Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 8.
Published in final edited form as: J Fam Issues. 2017 Mar 16;39(5):1236–1257. doi: 10.1177/0192513X17698184

Men’s Migration, Women’s Autonomy, and Union Dissolution in Rural Mozambique

Victor Agadjanian 1,**, Sarah R Hayford 2
PMCID: PMC6687330  NIHMSID: NIHMS881726  PMID: 31395999

INTRODUCTION

Research in a variety of contexts suggests that migration increases the risk of union dissolution (e.g., Boyle et al. 2008; Frank and Wildsmith 2004; Landale and Ogena 1995; Muszynska and Kulu 2007). These studies typically point to the strain caused by spousal separation as the underlying cause of this greater risk. However, migration also has the potential to increase earnings and household wealth, and improved economic conditions generally reduce the risk of divorce (e.g., Ono 1998; South and Lloyd 1995). The exact mechanisms of the influence of migration on marital stability are still not well understood, and the relative importance of spousal separation and economic contributions are likely to vary depending both on the specific nature of the migration experience and marriage relationship and the broader context of migration and marriage systems. In today’s sub-Saharan Africa, where marital instability is high (Clark and Brauner-Otto 2015; Reniers 2003) and where temporary or circular male migration is a widely-exercised household economic strategy and an accepted or even expected part of marriage (Agadjanian 2008), the relationship between migration and marital dynamics is a particularly important topic for investigation.

In this study, we use longitudinal data to examine the complex associations between male labor migration and union dissolution in rural southern Mozambique, a setting where rates of male labor migration are high and the economic returns to such migration are highly variable. We move beyond previous research that treats all migrants equally in assessing the impact of migration on marital stability. Instead, following microeconomic models of marriage that emphasize the importance of spouses’ financial contributions to the household (Becker 1981; Oppenheimer 1997), we use a more nuanced measure of migration that accounts for variation in the economic benefits of men’s migration for their left-behind household members. In addition, we integrate women’s decision-making autonomy into our conceptual framework in order to better understand how broader gender dynamics may condition the impact of migration on marriage outcomes.

SETTING

This analysis is based on data from a longitudinal survey carried out in rural areas of Gaza province in southern Mozambique, one of the poorest countries in the world. The study area is characterized by low-yield subsistence farming, with limited opportunities for non-agricultural employment. As in other parts of sub-Saharan Africa (Locoh and Mouvagha-Sow 2008; Weisner et al. 1997), the institution of marriage in the area has undergone considerable change in recent decades (Arnaldo 2004). However, the economic and social functions of marriage continue to be central to societal reproduction. The lineage system in southern Mozambique, as in much of southern Africa, is patrilineal. Traditional marriage there is typically virilocal and bridewealth-based: women move to their husband’s village on marriage, often into his parents’ compound, and children are considered part of their father’s family and stay with him in the case of marital dissolution (Junod 1962). Although women participate intensively in agricultural production and make substantial contributions to household maintenance, men are considered to have primary responsibility for the financial stability of the family, particularly for wage labor. Male labor migration has both reflected and strengthened this gendered division of labor (Loforte 2000).

Male labor migration from rural southern Mozambique has been going on for generations. While some local men look for work in the nation’s cities, especially the capital Maputo, most labor migration from rural southern Mozambique has been to the Republic of South Africa, Mozambique’s much more prosperous neighbor. Initially, that migration was directed toward South Arica’s mining centers (Crush, Jeeves, and Yudelman 1991). Both the entry of Mozambicans into South Africa and their employment in its mining sector were strictly regulated, and men who were recruited to work in the mines could rely on relatively stable income, part of which was transferred by employers to migrants’ families directly. The dismantlement of the South African apartheid regime in the middle of the 1990s led to both a gradual liberalization of border crossing and increased political pressures within South Africa against using migrant labor. As a result, labor migration to South Africa from neighboring countries, including Mozambique, has increased in volume yet has also shifted toward the informal economy (Crush and Frayne 2010; de Vletter 2007). For the growing number of Mozambican migrants who are employed in the informal sector, often illegally, jobs and earnings are insecure and unpredictable. At the same time, the temporal pattern of male migration has remained rather consistent: migration is a multiyear practice, and because agricultural work rests almost entirely on women’s shoulders, migrant men are typically absent for most of the year, returning home for brief periods around Christmas and New Year and sometimes for Easter (Agadjanian, Menjívar, and Cau 2013). Finally, while male labor migration in this setting has been normative and widespread, women’s migration for work remains uncommon.

THEORETICAL FRAMEWORK

Decisions about marital dissolution reflect a complex process that incorporates assessments of marital quality or satisfaction, evaluation of financial benefits of marriage, and possible attempts to change aspects of the relationship or distribution of resources (Amato 2010; Brines and Joyner 1999; Gottman and Notarius 2000; Lyngstad and Jalovaara 2010). In our conceptualization, migration influences all the three components of this process.

Migration-induced strains and marital stability

In the study site, men’s labor migration typically entails prolonged separation of migrant men from their non-migrating marital partners. Physical separation of spouses limits contact and communication between them, reduces emotional intimacy, and makes it difficult to provide social, psychological, and practical support. Hence, men’s migration may reduce positive social support and marital satisfaction, thereby increasing the risk of marital dissolution (Amato and Rogers 1997; Bradbury and Karney 2004; White and Booth 1991). Furthermore, living apart makes it easier for both husband and wife to form alternative partnerships, which in turn may weaken the spousal relationship and increase risks of marital dissolution (Luke 2010; South and Lloyd 1995). Our first hypothesis posits:

Hypothesis 1: Women with migrant husbands will be more likely to experience marital dissolution than women with non-migrant husbands.

Diversity of migration’s economic outcomes and risks of marital dissolution

According to microeconomic theories of marriage, decisions to get or stay married are motivated by financial as well as emotional concerns (Becker 1981; Oppenheimer 1997). These theories argue that women choose husbands at least in part based on men’s earning power, and that divorce occurs when spouses believe they can improve economic wellbeing by separating or by finding a different partner. Applied to our setting, this argument would imply that migration has the potential to lower divorce rates, but only if it generates sufficient economic returns for the migrant’s household. Previous research in this setting has demonstrated variation in the association between migration and several family-related outcomes based on the economic success of the migrant as assessed by non-migrating family members (e.g., Agadjanian, Arnaldo, and Cau 2011; Agadjanian, Yabiku, and Cau 2011; Yabiku, Agadjanian and Cau 2012). Building on this research we expect the association between migration and marital outcomes to vary according to migration success. Specifically, we posit:

Hypothesis 2: Migrant’s economic failure will be associated with an increased risk of marital dissolution, whereas migrant’s economic success will be associated with a decreased risk of marital dissolution.

Note that Hypothesis 1 and Hypothesis 2 may both hold if both mechanisms (physical separation and economic outcomes) are at work in this setting. Alternatively, if only one mechanism is at work or if one is substantially stronger than the other, only one of the two hypotheses may be supported.

The role of women’s autonomy

In order to understand divorce, it also necessary to consider women’s overall ability to make and carry out decisions for themselves, an ability that is traditionally limited in this male-dominated and gender-stratified context (Loforte 2000). Women’s decision-making autonomy is a complex and often elusive notion (e.g., Bloom, Wypij, and Das Gupta 2001; Jejeebhoy 2000). While women’s autonomy may be influenced by their schooling, outside-the-home labor force participation, and access to their households’ economic resources, it is not reducible to these characteristics (Yabiku, Agadjanian, and Sevoyan 2010). Instead, autonomy implies the capacity both for self-determination and for influencing others to effect change in relationships or behaviors.

Research on divorce has argued that women’s independence (financial or otherwise) may increase the risk of divorce by improving women’s ability to support themselves without a husband or to find a better partner (e.g., Ruggles 1997; Takyi and Broughton 2006). That is, women’s autonomy may increase divorce rates by improving women’s alternatives to marriage. However, this association is complex and contingent on other individual and relationship characteristics. Most notably, autonomy increases divorce risk only in cases where women are unhappy in their marriage (Kalmijn, de Graaf, and Poortman 2004; Sayer and Bianchi 2000). Furthermore, independence and status also increase women’s voice and power within a relationship, allowing them to negotiate changes in spousal behavior or distribution of resources (Katz 1997; Lundberg and Pollak 1996). Research showing that women’s autonomy is associated with higher divorce rates has primarily been carried out in Western contexts where women have economic opportunities and social identities outside of marriage. However, in contexts where women’s social position is primarily determined by marriage and family relationships and where employment options outside subsistence agriculture are very limited, women may strongly prefer to stay married, and thus may seek to influence men’s behavior within a marriage rather than leave the marriage. Consistent with this possibility, some research in sub-Saharan Africa has found that women’s autonomy is associated with lower divorce rates (e.g., Hertrich 2014). This association, however, is contingent on the marital partners’ communication. Qualitative research in our study setting has shown that husband’s migration greatly constrains spousal communication, especially on such sensitive matters as marital fidelity or HIV risks (Agadjanian, Menjívar, and Cau 2013). The ability of migrants’ wives to negotiate with their husbands and to effect changes within marital relationships is therefore also likely to be limited compared to that of non-migrants’ wives. Of course, the fate of both migrant and non-migrant marriages in this patriarchal setting is determined largely by men’s decisions; yet, prior qualitative research suggests that women can exercise considerable agency and influence in shaping marital outcomes.

To summarize, building on existing research literature, we propose that (a) women’s autonomy is most salient for marital stability when marital satisfaction is low; and (b) women’s autonomy is more likely to result in divorce when husbands are migrants, and may even be positively associated with marital stability when husbands are present. Consistent with hypotheses 1 and 2, we assume that marital satisfaction is lower for women married to migrant husbands and especially for migrant husbands who are not economically successful. Hence, we hypothesize that:

Hypothesis 3: Among women married to migrants, autonomy will be associated with increased risks of marital dissolution, whereas among women married to non-migrants, autonomy will be associated with reduced risks of marital dissolution.

Hypothesis 4: The positive association of women’s autonomy with divorce among women married to migrants will be stronger for women married to unsuccessful migrants.

DATA AND METHODS

The survey on which this analysis is based was carried out in 56 randomly selected villages in four districts of Gaza province. In the first wave of the survey, conducted in the middle of 2006, 1680 women (30 women in each village) in marital union aged 18–40 were interviewed. The notion of marriage in this setting is fluid and its complete formalization even within the customary, bridewealth-based system, is a prolonged process (cf. Meekers 1992). Because of this fluidity, an inclusive definition of marital union was applied in the survey: any woman who, in her own words, had a husband, was considered to be in a marital union regardless of whether that union was a customary (bridewealth-based) or legal marriage (very rare in the study setting) or whether it was what in the western context is usually referred to as cohabitation. Accordingly, we use the terms “married” and “being in union” interchangeably.

In each village, the survey sample was designed so as to assure a balanced representation of women married to migrants (regardless of destination) and non-migrants. The village population was first canvassed to identify households with women married to migrants and women married to non-migrants. Each of the two categories of households constituted a separate sampling frame from which fifteen households were drawn randomly. In each selected household, one married woman of eligible age was selected for a face-to-face survey interview. The total sample consisted of 1678 women; 41% of them were married to migrants (more than 80% of whom worked in South Africa). The survey covered a variety of sociodemographic and ethnocultural characteristics and collected detailed information on husbands’ work and migration (for women married to migrants). A separate module of the instrument was devoted to respondents’ decision-making autonomy.

In the middle of 2009 the same women were revisited for a second wave of interviews. The Wave 2 survey had a similar format and included similar modules. After the main fieldwork two more attempts were made to locate and interview respondents who could not be found or were not available at the first attempt. Overall, the sample retention rate was around 84%, which is comparable or superior to that of longitudinal surveys in similar settings (cf. Thornton 2008); almost all women who could not be interviewed had either died or moved out of the area. For all women not interviewed in the three data collection attempts of the second wave, interviews were carried out with a household member, neighbor, or other person who knew the women well; these “proxy interviews” (n=240) were aimed in part at reconstructing the women’s marital, reproductive, and mobility history. Combining information from interviews with available original respondents and from proxy interviews, we were able to ascertain marital status in the middle of 2009 for almost all Wave 1 respondents. Our analytical sample, for which both the Wave 2 marital status and information on all covariates are available, consists of 1568 women, or 96% of the 1638 respondents of Wave 1 who were still alive in mid-2009.

The outcome variable is whether the union reported by a respondent in Wave 1 was dissolved by Wave 2. Due to the earlier mentioned fluid nature of marriage in the study setting, and correspondingly, the fine line that separates “formal” and “informal” marital unions, we cannot distinguish between “divorce” and “separation” (and in fact, would argue that such a distinction has little heuristic or practical value). Even in more developed contexts, studies often combine divorce and separation analytically (e.g., Hirschman and Teerawichitchainan 2003; Martin and Bumpass 1989; Schoen 1992). Accordingly, we use the terms “marital dissolution,” “divorce,” and “separation” interchangeably. Because divorce and separation, like marriage, are often an extended process rather than a discrete event, it is difficult to pinpoint the exact timing of marital dissolution. Informants in the proxy interviews were therefore not asked for the month of marital dissolution.

The main predictor of interest is husband’s migration status as reported in Wave 1. First, we contrast women whose husbands were migrants with those whose husbands were not. The vast majority of migrant husbands in the sample were in South Africa, but some men were working in Mozambique’s capital or another city in Mozambique. Because in this setting the nature of migration and its social and economic consequences for sending households are similar for internal and international migrants, we do not distinguish between these two categories. However, within the migrant-husband subsample we distinguish between women married to more and less successful migrants. Following an earlier approach (e.g., Agadjanian, Arnaldo, and Cau 2011; Agadjanian, Yabiku, and Cau 2011; Yabiku, Agadjanian and Cau 2012), economic success is defined based on a respondent’s assessment of whether the living conditions in her household had improved, worsened, or remained the same as a result of her husband’s migration; “successful” migrants are those for whom economic conditions improved.

The second predictor of interest is women’s decision-making autonomy. We operationalize autonomy using a scale constructed on the basis of respondents’ answers to eight questions in the autonomy module of the Wave 1 questionnaire. The questions were formulated so as to find out whether respondents needed permission from their husbands or in-laws (with whom they typically reside in their husbands’ absence) to visit relatives or friends, travel to the district capital, spend money on family and own needs, work outside the home, use family planning, or take an HIV test. The scale has high internal consistency, with Cronbach’s alpha of .78. Although this variable does not directly measure how decisions about maintaining or dissolving a marital union are made, information about women’s ability to make a variety of important personal and household decisions independently from their husbands and in-laws can serve as a proxy for their ability to influence the outcome of their marital unions. We tested both linear and non-linear (categorical, quadratic) specifications of the relationship between women’s autonomy and marital dissolution. Model fit was not significantly better in the non-linear vs. linear specifications. We therefore opted to treat autonomy as continuous variable and test for its linear association with marital dissolution.

Models also include individual-, couple-, household-, and village-level characteristics that have been shown to be associated with divorce in previous research (Fan and Lui 2004; Heaton 2002; Lee 2006; Lehrer 2004; Martin 2006; Pison 1986; Reniers 2003). These characteristics are measured at Wave 1. The models control for respondent’s age, duration of current marriage (both linear and quadratic), number of children with the Wave 1 husband, education, non-family work, and religious affiliation. Four measures of marital relationship and stability are included. The first of them is a dichotomy of whether or not the respondent had been previously in a marital union. The models also control for whether the respondent’s marriage was monogamous or polygynous at Wave 1. The bridewealth status of the union is operationalized as a dichotomy—at least some bridewealth paid vs. no bridewealth paid. In parts of sub-Saharan Africa greatly affected by the HIV/AIDS epidemic, such as our study setting, concerns about HIV transmission have been a strong motivation for divorce (Porter et al. 2004; Reniers 2008); we therefore control for respondents’ worries about getting infected with HIV by their husbands (respondents’ HIV status is not available).

Two measures of household economic status are also used as controls. One is a material possession scale based on household ownership of such consumer items as radio, bicycle, motorcycle, and automobile; the other is cattle ownership (a dichotomy—household owns any vs. none), which is a largely symbolic measure of household wealth and status in that rural society. In addition, models include a measure of whether in-laws are living in the household. Finally, the prevalence of migration in the community is approximated by the ratio of the number of households with at least one migrant man to the number of households without migrant men in the village households as recorded during the pre-survey canvassing.

Means and standard deviations for the dependent variable and all predictor variables are shown in Table 1. Multivariate analyses use binary logistic regression to model the dichotomous outcome (marriage dissolved vs. not dissolved). All predictor variables were measured at Wave 1; the outcome variable, whether the marriage ended, was measured at Wave 2 from interviews with respondents or from proxy interviews (the actual dissolution of marriage could have taken place at any point over the three-year period between Wave 1 and Wave 2). Because the survey respondents were clustered in villages and therefore could share some unobserved characteristics related to divorce, we employ a random-intercept approach allowing the intercept to vary randomly across villages. The models are fitted using the GLIMMIX procedure in SAS, version 9.

Table 1.

Distribution of all variables

Mean/proportion (SD)
Experienced marital dissolution by Wave 2 0.13 (0.34)
Migration status
 Husband is a migrant 0.42 (0.49)
 Husband is a more successful migrant 0.21 (0.41)
 Husband is a less successful migrant 0.21 (0.41)
Decision-making autonomy scale (1–7, mean) 4.60 (1.90)
Age
 Age 20 or under 0.06 (0.23)
 Age 21–25 0.29 (0.45)
 Age 26–30 0.28 (0.45)
 Age 31 or over 0.27 (0.45)
Duration of current marriage (years, mean) 7.61 (5.28)
Number of children (mean) 2.25 (1.66)
Education
 No formal education 0.26 (0.44)
 Education, 1–4 years 0.45 (0.50)
 Education, 5 or more years 0.28 (0.45)
Working for income 0.21 (0.41)
Marriage characteristics
 At least some bridewealth paid 0.40 (0.49)
 Not first marriage 0.14 (0.35)
 Polygynous marriage 0.21 (0.40)
Belongs to organized religion 0.14 (0.34)
Worried about getting HIV from husband 0.84 (0.37)
Household material possessions scale (1–4, mean) 2.09 (0.96)
Household owns cattle 0.31 (0.46)
Co-resident in-laws 0.39 (0.49)
HH with migrants/HH without migrants in village 1.29 (0.57)
Open in a new tab

Note: All variables measured in Wave 1 except marital dissolution.

To test our first hypothesis, we compare women who were married to migrants with women married to non-migrants. To test Hypothesis 2, we subdivide the migrant-husband subsample on the basis of migration success. To test Hypotheses 3 and 4, we add interactions of the autonomy score with the husband’s migration status variables.

RESULTS

Descriptive results

Table 2 presents the percentage of Wave 1 respondents who were divorced or separated by Wave 2 among all respondents for whom this information is available for the entire sample and by husband’s migration category. The rate of marital dissolution in the sample was quite high: overall, 13.0% of the married women interviewed at Wave 1 were divorced or separated three years later. The percentage of those divorced/separated was only slightly higher among women married to migrants than women married to non-migrants, 13.3% vs. 12.7%. However, this similarity conceals the contrasting experiences of women married to more successful migrants and those married to less successful migrants: among the latter, the share of those whose unions ended in dissolution was more than double that among the former (18.1% vs. 8.7%). Table 2 also shows the percentage of those divorced/separated by their autonomy score (from 1 to 7). The percent divorced/separated is highest at the lowest end of the autonomy scale, declines as autonomy rises, and then increases again slightly at the highest end of the autonomy scale distribution.

Table 2.

Marital dissolution rates by husband’s migration status and wife’s autonomy score

Characteristics in Wave 1 Percent divorced/separated by Wave 2
All 13.0
Husband’s migration status
 Husband is not a migrant 12.7
 Husband is any migrant 13.3
 Husband is a more successful migrant 8.7
 Husband is a less successful migrant 18.1
Wife’s autonomy score
 1 19.6
 2 15.3
 3 13.7
 4 11.2
 5 9.9
 6 12.4
 7 13.4
Open in a new tab

Table 3 displays average autonomy scores for respondents whose union ended in divorce/separation and for respondents whose union remained intact at Wave 2 by their husbands’ migration status at Wave 1. Overall, women who would eventually divorce or separate scored slightly lower on the autonomy score than women who would remain married. However, the gap between the two categories appeared large only among women married to non-migrants: non-migrants’ wives who found themselves divorced/separated by Wave 2 had had a lower autonomy score at Wave 1 than women married to migrants. While the average autonomy scores were identical between women in intact and dissolved unions in the less-successful migrant category, divorced/separated women who had been married to more successful migrants had a slightly higher mean autonomy score than their counterparts in intact unions (5.1% vs. 4.6%). However, these differences are not statistically significant.

Table 3.

Average autonomy score (1–7) by husband’s migration status in Wave 1 and divorce/separation status in Wave 2

Divorced or separated Not divorced or separated
All 4.4 4.6
Husband’s migration status
 Husband is not a migrant 3.9 4.5
 Husband is any migrant 5.0 4.9
 Husband is a more successful migrant 5.1 4.6
 Husband is a less successful migrant 5.0 5.0
Open in a new tab

Multivariate results

Table 4 displays the results of four random intercept logistic regression models predicting divorce or separation by Wave 2 from individual and household characteristics reported at Wave 1. Models A and B include only main effects of migration and autonomy, while Models C and D include interactions between the two predictors. In each set of models, the first models (A and C) use a dichotomous measure of migration, while the second models (B and D) distinguish between more and less successful migrants.

Table 4.

Random-intercept logistic regression of marital dissolution by Wave 2 (parameter estimates and standard errors)

Predictors (all measured in Wave 1) A B C D
β SE β SE β SE β SE
Husband is a migrant 0.020 0.168 −1.204 0.448**
Husband is a more successful migrant −0.333 0.236 −0.300 0.072 −1.686 0.690**
Husband is a less successful migrant 0.248 0.191 −0.864 0.526+
[Husband is not a migrant]
[Age 18–20] 0.343 0.229 0.361 0.230 0.362 0.231 0.378 0.231
Age 21–25 0.343 0.229 0.361 0.230 0.362 0.231 0.378 0.231
Age 26–30 0.079 0.302 0.082 0.303 0.081 0.303 0.081 0.304
Age 31 or over 0.105 0.381 0.125 0.382 0.115 0.382 0.132 0.383
Duration of current union −0.226 0.055** −0.220 0.056** −0.231 0.056** −0.226 0.056**
Duration of current union, squared 0.009 0.003** 0.008 0.003** 0.009 0.003** 0.009 0.003**
Number of children −0.291 0.071** −0.291 0.072** ** −0.299 0.072**
[No education] Education, 1–4 years −0.208 0.197 −0.184 0.197 −0.231 0.198 −0.207 0.198
Education, 5 or more years −0.485 0.234* −0.455 0.235* −0.465 0.234* −0.436 0.235+
Working for income [Not working for income] −0.021 0.200 −0.039 0.200 −0.019 0.200 −0.035 0.201
At least some bridewealth paid [No bridewealth paid] −0.383 0.191* −0.341 0.193 + −0.387 0.192* −0.346 0.194+
[First marital union] Not first marital union 0.020 0.250 0.004 0.251 0.044 0.252 + 0.029 0.252+
[Monogamous union]
Polygynous union 0.695 0.197** 0.667 0.198** 0.701 0.198** 0.674 0.199**
Belongs to organized religion −0.207 0.247 −0.201 0.248 −0.173 0.248 −0.168 0.249
[Does not belong to organized religion]
Worried about getting HIV from husband 0.566 0.238* 0.567 0.238* 0.587 0.239* 0.586 0.239*
[Not worried about getting HIV from husband]
Household material possessions scale −0.059 0.091 −0.047 0.091 −0.061 0.092 −0.049 0.092
Household owns cattle −0.359 0.195+ −0.350 0.196+ −0.387 0.197* −0.378 0.198 +
[Household does not own cattle]
Has co-resident in-laws −0.137 0.176 −0.142 0.178 −0.094 0.177 −0.101 0.178
[Does not have co-resident in-laws]
HH with migrants/HH without migrants in village −0.216 0.162 −0.208 0.163 −0.203 0.163 −0.195 0.165
Decision-making autonomy scale −0.029 0.042 −0.031 0.043 −0.125 0.053* −0.125 0.053*
Interactions
Migrant husband*autonomy scale 0.266 0.089**
More successful migrant husband*autonomy 0.294 0.133*
Less successful migrant husband*autonomy 0.241 0.102*
Intercept −0.231 0.427 −0.298 0.429 0.136 0.443 0.061 0.445
Level-2 variance 0.067 0.082 0.073 0.084 0.071 0.084 0.076 0.085
Model Chi-Square 1554 1571 1561 1580
Number of cases 1568 1568 1568 1568
Open in a new tab

Notes: References categories in brackets; Significance levels:

+

p≤.10

*

p≤.05

**

p≤.01.

In Model A, the likelihood of experiencing marital dissolution does not vary significantly between migrants’ and non-migrants’ wives (and the magnitude of the coefficient is small). Hypothesis 1 is therefore not supported. The autonomy score coefficient is negative but is not statistically significant either. To test Hypothesis 2, we break down the migrant husband category by migration economic success (Model B). Paralleling the bivariate associations (Table 3), women married to more successful migrants are less likely than women married to non-migrants to experience marital dissolution, whereas women married to less successful migrants are more likely to experience dissolution. Although neither category of migrants’ wives are significantly different from non-migrants’ wives, women in the two migration success categories are significantly different from each other (the results of the model contrasting the two categories of migrants are not shown but are available upon request). Thus, Hypothesis 2 is supported with respect to the diverging effects of more successful vs. less successful migration.

In Model C, we add interaction between husband’s migration status (migrant vs. non-migrant) and wife’s autonomy score. Compared to Model A, the picture changes considerably: the main-effect coefficients for both migration status and the autonomy score are now statistically significant. The interaction term has a positive sign and is also significant. The results suggest that being married to a migrant is negatively associated with the probability of marital dissolution at the lowest level of autonomy, but as autonomy among migrants’ wives rises, the probability of their unions ending in divorce/separation also rises. Looked at from the autonomy perspective, the results suggest that higher autonomy is associated with a lower likelihood of marital dissolution among women married to non-migrants but this relationship is reversed among migrants’ wives, consistent with Hypothesis 3. However, Model C also yields the unexpected result that husband’s migration is negatively associated with marital dissolution at the lower levels of autonomy.

The last model, Model D in Table 4, tests Hypothesis 4 by adding interactions of autonomy with perceived migration success. The results of this model are also graphically displayed in Figure 1. The main effect terms for migration in this model suggest that the negative association between migration and marital dissolution for women with low autonomy found in Model C is concentrated among women married to more successful migrants. Although the coefficient for less successful migrants is negative, it is not statistically different from zero. Thus, among women with low autonomy, it is primarily the women reaping economic benefits from their husbands’ absence who experience lower rates of marital dissolution. Both terms for interaction between migration success and autonomy have statistically significant positive coefficients; whereas the magnitude of the coefficient of the interaction term for more successful migrants is somewhat larger, the difference between the two interaction term coefficients is not statistically significant (not shown). These interactions show that higher autonomy is associated with a higher probability of marital dissolution among both wives of more successful migrants and wives of less successful migrants. Overall, contrary to Hypothesis 4, the results do not suggest substantial difference in the association between autonomy and union dissolution across the two categories of migrants’ wives.

Figure 1. Predicted probability of marital dissolution by Wave 2, by husband’s migration status and wife’s decision-making autonomy in Wave 1.

Figure 1.

Open in a new tab

Note: Predicted probabilities based on Table 4, Model D. Husband’s migration status and women’s autonomy score as shown. Control variables are set at the mean or modal values.

Results for other covariates are also noteworthy. The likelihood of divorce first rises and then declines as the duration of union increases. It is negatively associated with the number of children. Being in a polygynous marriage significantly increases the chances of divorce/separation, while being in a union cemented by at least partial transfer of bridewealth is associated with a lower likelihood of marital dissolution (the latter association is marginally significant). Having been in a marital union before tends to increase the likelihood of divorce (the coefficient is marginally significant). Education shows a net negative relationship with the likelihood of divorce. Being worried about contracting HIV from the husband is positively associated with the likelihood of marital dissolution. Cattle ownership is negatively (even if marginally) related to divorce, but household material status is not. The effects of the other controls do not reach statistical significance.

DISCUSSION AND CONCLUSION

The foregoing analysis produced instructive insights into the relationship between men’s migration and marital stability in a rural sub-Saharan setting of rapid social change with traditionally high levels of labor migration but also an increasing diversity of migration outcomes. It illustrated how male labor migration, once a normative process and an integral part of family life, may be associated with the erosion of the marital bond if it ceases to produce expected economic returns for the household. Contrary to the findings of studies in the North American and European settings (e.g., Boyle et al. 2008; Frank and Wildsmith 2004; Landale and Ogena, 1995; Muszynska and Kulu 2008), the analysis did not detect any net impact of husband’s migration on the likelihood of marital dissolution. However, the results showed that the apparent null relationship between migration and marital stability concealed divergent associations for more and less successful migration, and the analysis highlighted important connections between migration, economic outcomes, and women’s autonomy. These results illustrate what an earlier study of migration in this setting defined as the normatively disruptive nature of male labor out-migration (Agadjanian, Yabiku, and Cau 2011). While men’s migration separates marital partners and may cause some psychological and social tensions between them, this separation is built into the local marriage system—and is seen as a normal marital arrangement as long as it yields expected material gains.

Our analysis also detected instructive moderating effects. Specifically, while wife’s decision-making autonomy was associated with reduced likelihood of dissolution in unions of non-migrant husbands, greater autonomy among women married to migrants was associated with increased probability of union dissolution. Although we had predicted that the association between union dissolution and autonomy would be different for women married to more and less successful migrants, the magnitude of the association was similar in the two groups. The association between autonomy and union dissolution thus does not appear to be driven by women’s perception of benefits derived from their husbands’ migration. Instead, the physical presence of the husband may be the primary factor shaping this association between autonomy and divorce. Why is autonomy negatively associated with the likelihood of divorce among women married to non-migrants? In the absence of specialized data on the nature and content of marital relationship and interactions, we cannot offer a definite answer to this question. It may be, we surmise, that women married to non-migrant men are able to deploy their decision-making power to negotiate their relationships with their husbands and to effect changes in their behavior so as to maintain and even strengthen their marital bond. In comparison, negotiating with migrant husbands is greatly hindered by physical separations of the spouses. In addition, migrants’ prolonged immersion in a male-only social milieu that enhances the narrative of masculinity and men’s preeminence (e.g., Campbell 1997) may heighten the communication barriers between them and their left-behind wives.

Our study is not without limitations. Because our measure of marital dissolution does not include the date of divorce/separation, we are unable to use event history analysis to account for censoring of observations. Although a relatively short, three-year span between the two survey waves instills confidence in the use of predictors measured at the first wave, it does not allow for an analysis of longer-term trends. Furthermore, our longitudinal data, while enabling us to separate the predictors and the outcome in time, do not contain any information on the quality of marital relationships prior to the first survey wave. Thus some of the unions that dissolved between Wave 1 and Wave 2 were likely to have been on a trajectory toward dissolution prior to the first wave of the survey. In fact, less successful migration, i.e., perceived lack or insufficiency of material benefits from migration, can already be a sign of a weakening marriage, and vice versa, migration success can reflect a stronger marital bond. However, our inability to pinpoint the onset of changes in marital relationships relative to the dynamics of migration remittances does not challenge our general assumption that migration success or failure can be a step in the process of marital dissolution. In particular, men’s migration is unlikely to be caused by deterioration of marriage quality. Although in some contexts men might initiate migration in order to move away from an unhappy marriage, in this male-dominated setting a more common response to this situation might be to send a wife away or enter into a polygamous relationship. Besides, as we noted earlier, male migration in this setting is a socially normative and culturally routinized practice. Thus reverse causation is unlikely to affect our results.

Another limitation of the study is our inability to ascertain whether marital dissolution was initiated by the wife or the husband and especially how this varied by wife’s autonomy level. It is plausible, for example, that higher-autonomy women would be more likely to initiate divorce/separation than lower-autonomy ones. However, determining at whose initiative a marriage dissolves would be extremely challenging in a setting where the majority of marriages are not legally formalized. In fact, this information is rarely collected in surveys even in western settings, where both marriage and divorce can be more precisely defined and measured and where divorce is initiated by one or both spouses through a formal legal procedure. It is also possible that some of the respondents reclassified their status from having a husband to not having one simply based on their perception of the strength of their “marital” bond. This and other limitations notwithstanding, our study provides a valuable illustration of how labor migration and gender ideologies are jointly linked to the stability of marital unions.

In considering the effects of migration on marital dynamics in sub-Saharan Africa, it is important to see the role of changing migration patterns and returns in the broader context of the dramatic social transformation of Africa’s rural communities and of their marriage systems in particular. As in more developed settings (e.g., Bumpass, Rindfuss, and Tsuya 2009; Ono 2006; Ruggles 1997), these processes in the sub-Saharan context are rooted in economic modernization and the rise of individualism, which, in turn, change the relationship between the nuclear family and kinship networks, shift the relative positions and roles of men and women in society, and transform gender and marital norms, expectations, and relationships (Goode 1993; Reniers 2003; Takyi and Broughton 2006). In that context, trends in marital dissolution are also increasingly conditioned on the acceptability of remarriage. While remarriage of widows, often into polygynous unions, has always been common, remarriage of divorced women, once strongly frowned upon, is also on the rise, especially if the bridewealth requirements in the dissolved marriage were not met. Future research on migration and marriage must take these complex interconnections into account.

Acknowledgments

The support of the Eunice Kennedy Shriver National Institute of Child Health & Human Development (grants # R01HD058365, R21HD048257) is gratefully acknowledged.

Contributor Information

Victor Agadjanian, University of California – Los Angeles.

Sarah R. Hayford, Ohio State University

REFERENCES

  1. Agadjanian V (2008). Research on international migration in sub-Saharan Africa: Foci, approaches, and challenges. The Sociological Quarterly 49(3): 407–421. [Google Scholar]
  2. Agadjanian V, Arnaldo C, & Cau B (2011). Health costs of wealth gains: Labor migration and perceptions of HIV/AIDS risks in Mozambique. Social Forces 89(4): 1097–1118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Agadjanian V, Menjívar C, & Cau B (2013). Economic uncertainties, social strains, and HIV risks: Exploring the effects of male labor migration on rural women in Mozambique In Eckstein SE, & Najam A(Eds.), How migrants impact their homelands (pp. 234–251). Durham, NC: Duke University Press. [Google Scholar]
  4. Agadjanian V, Yabiku ST, & Cau B (2011). Men’s migration and women’s fertility in rural Mozambique. Demography 43(3): 1029–1048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Amato PR (2010). Research on divorce: Continuing trends and new developments. Journal of Marriage and Family 72(3): 650–666. [Google Scholar]
  6. Amato PR, & Rogers SJ (1997). A longitudinal study of marital problems and subsequent divorce. Journal of Marriage and the Family 59(3): 612–624. [Google Scholar]
  7. Arnaldo C (2004). Ethnicity and marriage patterns in Mozambique. African Population Studies 19(1): 143–164. [Google Scholar]
  8. Becker GS (1981). A Treatise on the family. Cambridge, MA: Harvard University Press. [Google Scholar]
  9. Bloom SS, Wypij D, & das Gupta M (2001). Dimensions of women’s autonomy and the influence on maternal health care utilization in a North Indian city. Demography 38(1):67–78. [DOI] [PubMed] [Google Scholar]
  10. Boyle PJ, Kulu H, Cooke T, Gayle V, & Mulder CH (2008). Moving and union dissolution. Demography 45(1): 209–222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bradbury TN, & Karney BR (2004). Understanding and altering the longitudinal course of marriage. Journal of Marriage and Family 66(4): 862–879. [Google Scholar]
  12. Bramlett MD, & Mosher WD (2002). Cohabitation, marriage, divorce, and remarriage in the United States Vital and Health Statistics, Series 23, Number 22. Washington, DC: National Center for Health Statistics. [PubMed] [Google Scholar]
  13. Brines J (1994). Economic dependency, gender, and the division of labor at home. American Journal of Sociology 100(3): 652–688. [Google Scholar]
  14. Brines J, & Joyner K (1999). The ties that bind: Principles of cohesion in cohabitation and marriage. American Sociological Review 64(3): 333–355. [Google Scholar]
  15. Bumpass LL, Rindfuss RR, & Tsuya NL (2009). The institutional context of low fertility: The case of Japan. Asian Population Studies 5(3): 215–236. [Google Scholar]
  16. Campbell C (1997) Migrancy, masculine identity and AIDS: The pshychosocial context of HIV transmission on the South African gold mines. Social Science & Medicine 45(2): 273–281. [DOI] [PubMed] [Google Scholar]
  17. Cherlin A (1992). Marriage, divorce, remarriage. Cambridge, MA: Harvard University Press. [Google Scholar]
  18. Clark S, & Brauner-Otto S (2015). Divorce in sub-Saharan Africa: Are unions becoming less stable? Population and Development Review [Google Scholar]
  19. Crush J, & Bruce F (Eds). (2010). Surviving on the move: Migration, poverty and development in Southern Africa. Cape Town: The Institute for Democracy in South Africa. [Google Scholar]
  20. Crush J, Jeeves A, & Yudelman D (1991). South Africa’s labor empire: A history of black migrancy to the gold mines. Boulder, CO: Westview Press. [Google Scholar]
  21. De Snyder VN (1993). Family life across the border: Mexican wives left behind. Hispanic Journal of Behavioral Sciences 15(3): 391–401. [Google Scholar]
  22. de Vletter F (2007). Migration and development in Mozambique: Poverty, inequality and survival. Development Southern Africa 24(1): 137–153. [Google Scholar]
  23. Fan CS, & Lui H-K. (2004). Extramarital affairs, marital satisfaction, and divorce: Evidence from Hong Kong. Contemporary Economic Policy 22(4): 442–452. [Google Scholar]
  24. Frank R, and Wildsmith E (2005). The grass widows of Mexico: Migration and union dissolution in a binational context. Social Forces 83(3): 919–947. [Google Scholar]
  25. Goode WJ 1993. World changes in divorce patterns. New Haven, CT: Yale University Press. [Google Scholar]
  26. Gottman JM, & Notarius CI (2000). Decade review: Observing marital interaction. Journal of Marriage and Family 62(4): 927–947. [Google Scholar]
  27. Hirschman C, & Teerawichitchainan B (2003). Cultural and socioeconomic influences on divorce during modernization: Southeast Asia, 1940s to 1960s. Population and Development Review 29(2): 215–253. [Google Scholar]
  28. Jejeebhoy SJ (2000). Women’s autonomy in rural India: Its dimensions, and the influence of context” In Presser H & Sen G (Eds) Women’s empowerment and demographic process: Moving beyond Cairo, (pp. 204–238). New York: Oxford University Press. [Google Scholar]
  29. Jesmin S, & Salway S (2000). Marriage among the urban poor of Dhaka: Instability and uncertainty. Journal of International Development 12(5): 689–705. [Google Scholar]
  30. Jones GW (1997). Modernization and divorce: Contrasting trends in Islamic South-East Asia and the West. Population and Development Review 23(1): 95–114. [Google Scholar]
  31. Junod H-A. (1962). The life of a South African tribe. New York: University Books. [Google Scholar]
  32. Heaton Tim B. 2002. Factors contributing to increasing marital stability in the United States. Journal of Family Issues 23(3): 392–409. [Google Scholar]
  33. Heaton TB, Cammack M, & Young L (2001). Why is the divorce rate declining in Indonesia? Journal of Marriage and the Family 63(2): 480–490. [Google Scholar]
  34. Hertrich V (2014) Migration adolescente et autonomie féminine en matière de divorce en Afrique: Réflexions à partir d’un suivi de population au Mali. Cahiers Québécois de Démographie 43(2): 279–313. [Google Scholar]
  35. Kalmijn M, de Graaf PM, & Poortman A-R (2004). Interactions between cultural and economic determinants of divorce in The Netherlands. Journal of Marriage and the Family 66(1): 75–89. [Google Scholar]
  36. Katz E (1997). The intra-household economics of voice and exit. Feminist Economics 3(3): 25–46. [Google Scholar]
  37. Landale NS, & Ogena N (1995). Migration and union dissolution among Puerto Rican women. International Migration Review 29(3): 671–692. [Google Scholar]
  38. Lee Y-J. (2006). Risk factors in the rapidly rising incidence of divorce in Korea. Asian Population Studies 2(2): 113–132. [Google Scholar]
  39. Lehrer EL (2004). Religion as a determinant of economic and demographic behavior in the United States. Population and Development Review 30(4): 707–726. [Google Scholar]
  40. Locoh T, & Mouvagha-Sow M (2008). An uncertain future for African families In Jayakody R, Thornton A, & Axinn W (Eds), International family change (pp.45–80). New York, NY: Lawrence Erlbaum Associates. [Google Scholar]
  41. Loforte AM 2000. Género e poder entre os Tsonga de Moçambique. Maputo, Mozambique: Promédia. [Google Scholar]
  42. Luke N (2010). Migrants’ competing commitments: Sexual partners in urban Africa and remittances to the rural origin. American Journal of Sociology 115(5): 1435–1479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Lundberg S, & Pollak RA (1996). Bargaining and distribution in marriage. Journal of Economic Perspectives 10(4): 139–158. [Google Scholar]
  44. Lyngstad TH, & Jalovaara M (2010). A review of the antecedents of union dissolution. Demographic Research 23: 257–292. [Google Scholar]
  45. Martin TC, & Bumpass LL (1989). Recent trends in marital disruption. Demography 26(1): 37–51. [PubMed] [Google Scholar]
  46. Martin SP (2006). Trends in marital dissolution by women’s education in the United States. Demographic Research 15, Article 20: 537–560. [Google Scholar]
  47. Meekers D (1992). The process of marriage in African societies. Population and Development Review 18(1): 61–78. [Google Scholar]
  48. Menjívar C, & Agadjanian V (2007). “Men’s migration and women’s lives: Views from rural Armenia and Guatemala. Social Science Quarterly 88(5): 1243–1262. [Google Scholar]
  49. Muszynska M, & Kulu H (2007). Migration and union dissolution in a changing socio-economic context: The case of Russia. Demographic Research 17, Article 27: 803–820. [Google Scholar]
  50. Ono H (1998). Husbands’ and wives’ resources and marital dissolution. Journal of Marriage and the Family 60(3): 674–689. [Google Scholar]
  51. Ono H (2006). Divorce in Japan: why it happens, why it doesn’t” In Blomström M, & La Croix S (Eds.) Institutional change in Japan (pp. 221–236). London: Routledge. [Google Scholar]
  52. Oppenheimer VK (1997). Women’s employment and the gain to marriage: The specialization and trading model. Annual Review of Sociology 23: 431–453. [DOI] [PubMed] [Google Scholar]
  53. Porter L, Hao L, Bishai D, Serwadda D, Wawer MJ, Lutalo T, et al. (2004). HIV Status and union dissolution in Sub-Saharan Africa: The case of Rakai, Uganda. Demography 41 (3): 465–482. [DOI] [PubMed] [Google Scholar]
  54. Reniers G (2003). Divorce and remarriage in rural Malawi. Demographic Research, Special Collection 1, Article 6: 175–206. [Google Scholar]
  55. Reniers G (2008). Marital strategies for regulating exposure to HIV. Demography 45(2):417–438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Ruggles S (1997). The rise of divorce and separation in the United States, 1880–1990. Demography 34 (4): 455–466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Sayer LC, & Bianchi S (2000). Women’s economic independence and the probability of divorce: A review and reexamination. Journal of Family Issues 21(7): 906–943. [Google Scholar]
  58. Schoen R (1992). First unions and the stability of first marriages. Journal of Marriage and the Family 54(2): 281–284. [Google Scholar]
  59. South SJ (2001). Time-dependent effects of wives’ employment on marital dissolution.” American Sociological Review 66(2): 226–245. [Google Scholar]
  60. South SJ, & Lloyd KM (1995). Spousal alternatives and marital dissolution. American Sociological Review 60 (1): 21–35. [Google Scholar]
  61. Sweeney MM (2002). Two decades of family change: The shifting economic foundations of Marriage. American Sociological Review 67(1): 132–147. [Google Scholar]
  62. Takyi BK, & Broughton CL (2006). Marital stability in Sub-Saharan Africa: Do women’s autonomy and socioeconomic situation matter? Journal of Family and Economic Issues 27(1): 113–132. [Google Scholar]
  63. Takyi BK, & Gyimah SO (2007). Matrilineal family ties and marital dissolution in Ghana.” Journal of Family Issues 28(5): 682–705. [Google Scholar]
  64. Thornton RL (2008). The demand for, and impact of, learning HIV status. American Economic Review 98(5): 1829–1863. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Weisner TS, Bradley C, & Kilbride PL (Eds). (1997). African families and the crisis of social change. Westport, CT: Bergin & Garvey. [Google Scholar]
  66. White LK, & Booth A (1991). Divorce over the life course the role of marital happiness. Journal of Family Issues 12(1): 5–21. [Google Scholar]
  67. Yabiku ST, Agadjanian V, & Cau B (2012). Labor migration and child mortality in Mozambique. Social Science & Medicine 75(12): 2530–2539. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Yabiku ST, Agadjanian V, & Sevoyan A (2010). “Husbands’ Labor Migration and Wives’ Autonomy.” Population Studies 64 (3): 293–306. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Yi Z, & Deqing W (2000). Regional analysis of divorce in China since 1980. Demography 37 (2): 215–219. [PubMed] [Google Scholar]
  70. Yi Z, Schultz TP, Wang DD, & Gu D (2002). Association of divorce with socio-demographic covariates in China, 1955–1985: Event history analysis based on data collected in Shanghai, Hebei, and Shaanxi. Demographic Research 7, Article 11: 407–432. [Google Scholar]

RESOURCES