same dimensions in 2008, as well as marital characteristics, key factors assessed at baseline, and caste/ethnicity. Models cluster the errors by neighborhood to account for the clustering of the CVFS sampling design. This lagged outcome model allows for estimation of the effects of husbands’ migration on the change in spouses’ marital quality between 2008 and 2014, given that the earlier dimensions are held constant. It can be represented as follows:

$$Y=b_0+\sum (b_k)(X_k)+\epsilon ,$$

where Y is the predicted value of the dimension of marital quality in 2008, b₀ is a constant term, X_k are the explanatory variables in the model, and ε is the error term. Tables present estimates for b_k, which represent the mean change in 2014 marital quality for one unit of change in x while holding all other factors constant, including the same respective dimensions of marital quality measured exactly the same way in 2008 (Axinn & Thornton, 1992). The main coefficient of interest—a husband’s international migration status in 2014—thus represents the difference in mean changes in 2014 marital quality between spouses in a “migrant” marriage (i.e., the husband is currently living and working outside Nepal/India) and spouses in a “nonmigrant” marriage (both husband and wife are currently living in Chitwan, Nepal).

It important to note the sample size: 267 couples, or 534 individual spouses. While relatively small, the study design is strong. Couples were randomly selected from a larger panel study, allowing for repeated assessment of the same dimensions of marital quality among the same husbands and wives, 6 years apart. Moreover, the fact that the study consists of migrant and non-migrant couples allows analyses to account for changes in marital quality that occurs over time, regardless of whether a husband migrates or not. This difference-in-difference approach has been used in other migrant studies, such as the Mexican Family Life Survey (e.g., Antman, 2011). Importantly, descriptive statistics presented in Table 1 show that migrant spouses do not significantly differ from nonmigrant spouses on most key covariates (indicated in superscript; two sample t-tests). Related to the small sample size is the analytic inability to include a broad array of covariates in one model. The CVFS allows models to account for many potentially confounding variables, including those related to socio-economic status—a key factor associated with both a husband’s propensity to migrate and spouses’ marital quality. Additional models including these (e.g., whether the household owns farmland, neighborhood migration experience, ownership of a business, salary, or military work experience) yield similar results. Analyses presented here focus on those factors most influential for migration and marriage.

An Additional Note on Selection.

Marital dissolution is an important indicator of marital quality that is not accounted for in analyses. This selection criterion poses a significant challenge to the study of changes in marital quality in all settings. To understand the potential threat of dissolution to the association of interest, additional analyses examined divorce using the full CVFS sample in 2008. Analyses focused on respondents aged 17–40 and married in 2008 (N = 3,217; 1,132 males and 2085 females) to examine the risk of divorce among a sample similar to that used for this study. Roughly 1% of this larger sample experienced a divorce between 2008 and 2014 (N = 43 divorces). Death of a spouse was even rarer: less than 1% became a widow (N = 27 deaths). Still, while divorce remains rare, it is increasing and has been linked to reports of marital discord in Nepal (Jennings, 2016). Additional analyses using this full CVFS panel study data examined the relationship between husbands’ migration and marital dissolution. Results show that being an international migrant (Model 1), having any international migration experience (Model 2), the number of international migration trips (Model 3), and/or the cumulative number of months as an international migrant (Model 4) do not significantly increase a husband’s odds of dissolution.

A second selection issue is the nature of the marital relationship prior to a spouse’s migration. Fortunately, a large literature has identified a number of factors influencing spouses’ marital quality. Models test for the predictive power of these factors, assessed in 2008—marital characteristics, educational attainment, financial status, and caste/ethnicity—on spouses’ marital quality in 2008. Results, shown in Table A1 in the Appendix, are similar to those found in previous studies (see Allendorf & Ghimire, 2013; Hoelter et al. 2004 for more information on the determinants of marriage in this setting). Models account for these 2008 factors as well as 2008 migration experience. However, given that the focus of this study is on the change in marital quality between 2008 and 2014, results presented emphasize the main coefficient of interest—husbands’ migration status—and include spouses’ marital quality in 2008 as well. Results from models including varying combinations of key covariates are robust to those presented.

Results

Bivariate

Table 1 displays descriptive statistics for measures used in models, stratified by respondent gender, then further by husbands’ 2014 migration status. Statistics for husbands are in the left most columns, with nonmigrant husbands’ to the left of migrant husbands’; statistics for wives are in the right most columns, with non-migrant husbands’ wives’ to the left of migrant husbands’ wives’. Marital quality measures in 2008 are presented in the top most rows and those assessed in 2014 are below these. Again, larger numbers represent higher marital quality (e.g., fewer disagreements, less criticism, more love).

Regarding marital quality across the two time points, respondents reported high marital quality, with nearly all means measuring above 3 on a scale from 1 to 4 and only 1–2% of respondents reporting frequently or not at all for each of the dimensions. Looking across migration status, migrant spouses reported higher marital quality on five out of six marital quality measures in 2008 than did nonmigrant spouses (all but husbands’ reports of disagreement). This further supports the need to examine marital quality not just while the spouse is away, but to account for it prior to migration as well. Migrant couples also reported significantly higher marital quality on all three dimensions (a total of six measures, three for each spouse) in 2014 than did nonmigrant couples.

Across gender, husbands and wives experienced their marriages differently. Overall, wives reported more disagreements than husbands did, but less criticism and more love, in both 2008 and 2014. However, consistent with above, husbands’ and wives’ marital quality varied according to 2014 migration status. Among husbands (the left-most columns), migrants reported higher marital quality than did nonmigrants for all three measures in both 2008 and 2014. These differences are statistically significant for love in 2008 and for all three dimensions in 2014 (p < .05, noted with a letter “a”). Among wives (the right-most columns), migrants’ wives reported higher marital quality than non-migrants’ wives for all three dimensions in both 2008 and 2014, although only frequency of disagreements in time 1 is significant (p < .05, noted with a letter “b”).

Multivariate Models

Hypothesis 1 (spouses’ marital quality improves).

This association is tested first at the couple-level (mean of husbands’ and their wives’ reports of the same dimension of marital quality). Results, not presented for parsimony due to their similarities with subsequent models, reveal two patterns. One, both dimensions of marital conflict—frequency of disagreement and of criticism—assessed in 2008 are strong predictors of the same respective dimension in 2014 (p < .001), whereas an earlier assessment of love for spouse is not. Two, net of these 2008 marital quality measures, migrant couples reported less criticism (p < .05) and more love for each other (p < .05) than did nonmigrant couples. That migrant couples reported higher marital quality on two distinct dimensions of marital quality supports Hypothesis 1.

Hypothesis 2 (husbands’ marital quality improves).

Table 2 presents results from models teasing apart this couple-level effect and testing the effect of husbands’ international labor migration on changes in each of the three dimensions on husbands’ and wives’ marital quality, respectively. Again, dimensions of marital quality are coded so that higher values indicate higher marital quality. A positive coefficient for “migrant” thus represents a positive effect, or higher marital quality, for couples engaged in migration than those who are not. These measures are standardized to account for skewness and allow for comparability across measures. All models shown in Table 2 control for marital characteristics, other important factors, and caste/ethnicity.

Table 2.

Models of the Relationship between Husbands’ Temporary International Labor Migration and Husbands’ and Wives’ Marital Quality in 2014 (standardized)

	Husbands			Wives
	Fewer disagreements (1)	Less criticism (2)	More love for spouse (3)	Fewer disagreements (4)	Less criticism (5)	More love for spouse (6)
Husbands’ migrant status
Is a migrant (ref: not a migrant)	0.21*	0.27*	0.25*	0.04	0.16	0.22†
	0.11	0.13	0.12	0.14	0.12	0.12
Spouses’ marital quality in 2008, standardized (larger values = higher marital quality)
Disagree	0.30***			0.15*
	0.07			0.07
Criticism			0.35***		0.17***
			0.06		0.05
Love for spouse			0.08			0.13*
			0.06			0.06
Constant	−0.10	−0.25	1.16**	−0.30	−0.35	−0.27
	0.64	0.73	0.48	0.63	0.59	0.57
p-value	.00	.00	.00	.00	.00	.00
R	13.2	14.5	9.9	10.1	10.2	16.7
N	267	267	267	267	267	267

Note: Estimates represent results from linear regression models, with SEs. All models control for marital characteristics and other important factors, assessed in 2008. Marital quality measures are standardized; greater values indicate higher quality marriage.

^†p < .10

^*p < .05

^**p < .01

^***p < .001; two-tailed tests.

Results show that effects at the couple-level discussed earlier were largely driven by husbands’ experiences. Previous assessments of both dimensions of marital conflict (2008) significantly predicted the same respective dimension in 2014: husbands who reported fewer disagreements and less criticism in 2008 also reported fewer disagreements and less criticism in 2014 as well (p < .001; Models 1 and 2). Net of 2008 marital quality, however, husbands reported fewer disagreements (Model 1; p < .05), less criticism (Model 2; p < .05), and more love for their spouse (Model 3; p < .05). Additional models including 2008 reports of all dimensions of marital quality predicting individual dimensions of marital quality in 2014 (see Axinn et al., 2017) yielded similar results. These results support Hypothesis 2.

Hypotheses 3a, 3b, and 3c (wives’ marital quality improves/declines/neither).

Models 4–6 in Table 2 (rightmost columns) display results from models for wives. Wives’ 2008 reports of disagreement, criticism, and love all positively and significantly predicted these same reports in 2014. Net of these early reports, migrants’ wives reported higher marital quality than non-migrants’ wives on all three dimensions in 2014, although only love for spouse was marginally statistically significant (p = .10; Model 6). Of note, however, was the magnitude of the effect of the latter and its similarity to that of the effect for husbands: both migrants and their wives reported nearly one-fifth a standard deviation more love for their spouse in 2014, net of 2008 reports, than do nonmigrants. Results provide limited support for Hypothesis 3a (higher marital quality) and more support for Hypothesis 3c (benefits and costs cancel each other out for a null effect).

A Deeper Look: Husbands.

Additional analyses first dive deeper into the positive effect for husbands. In 2014, 77% of migrant husbands were in the Middle East (N = 99); 23% were elsewhere (N = 30). Of this latter group, 87% were in East Asia (92% of whom were in Malaysia), with the remaining in Afghanistan and Iraq. As far as migration experience between 2008 and 2014, migrant husbands averaged 1.9 trips and 42 months away during this period.

Financial Investments

Although the 2014 survey did not directly assess remittances, it did ask respondents about investments since the time of the husband’s migration/since 2008 for migrant/nonmigrant spouses, respectively. Table A2 in the Appendix presents descriptive statistics of investments by migrant status. Results suggest that migrant households were more likely to have invested in land, a house plot, building (all significant at the p < .001 level), farming, and private schooling for their children’s education (both p < .01), and less likely to have invested in livestock and business (both p < .01). Of course, these items were not repeated assessments and do not capture baseline financial status, although do provide insight into whether the ability to and the experience of making substantial investments help explain the observed benefits for husbands. Inclusion of these investment items in linear regression models resulted in none of the dimensions of marital quality remaining statistically significant (see Table A3 in the Appendix for full model results), suggesting investments may explain some of the migration effect. Results from separate models suggest that some investments were less powerful predictors—investing in business did not explain migrant husbands’ reports of less criticism, and investing in a houseplot or livestock did not explain more love for their wives—although these results should be interpreted with caution.

Destination. Expanding on the financial component to capture larger social and cultural contexts is migrant destination. Nepali migrants often live and work abroad for extended periods of time due to the nature of foreign labor contracts and the upfront investment of international travel. Results from additional models, presented in Table A4 in the Appendix, show that the main effects presented in Table 2 did indeed vary by destination. Specifically, men in GCC member states reported significantly less criticism and more love for their spouses than do nonmigrants (both p < .05; Models 2 and 3, respectively). However, husbands in other destinations, largely Malaysia, reported significantly fewer disagreements than do nonmigrant husbands (p < .05; Model 1). Worth noting, however, were the similarities in magnitude between the effects for both destinations with regard to husbands’ reports of criticism (Model 2) and love for spouse (Model 3): they were nearly identical in magnitude but with differing SEs due to varying sample sizes. Relatedly, no dimension of marital quality significantly differed between migrant groups.

Number of trips. It is possible that the number of migrant trips—in this case, signing longer-term labor contracts very far away—may influence husbands’ marital experiences as well. Analyses leveraged highly detailed monthly household registry data to create a continuous measure indicating the number of trips migrants made to any destination between 2008 and 2014. Results from additional models (Table A4) show that husbands who had made more trips reported less criticism and more love for their spouse than did husbands who had made fewer or no trips (p < .05, Model 2; p < .01, Model 3, respectively). Again, the effects of the number of trips were not significant between migrant groups, only when compared to nonmigrants.

Number of months away. Related is the cumulative number of months spent abroad between 2008 and 2014. Unlike the number of trips, however, the number of months is an indication for the amount of time spouses spent apart during this period. Studies show that extended periods of geographic separation can negatively impact spouses’ well-being (e.g., Farooq & Javed, 2009; Wilkerson et al., 2009), ultimately weakening an emotional connection. Of course, the less time spouses spend together, the less they are at risk for conflict as well. Results from Models 4–6 in Table A4 suggest that time spent away may help explain the effect of migration on husbands’ marital quality: more months away, regardless of destination, did significantly improve husbands’ reports of criticism and love for their spouse.

A Deeper Look: Wives.

Presence and relationships with in-laws. Additional analyses next examined the weaker (null) finding for wives by first analyzing whether or not their in-laws were alive at baseline. Because of the historical patrilocal norms in this setting, the wives are overwhelmingly likely to live with or near their in-laws, and relationships with them are historically tense. Analyses also considered whether wives’ relationships with their in-laws at baseline moderates or mediates the findings presented in Table 2. Neither set of measures was significantly associated with marital quality, although one finding is noteworthy. A categorical measure constructed using husband’s migrant status and in-law(s)’ living status found that migrants’ wives whose mother-in-law was not alive in 2008 reported more love (p < .01), and those whose father-in-law was not alive reported less criticism (p < .01) and more love (p < .05), than nonmigrants’ wives whose respective in-law was alive.

Fertility experience. Analyses next examined fertility experiences using baseline assessments of motherhood status and number of children. Analyses also leveraged the monthly household registry data on women’s reports of pregnancy and pregnancy outcomes. Twenty-eight percent of wives experienced a live birth during the observation period; this experience did not significantly differ by husband’s migrant status. Results from all models were similar to those presented in Table 2, with no fertility measures statistically significant.

Time use. Finally, analyses utilized supplemental data collected as part of the larger panel study in 2015, which is outside the observation period of interest for this present study. Nevertheless, analyses merged Women’s Time Use Study data with the present study data, as well as the larger household panel study in order to identify husbands’ migration experience and status between their 2014 interview and the 2015 women’s time use interview. Importantly, only 20 migrant husbands were still migrants at the time of the 2015 time use survey. Results from two sample t-tests, which should be interpreted with caution, suggest that migrants’ wives spent marginally less time caring for children and/or elderly people (0.74 vs. 1.82 hours/day; p < .10) and significantly more time in leisure (e.g., watched TV, read, on Facebook; p < .05) than did nonmigrants’ wives. These results are consistent with studies suggesting that remittances provided may mediate the burden on wives, thereby increasing autonomy and enabling them to spend their time on other activities (Maharjan et al., 2012).

Discussion

Research has identified marital quality as a useful lens through which to examine relationship change, as well as its key correlates. Despite strong interest, however, how it relates to temporary international labor migration remains understudied. Previous research finds that labor migration increases risk of divorce and of infidelity, whereas research on marital quality, much of which is cross-sectional, is more mixed. That little consensus exists is not surprising given the challenges associated with assessing marital quality over time, particularly among the same spouses, and even more so when the marriage crosses international borders. Additional concerns regard biases stemming from selection into marriage and migration. This study addresses these gaps in two ways. The first is analytic: it randomly selects couples from a larger, detailed panel study, leveraging repeated measures of multiple dimensions of marital quality among the same couples across 6 years, during which time half of the sample experienced the international migration of the husband, and half did not. The second is blending existing literatures to approach this association: acknowledging key selection factors, and using them to re-examine the influences of labor migration on the marital relationship in new light.

Specifically, this study identifies and tests setting-specific hypotheses in Nepal. The setting is useful to address these questions given the shift toward international labor markets (and subsequent increase in labor migration abroad in response to the significantly higher wages) and concurrent changes in the marital relationship in recent decades (e.g., a shift toward love marriages). Moreover, marriage remains largely universal and divorce rare, which is in stark contrast to Western settings, where high divorce rates have led to considerable bias in estimates of couples’ marital quality (Cohen, 2014). Results from linear regression models show that transnational marriages reported higher marital quality than nonmigrant marriages, net of marital quality assessed 6 years earlier. Importantly, migrant husbands report higher marital quality on multiple dimensions than do nonmigrant husbands. The impact on wives’ marital quality, however, is less clear: while they also report higher marital quality than non-migrants’ wives, only their love for spouse is (marginally) statistically different from zero.

These results—which counter existing assumptions and vary by gender—help identify three conditions shaping this association. The first is social and structural forces supporting men as breadwinners. Regarding husbands, results show that living and working away from home improves two fundamentally different dimensions of their marital relationship. These results are consistent with literatures linking gender, work, and marriage: in settings like Nepal, migration is an increasingly viable and desirable livelihood strategy and, given enduring historical social norms, households are overwhelmingly likely to send an adult male. As such, migrant husbands’ ability to send money home fulfills gendered expectations of the male provider role, which has been linked to marital quality in high-income settings (e.g., Killewald, 2016). Additional analyses presented in the Appendix are consistent with this finding: migrant households are significantly more likely to have made major investments during the observation period than nonmigrant households and, although only assessed in 2008, regression results suggest that financial investments do indeed influence husbands’ marital quality. Additionally, the positive effects of migration are stronger for migrants in regions in which wages are higher (GCC member states), as well as for those who have made more trips and spent more months abroad, suggesting that “successful” migration—staying longer and/or repeating migration—may help explain the benefits. Of course, the effects are not statistically significant across migrant groups, suggesting that, beyond financial gains, migrants’ ability to contribute to household income in general may provide migrants a sense of “mattering” (Afulani et al., 2016; Ambugo & Yahirun, 2016).

Regarding migrants’ wives, they historically enjoy little autonomy in this setting: they marry early and devote much of their time caring for family, rather than pursuing non-family activities, such as education and work. Scholars often identify remittances from a migrant husband as a potential pathway through which women may enjoy new financial and social opportunities (e.g., Lu 2012; Yabiku et al., 2010). But studies also note limitations to these benefits, which may net the positives out. For example, any realized gains for women in these domains must be balanced with the loss of an able-bodied adult who can no longer help in daily household management. Although limited, preliminary findings from a time use survey implemented to a small subsection of respondents in 2015 (outside the observation period) suggest that migrants’ wives spend less time providing care for others and more time in leisure. This finding supports the assertion that remittances may mediate wives’ burdens, thereby enabling them to spend their time on other activities. At the same time, the nature of temporary labor migration means that any changes in autonomy may not be permanent, but only last as long as a husband is away. This is consistent with other studies in low-income settings, which found that husbands’ migration can maintain a division of labor that further reinforces gender inequality: men remain earners, with women in subordinate positions in the household (e.g., Menjívar and Agadjanian (2007)), as well as with literatures examining commuter marriages in high-income settings (e.g., Bergen et al., 2007). The converse follows here as well: studies in settings in which there is a feminization of migration or household management show that conflict often arises, with consequences for key social relationships. This is consistent with additional analyses that suggested, to some extent, that the presence of in-laws may be constraining potential benefits to wives.

The second is a strong husband–wife bond facilitating a spouse’s migration. Part of the challenge in studying this association is the unevenness in marital quality among couples to begin with, prior to migration. Although this study design allows for analyses to control for this empirically, it is likely that this baseline marital dynamic influences not just the likelihood that the couple engages in temporary international migration, but that this—a strong foundation of love, respect, commitment—influences the way husbands and wives experience migration with regards to their marriage as well. Studies highlight the coordination required to successfully pursue and maintain a migrant spouse abroad; separate studies show that high functioning couples are able to pursue high-risk, high-reward strategies in the short-term if they lead to long-term gains. As the institution of marriage continues to shift worldwide, married international labor migrants may be more likely to belong to more highly functioning marriages than before.

The third condition is marriage-protective social environments at both ends of migration. In Nepal, as a sending country, social norms dictate that marriage is a lifelong commitment, although divorce is growing slightly more common. The emphasis on marriage, coupled with an emerging culture of migration, creates an environment in which wives “left behind” may not feel unique or vulnerable in their circumstance, but rather connect with others and receive support from the larger community in times of need. On the receiving end, common destination countries, primarily in the Middle East, maintain restrictive migrant monitoring systems, which, coupled with the region’s historically patriarchal and family-centric norms, greatly limit migrants’ opportunities to engage with the local community—the opposite of what occurs among married Nepali migrants in India (Poudel et al., 2004). Destination-specific results in the Appendix support this assertion.

As with any study, limitations exist. Of primary concern is the small sample size. Although the study design here is strong, a larger sample size would bolster these findings and allow for deeper examination of the mechanisms considered in supplemental analyses here. Nevertheless, results here are consistent with the theoretical reasoning and provide a strong foundation for future work. In particular, studies should apply this framework to marital quality among spouses who migrate to diverse destinations, such as India, as well as upon migrants’ return. Future work should also connect these findings with related literatures of migrants’ success (e.g., Agadjanian & Hayford, 2018), with particular focus on remittances and subsequent investments. Additionally, discord—differences in husbands’ and their wives’ reports—is an additional dimension of marital quality that warrants examination in future work. Last, studies should account for health factors, both as a key covariate and as a potential mechanism behind the largely null findings among wives. Specifically, studies have shown that family separation can be more detrimental for female compared to male migrants (Gutierrez-Vazquez et al., 2018), suggesting that such gendered differences in mental health may help capture the nuances of migrants’ wives’ experiences left behind as well.

In sum, a deeper understanding of how an increasingly common livelihood strategy influences the marital relationship informs current understandings of marital processes. Yet empirical challenges surrounding selection and data collection have greatly limited our ability to study this association. This study leverages a strong study design in a unique setting in which both marriage and migration have shifted in recent decades. Moreover, it blends literatures and considers how factors selecting married individuals into migration continue to shape how couples experience the migration even as the spouse is away. As wage disparities widen across borders and the global population becomes more mobile, the prevalence of husbands and wives living and working apart is likely to grow ever more common. Identifying factors shaping this association—social and structural forces supporting men as breadwinners (clear gender–work preferences), a strong husband–wife bond facilitating husbands’ migration, and marriage-protective social environments at both ends of migration—is particularly timely.

Supplementary Material

Appendix

Appendix S1: Supporting Information