Negotiated Identities: Male Migration and Left-Behind Wives in India

Abstract

This paper examines the impact of husbands’ migration on the lives of women left behind. Using data from the India Human Development Survey 2005, we focus on two dimensions of women’s lives: women’s autonomy and control over their lives; and women’s labour force participation. Results suggest that household structure forms the key mediating factor through which husbands’ absence affects women. Women not residing in extended families are faced with both higher levels of responsibilities and greater autonomy, while women who live in extended households do not experience these demands or benefits.

Research on migration has increasingly focused on women migrants (Bilsborrow 1992) and has also recognized the gendered nature of the migration process (Whiteford 1978; Hondagneu-Sotelo 1992; Chin 1997; Yeoh et al. 1999; Lutz 2002; Curran and Rivero-Fuentes 2003; Morokvasic 2004; Shah 2004). However, surprisingly little attention is directed to women who are not migrants themselves but are deeply affected by the migration process: women whose husbands have migrated in search of work leaving them behind (Hugo 2000); a gap this paper seeks to fill.

Unlike other demographic phenomena such as birth and death, migration is a process rather than an event. Diversity seems to be the norm when it comes to characterizing migration with respect to reasons for departure, length of migration, frequency of return to place of origin and ties to home communities (Goldscheider 1987; Massey et al. 1990; Lucas 1997).

Nevertheless, a common pattern characterizing migration of men in developing countries is that they leave their wives and children in the place of origin while they migrate in search of work. For instance, a ‘well-documented strategy’ among Mexican rural households is for the men to migrate in search of employment while their families remain at home (Kanaiaupuni 2000). US immigration policies have supported and strengthened this form of migration through provisions that make it easier for women to legally migrate as spouses of male migrants, but until 1952 women could not legally sponsor their husbands as migrants. The rotation system or guest worker immigration policy that has been adopted in postwar Western Europe also encourages this pattern of solo male migration. This policy is tailored to meet a short-term demand for labour, with single males rather than entire families being given temporary work permits which are renewed annually. When economic conditions change and the services of the immigrant work force are no longer required, the work visas are not renewed, in the hope that these immigrants will return to their native countries (Laczko 2002).

State policies may also encourage solo male migration within the country. An example of this was apartheid in South Africa which resulted in the influx of male mine workers who were prohibited from bringing their families with them (Brown 1983). At other times, irregularity of work in urban areas, urban housing shortages or the need for farm labour separates the families (de Haan 2006). While sole female migration is increasing (Zlotnik 1995; Roberts 2002; Shah 2004), particularly for specific occupations such as nurses or domestic workers, it appears that human capital, family responsibilities and gender roles will continue to ensure that labour migration will remain dominated by male migration in years to come (Donato 1993; De Jong 2000).

Solo male migration and women

A review of the available literature on the impact of male migration on families documents two types of effects, the first being on women’s autonomy since husbands’ absence forces or frees them to take on roles that they would not normally undertake (Hugo 2000). A study of 44 migrant Mexican men and women in the San Francisco Bay Area (Hondagneu-Sotelo 1992) suggests that the migration of men typically entails expansion of their wives’ responsibilities and acquisition of skills in tasks not traditionally undertaken by women; for instance, to cope with the low level of remittances women often take up employment in the informal sector and learn to administer household budgets. Thus, men’s absence from home provides conditions for fostering women’s autonomy, self-esteem and role expansion. For men, separation from their wives necessitates undertaking domestic tasks that they would not have otherwise done. However, the dismantling in gender-segregated roles is only partial since frequently families return to a patriarchal division of labour once they reunite. Another study of male migration from rural to urban areas for skilled manual and white collar employment in the coastal state of Goa, India, suggests that in the absence of husbands women are de facto household heads and execute various responsibilities such as hiring and supervision of agricultural labour (Mascarenhas- Keyes 1990).

The other line of research emphasizes financial hardships and increased responsibilities for women. Further, the assumption of responsibilities outside the home may in some instances increase the work burden of women. For instance, in rice-producing villages of eastern UP, if remittances are not large enough, women’s work load is likely to increase as they have to compensate for the absence of their husbands’ farm labour (Paris et al. 2005). Other studies also note that frequently remittances from the migrants are not enough and women who get left behind in the native villages have to assume the role of sole breadwinner in addition to added familial and domestic responsibilities (Jetley 1987). Domestic responsibilities may be shared by the older daughter who acts as a little surrogate mother to her brothers and sisters.

While these arguments appear logical, reality is often far more complex. As Gupta and Ferguson note in a slightly different context, ‘Representations of space in the social sciences are remarkably dependent on images of break, rupture, and disjunction.’ (1992: 6). However, this rupture may be less severe than one imagines. We have already noted earlier that migration is a messy process; migrant husbands often leave behind a very large presence, reinforced by periodic visits to their homes. More importantly, the vision of a static rural community may also be unrealistic. Communities and households may well reconfigure themselves to respond to male migration to preserve gender patterns. An interesting study of male migrants to the Gulf countries from Cairo (Hoodfar 1996) finds that male migration has tended to strengthen rather than weaken the traditional gender ideologies whereby women are seen as being dependent on their menfolk and many households have reorganized themselves so that brothers or other male relatives moved in to take care of the women left behind. Studies in India have also found that migrants often delayed their migration until some male relatives were available to care for the families being left behind (de Haan 2006). Thus, how households and communities respond to migration is likely to be the key to shaping the effect of male migration on women’s autonomy and empowerment which forms the focus of the present paper.

Migration in Indian society

Migration in India is a highly localized phenomenon. A report by the National Sample Survey Organisation provides an interesting description of migration in India: almost 99 per cent of the migration takes place within India, although given the overall population size, India also contributes substantially to international migration. While 27 per cent of the Indian population is identified as ‘migrant’, consisting of individuals who no longer live in the town or village where they were born, an overwhelming majority, 77 per cent, are women who migrated in the context of an arranged marriage. Over 40 per cent of the female population in rural as well as urban areas consists of migrants. However, male migration is not insubstantial. Nearly seven per cent of the rural population and 27 per cent of the urban male population consist of migrants. Of these, 30 per cent of the rural male migrants and 53 per cent of the urban male migrants migrated for employment-related reasons; an increasing number also travel to urban areas to study.¹ Most of this migration is relatively short-distance. Among rural male migrants, 57 per cent come from same district and 82 per cent from the same state; among urban male migrants, 34 per cent from the same district and 74 per cent from the same state (National Sample Survey Organisation 2001).

However, it appears that short-distance migrants often take their families with them. In contrast, individuals migrating to another state or abroad may well be more likely to leave their families behind. A recently-conducted household survey in India, the India Human Development Survey 2005, documents that among ever-married women aged 15–49, about 4.5 per cent of the rural women and about 1.5 per cent of the urban women had husbands who lived elsewhere. Among migrant husbands, 26 per cent are living in the same state, 62 per cent in a different state and 12 per cent in a different country. However, this phenomenon is highly geographically clustered. In the mountainous state of Uttarkhand, nearly nine per cent of the ever-married women have husbands living elsewhere, as do eight per cent of the rural women in the central plains of Uttar Pradesh and 11 per cent in Bihar. In contrast, in the more prosperous southern states of Karnataka and Tamil Nadu, few women reside away from their husbands. Socio-economic characteristics of the families with male migrants as well as geographic distribution are presented in Tables 1 and 2.

Table 1.

Distribution of marital status by socio-economic characteristics

Characteristics	Co-resident husband	Migrant husband	Widowed/separated
All India	91.65	3.65	4.7
Age of the woman
20	92.56	5.99	1.46
30	93.36	4.4	2.23
40	91.26	3.32	5.42
49	88.81	2.06	9.13
Woman’s education
Illiterate	90.39	3.97	5.64
1–5 grade education	91.71	3.21	5.08
6–9 grade education	93.2	3.51	3.29
10 grade-some college	93.62	3	3.38
College graduate	93.8	3.41	2.79
Place of residence
Rural	90.85	4.5	4.65
Urban	93.7	1.48	4.82
Social group
Forward castes	92.45	3.46	4.09
Other backward classes	91.27	3.99	4.75
Dalit	91.16	3.45	5.4
Adivasi	94.56	0.9	4.55
Muslim	90.75	4.98	4.27
Christian, Sikh, Jain	90.68	4.13	5.19
Household occupation
Agricultural labour	91.62	1.91	6.47
Non agricultural  labour	92.71	2.97	4.32
Small farmer	90.9	4.44	4.66
Med/large farmer	94.5	2.01	3.49
Trade/artisan	95.19	1.3	3.51
Salaried/professional	90.33	6.06	3.61
Retired/other	73.48	13.09	13.43
Per capita household consumption expenditure	826.48	892.45	846.55
No of persons in HH	5.62	5.62	4.20
No. of adults in HH	2.89	2.34	2.05
No. of older women in HH	0.34	0.56	0.33
N	33951	992	1539

Source: India Human Development Survey, 2005.

Table 2.

Distribution of marital status by state of residence

State of residence	Co-resident husband	Migrant husband	Widowed/separated
All India	91.65	3.65	4.7
Jammu & Kashmir	96.25	1.09	2.66
Himachal Pradesh	87.73	7.30	4.97
Uttarakhand	78.45	14.53	7.02
Punjab	93.33	2.04	4.62
Haryana	95.46	1.12	3.42
Delhi	95.95	0.00	4.05
Uttar Pradesh	86.59	9.56	3.85
Bihar	85.13	12.23	2.65
Jharkhand	95.34	0.88	3.78
Rajasthan	89.81	6.92	3.27
Chhattisgarh	94.29	0.63	5.09
Madhya Pradesh	96.24	0.40	3.36
Northeast States	94.48	0.09	5.44
Assam	95.88	0.18	3.94
West Bengal	92.88	2.51	4.61
Orissa	94.03	1.55	4.42
Gujarat	96.22	0.51	3.28
Maharashtra, Goa	93.49	0.90	5.61
Andhra Pradesh	92.65	1.03	6.31
Karnataka	92.09	1.28	6.64
Kerala	86.63	8.31	5.07
Tamil Nadu	92.00	1.17	6.83

Source: India Human Development Survey, 2005.

A few characteristics of women with migrant husbands, compared to their counterparts who have co-resident husbands or those who are currently not in a union, are noteworthy. Younger wives are more likely to be living away from their husbands than older wives. This is reflective of labour migration being highly selective of younger men. It is also plausible that older wives have over time figured out a living situation in which they reside with their husbands. Education does not seem to play a role in women’s living away from their husbands and caste or ethnicity or religion plays only a minor role, with non-Hindu women being substantially more likely to have migrant husbands. There appears to be a preference for Muslim employees among the Middle East recruiters, which may partly explain this phenomenon. Families whose primary source of income is salaried or professional work seem to be more likely to have migrant males, as do families subsisting on retirement income or remittances. The lack of formal-sector employment in rural areas may drive informal- sector workers to leave their families behind as they migrate to urban areas in search of work. Interestingly, women with migrant husbands live in slightly betteroff households than women who live with their husbands. It is difficult to identify the direction of this relationship. It has been noted that privileged individuals are more likely than poorer ones to migrate (Massey et al. 1998), but at the same time, having a migrant family member increases income from remittances, increasing the overall standard of living.

The effect of migration on household size is not clear: we expect migrant households to be smaller than other households since one of their members is no longer present. At the same time, migrant households could be larger given that migration is selective of younger men who are most likely to be in households with children. Data from the India Human Development Survey 2005 indicate that household size does not seem to differ by husband’s presence in the household; even the number of adults across migrant and non-migrant households differs by less than 0.5. It seems that other family members, particularly older relatives, fill the gap. This suggests that many women live with other family members in the absence of their husbands; in contrast widows and divorced or separated women live in smaller households with fewer adults.

Solo male migration and left-behind women’s lives in India

It would be reasonable to expect that the absence of husbands has a deep impact on women’s lives. In an Indian context, two areas seem to be particularly affected. First, various studies of women’s empowerment in India have noted limited autonomy and decision-making ability on the part of women (Desai 1994; Mason and Smith 2000; Jejeebhoy and Sathar 2001; Bloom et al. 2001). We expect that in their husbands’ absence women may have a greater role in family decision-making and may be able to put aside norms of female seclusion since the husbands’ absence would increase the need for their participation and leadership in the day-to-day affairs of the family. Moreover, migration may introduce new ideas and attitudes in men which may ultimately change gender roles in the family. Second, while male migration may be associated with a higher likelihood of remittances, women may need to fill in for absent husbands in many ways including care of animals, and work on the family farm or in the family business (Jetley 1987; Paris et al. 2005). Moreover, in some instances the sporadic nature of remittances may also force women to generate cash income through wage work (Gulati 1993).

Gulati’s (1993) interviews with women in Kerala provide insights into the processes shaping women’s lives in the context of male migration, and document both the constraints and the opportunities provided by male migration. Hameeda, one of Gulati’s informants, reports (p.31):

When Jamal [husband] is visiting home, he takes me out to movies. He never insists that I should cover my head. My mother is very orthodox and would never have permitted me such liberty. Actually, now several women in our neighbourhood have stopped covering their heads and go to the movies in short-sleeved blouses. You need someone to take the initiative and introduce these small changes.

In contrast, Rehana reports (p. 38):

My position is rather awkward. Of the 13 persons living in this house, I am the only outsider. Although I am married to the most important person who is everyone’s hope, I have very little freedom of movement, speech or action. I cannot go out of the house without the permission of my mother-in-law. … In my position, I simply cannot afford to displease anyone.

Similarly, Gulati’s respondents also note a diversity of economic outcomes. Sultana reports (p. 55):

My husband’s migration has descended on me as a curse … After Shams stopped sending me money, I registered with the government employment exchange for a job.

In contrast, Kumari documents an incredible growth in self-confidence and control over finances (p. 107):

In the beginning after Mani departed for Abu Dhabi, I had all kinds of doubts about my ability to cope with things in his absence. My problems started with writing letters … I had to ask someone to write it for me. I was not certain also how I would be able to manage finances. … Keeping money in the bank, I found, was a help. You keep it there safely until you really need it.

These interviews, and others, document both rigid constraints on women and spaces in which they manage to find their voice. They also raise an interesting question. Why do some women find freedom and responsibility in their husband’s absence while others do not?

We suggest that living arrangements form the sieve through which migration experience is filtered. Some women establish or maintain their own households and gain increased autonomy as well as responsibility. Others live with extended family and are subject to strict supervision and regulation and must cope without help from their husbands mediating between them and the extended households.

While extended-family living remains prevalent in India, and in our survey over 90 per cent of women started married life by residing with the husband’s parents, over time families frequently divide, particularly as children are born and parents die. In the India Human Development Survey, nearly half the households are nuclear, the rest are extended. However, it is considered unusual for women to live alone and husbands’ migration may be feasible only if young women are able to live with other family members (de Haan 2006). Moreover, it would not be acceptable for a young woman to live alone with older male relatives of her husband; chaperonage by a female relative would be required. Thus, we argue that the gender effect of male migration on women is moderated through household structure, with greater freedom in households where no older woman is present.

Research questions and data

This paper asks the following questions: is husbands’ migration empowering for women who are left behind, by increasing their autonomy and decision making power? Is husbands’ migration associated with higher work demands on women? Is this effect conditional on the living arrangements of the women left behind?

In addressing these questions we compare three groups of women, women living with their husbands, women whose husbands live elsewhere, and women who do not have a husband: widowed, separated and divorced women.

Most of the literature in this area, including the literature cited above, is based on qualitative research. While these studies help in developing a sophisticated theoretical framework, testing this framework is very difficult. For example, the study of Egyptian urban households cited above argues:

Studies of the impact of male migration on the position of women must take into consideration other important variables such as age, social class, education, rural versus urban context, and the duration of migration. (Hoodfar 1996: 72).

However, by their very nature, qualitative studies focus on small homogeneous samples and find it difficult to address these differences. On the other hand, quantitative studies based on sample surveys need a fairly large sample to find sufficient cases of non-traditional households, such as households with migrant husbands. We are fortunate to have access to the India Human Development Survey, 2005 (Desai et al. forthcoming).

The India Human Development Survey 2005 (IHDS) was organized by researchers from the University of Maryland and the National Council of Applied Economic Research, New Delhi, and funded by two grants from the National Institutes of Health. The IHDS is a nationally representative household survey of 41,554 urban and rural households. It covers all states and union territories of India, with the exception of Andaman-Nicobar and Lakshadweep which contain less than one per cent of the Indian population. These households are spread across 33 states and union territories, 384 districts, 1503 villages and 971 urban blocks. The sample was selected using a clustered sampling procedure designed to provide a nationally representative sample of India. The sampling procedure is described in detail elsewhere (Desai et al. forthcoming).

The survey instruments were administered in 13 languages and used interviewers fluent in the local language and well-versed in the culture. A team of female and male interviewers visited these households and collected data in face-to-face interviews from one respondent knowledgeable about household income and employment, usually the male head of the household, and one ever-married woman aged 15–49 (if any). Women were interviewed by female interviewers.

IHDS is a multi-topic survey with information on income, employment, education and health. The data are a public resource to Indian and international researchers interested in studying human development in India and are freely downloadable from ICPSR. Household information was conducted through structured interviews using two questionnaires. The first was administered to a key informant, usually the male household head, and collected detailed information about the household: demographic characteristics of household members, economic condition of the household (level of income, consumption patterns and asset ownership), education level and morbidity and mortality patterns of household members. The second questionnaire was directed at one ever-married woman aged 15–49 in the household and collected information about various dimensions of gender relations within the household and the community, in addition to information about health and fertility.

This provides us with a large enough sample of ever-married women aged 15 to 49 (a total of 33,482 women, including 992 with absent husbands) to study the effect of migration on women’s lives. Further, this large sample size allows comparison on various dimensions of gender relations between women in non-traditional households, that is, women with migrant husbands, and women in traditional households (women who live with their husbands), as well as women in other non-traditional households such as women who do not have a husband because of death or divorce. Data quality assessment suggests that the socio-economic composition of the sample is comparable to population characteristics documented in other large surveys and the Indian census (Desai et al. forthcoming)

While use of survey data offers many benefits, its structured nature limits the kind of questions that can be addressed. Consequently, empowerment in this paper has a very specific meaning. It draws from a focus on subjective sense of self-efficacy and entitlement (Kabeer 1999) and focuses on women’s role in household decisionmaking and their ability to move freely outside their home. We focus on the following markers of empowerment.

Women’s role in decision making

The survey asked ever-married women respondents in the age group 15–49 years:

Please tell me who in your family decides the following things: what to cook on a daily basis; whether to buy an expensive item such as TV or fridge; how many children you have; what to do if a child falls sick; whom your children should marry.

The respondent was able to offer a response ‘yes’ or ‘no’ for each of the following household members: the respondent herself, husband, senior male, senior female and other. When the respondent identified multiple decision makers, she was asked who the primary decision maker was. In this paper we focus on a decision-making index that counts the number of items on which respondents are primary or sole decision-makers.² This index ranges from 0 to 5 with 0 indicating no autonomy in the above decisions and 5 indicating full autonomy in all five of the decisions asked about.

Women’s mobility and freedom of movement

It has been noted by a variety of scholars of South Asia that women’s physical mobility is severely restricted in most parts. Norms of female seclusion expressed in ghunghat or purdah play a role, but even for women who do not practise purdah, there is an expectation that they will seek permission from their husband or older family members to go to a shop or visit friends (Bloom et al. 2001). IHDS asked eligible women respondents whether they needed permission of their husband or any other senior family member to go to a grocery store, to a health clinic and to visit friends or neighbours. The respondents had the option of answering ‘yes’ or ‘no’. The number of places where they could move freely without needing permission was added up to construct a mobility index,³ ranging from 0 to 3.⁴ Women who do not need permission from any other person in the household to make these visits are most mobile with an index score of 3.

In addition to a focus on gender roles, we also argued that male migration changes the labour balance in the household. In some cases women may need to fill the gap left by their husbands by taking care of farms or businesses, in other cases the remittance income may allow them to withdraw from the labour force, or make it possible to hire labour in the farm or business (Arya and Roy 2006). In order to examine these effects we focus on two types of labour force activities.

Women’s participation in the labour force

This is measured by whether women worked on the family farm or in the family business, looked after animals, or participated in any kind of wage work (0 = no labour force participation, 1 = work for wages, on farm or business, care for animals). Since it argued that women’s work on the family farm or in animal care is often not recorded, the IHDS made special efforts to capture this information using a specially designed employment module.

Women’s participation in wage labour

Participation in wage labour is measured by whether they worked for pay in agricultural or non-agricultural work (0 = no, 1 = casual wage work or regular salaried work).

The two gender roles variables are indices which are analysed using an ordinal logit regression while the labour force variables are analysed using a logit model. In each model, in addition to the migration and marital status, we also control for state of residence, urban residence, age of the woman, number of children, caste-religion, log of per capita household consumption expenditure (a marker of permanent income), woman’s education and household’s primary source of income. Descriptive statistics are presented in Appendix 1. The primary independent variables of interest are: (1) Marital status divided into three categories: co-resident husband, migrant husband, and not being currently in a union, which includes divorced, widowed, and separated women, as well as a handful for whom we could not find out the location of the husband; and (2) living in an extended family where an older woman is present.

Appendix 1.

Descriptive statistics

Variable	Per cent
Distribution of dependent variables
Decision making autonomy index score
0	19.43
1	43.88
2	20.02
3	9.04
4	2.74
5	4.89
Mobility index score
0	49.43
1	17.13
2	11.46
3	21.98
Per cent in any employment	53.67
Per cent in wage employment	24.59
Distribution of independent variables
Residential status
Co-resides with husband	91.65
Migrant husband	3.65
Widowed/divorced/separated	4.70
Per cent living with older women	35.05
Distribution of control variables
Urban residence	28.24
Respondent’s age	32.85
No. of children	2.60
Social group
Forward castes	20.40
Other backward classes	35.75
Dalit	22.34
Adivasi	7.42
Muslim	11.54
Christian, Sikh, Jain	2.54
Woman’s education
Illiterate or missing education	47.90
1–5 grade education	16.64
6–10 grade education	26.36
10 grade-some college	4.92
College graduate	4.17
Log per capita expenditure	9.96
N	33,486

Source: India Human Development Survey, 2005

Results

Results from the ordinal logits and logit regressions for the four dependent variables of interest are presented in Appendix 2. Given the difficulties in interpreting results from non-linear models, we present these results by calculating the difference in outcome variables for different categories of independent variables of interest where all control variables are held at their mean value. These calculations are based on the full models from Appendix 2.

Appendix 2.

Coefficients from ordinal logit models for women’s autonomy and mobility scale and from logit models for women’s employment

	Score on Empowerment Index		Score on Mobility Index		Any work		Wage work
Variablea	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Urban residence	0.0921**	0.0999**	0.108**	0.113**	−1.351***	−1.350***	−0.709***	−0.708***
Age of the woman	0.0198***	0.0190***	0.0161***	0.0156***	0.0304***	0.0303***	0.0273***	0.0272***
No. of children	0.0435***	0.0396***	0.0229	0.0196	0.137***	0.136***	−0.00758	−0.00849
Other backward classes	−0.130***	−0.132***	−0.0493	−0.0527	0.394***	0.394***	0.360***	0.361***
Dalit	0.00456	−0.00358	−0.131**	−0.141**	0.330***	0.328***	0.683***	0.682***
Adivasi	0.0166	0.00698	−0.0826	−0.0963	0.893***	0.891***	1.114***	1.112***
Muslim	−0.128**	−0.131**	−0.190***	−0.192***	−0.212***	−0.213***	−0.269***	−0.268***
Other religion	−0.122	−0.132	−0.13	−0.126	0.306***	0.308***	0.133	0.133
1–5 grade education	0.0105	0.019	−0.0927	−0.0917	−0.438***	−0.438***	−0.459***	−0.458***
6–9 grade education	0.03	0.0315	0.0326	0.031	−0.630***	−0.630***	−0.884***	−0.885***
10 grade-some college	0.0861	0.0746	0.0259	0.0161	−0.576***	−0.576***	−0.224*	−0.226*
College graduate	0.276***	0.276***	0.303***	0.298***	−0.102	−0.104	0.756***	0.756***
Non-agricultural labour HH	0.246***	0.247***	0.174**	0.170**	−0.459***	−0.460***	−0.350***	−0.351***
Small farmer HH	−0.0036	−0.00681	0.0414	0.0344	0.451***	0.450***	−1.175***	−1.177***
Med./large farmer	−0.130*	−0.131*	−0.0966	−0.103	0.302***	0.300***	−1.718***	−1.721***
Trade/artisan HH	0.0634	0.0698	0.087	0.087	−0.767***	−0.768***	−1.441***	−1.441***
Salaried/professional HH	−0.00979	−0.00423	0.193***	0.190**	−1.016***	−1.020***	−1.227***	−1.229***
Retired/other HH	−0.215**	−0.255**	0.06	0.0554	−1.028***	−1.036***	−1.560***	−1.568***
Log per capita consumption expenditure	0.101***	0.0915***	0.106***	0.104***	−0.105***	−0.105***	−0.451***	−0.452***
Any older women in HH	−0.889***	−0.747***	−0.274***	−0.167***	−0.281***	−0.266***	−0.258***	−0.229***
Migrant husband	1.022***	1.954***	1.102***	1.742***	0.307***	0.555***	0.311**	0.533***
Divorced/widowed	3.012***	3.734***	1.757***	2.244***	1.029***	0.997***	1.530***	1.603***
Migrant * Older women		−1.918***		−1.407***		−0.432*		−0.521*
Divorced * Older women		−2.142***		−1.317***		0.0856		−0.217
Constant	0.767***	0.685**	1.535***	1.522***	0.248	0.25	1.559***	1.564***
Cut 2	3.062***	2.979***	2.355***	2.345***
Cut 3	4.347***	4.283***	3.024***	3.019***
Cut 4	5.457***	5.432***
Cut 5	6.058***	6.062***
N	30,972	30,972	27,125	27,125	33,370	33,370	33,370	33,370

^***p < 0.01

^**p < 0.05

^*p < 0.1.

^aEach regression also contains dummy variables for state of residence, results not reported. Source: India Human Development Survey, 2005.

Table 3, containing predicted average scores on the decision-making and mobility index as well as probabilities of employment, tells an interesting and consistent story. Women with migrant husbands are more likely to participate in household decisions, are better able to venture outside the home without seeking permission, and more likely to participate in the labour force than women whose husbands have not migrated. However, this effect is considerably less than that for women who have no husbands. For example, the average predicted score on the decision-making index, which ranges from 0 to 5 decisions on which the respondent is the primary decision maker, is 1.33 for women with co-resident husbands, 1.93 for women with migrant husbands and 3.41 for women who are widowed or separated. These effects are net of other socio-economic factors such as presence of older women in the household, state of residence, urban or rural residence, household consumption expenditure, household’s primary occupation and woman’s education. Similarly, women with migrant husbands are more likely to participate in the labour force than those with co-resident husbands but divorced or widowed women are the most likely to work.

Table 3.

Predicted average score on autonomy and mobility indices and predicted likelihood of being employed, by marital status

	Predicted average score		Predicted likelihood of
Marital status	Autonomy	Mobility	Any work	Wage work
Co-resident husbands	1.33	0.88	0.51	0.13
Migrant husbands	1.93	1.60	0.58	0.18
Divorced/widowed women	3.41	2.03	0.74	0.42

Source: Predicted estimates based on coefficientsfrom Model 1, Appendix 2.

This suggests that while husbands’ absence changes women’s lives in many different ways, a migrant husband is still very much present in the way women’s lives are shaped and makes their experiences different from those of the widowed and divorced. How is this influence exercised? Our review of the literature above suggests that household composition is an important intervening factor. When men migrate, instead of living alone or only with their children, women may be left in the care of other relatives. Frequently migration may only be possible if other household members are available to co-reside with women. When this co-residence is with older family members, women may not have to deal with the difficult yet empowering experience of coping on their own, with little enhancement in autonomy even in their husbands’ absence.

In order to examine this, we add an interaction term to our regression models between the presence of an older woman in the household and marital status. The interaction between the presence of an older woman and having a migrant husband is statistically significant at the 0.05 level for three variables, decision making, mobility, and participation in the labour force. It is significant at the 0.1 level for wage work. Predicted probabilities from this model are presented in Table 4. The results show an interesting difference between women in extended families and those in households where no older woman is present. Much of the positive effect of husbands’ migration on wives is limited to those who live in nuclear families and not those in extended families. For example, on the decision-making scale, the average predicted score in nuclear households is 1.46 for women with co-resident husbands, 2.77 for those with migrant husbands and 4.06 for widows. However, in extended households, these scores are 1.08, 1.10 and 1.97 respectively.

Table 4.

Predicted average score on autonomy and mobility indices and predicted likelihood of being employed, by residential and marital status

	Predicted average score		Predicted likelihood of
Residential and marital status	Autonomy	Mobility	Any work	Wage work
Women in households with no older woman
Co-resident husbands	1.46	0.91	0.53	0.14
Migrant husbands	2.77	2.05	0.66	0.22
Divorced/widowed women	4.06	2.33	0.75	0.46
Women in extended families with an older woman
Co-resident husbands	1.08	0.82	0.47	0.12
Migrant husbands	1.10	1.01	0.50	0.12
Divorced/widowed women	1.97	1.40	0.72	0.35

Source: Predicted estimates based on coefficients from Model 2, Appendix 2.

This table highlights several issues facing Indian women. First, when women live in an extended family, whether the husband is present or not, they are embraced within a family circle and bound by the same rules and customs as their peers whose husbands reside within the household. Second, widows and divorced women have different and unique experiences. Whether they live in an extended household or not, they are more likely to control their own lives and carry greater labour market responsibilities.

Table 5 presents descriptive statistics that shed some light on the nature of household composition. Each entry in the table represents the proportion of households in that cell which are extended, as defined by the presence of an older woman in the household. It indicates that among the women with co-resident husbands, about 34 per cent live in households with an older woman, but this proportion rises to 56 per cent among women with migrant husbands. Younger women are far more likely to live in extended families in their husbands’ absence than older women, as are rural women. But perhaps one of the most striking observations is that it is privileged women who are most likely to live with other relatives. For example, in results not reported here, residence in extended family for wives of migrants is significantly more likely for literate than illiterate women. Similarly, upper-caste Hindu women are far less likely to live alone in the absence of their husbands than the lower-caste dalit women. Interestingly, women’s extended-family living was not associated with the destination of husbands; wives of international migrants were as likely to live in extended families as wives of domestic migrants.

Table 5.

Proportion residing with older woman by socio-economic characteristics

Characteristic	Co-resident husband	Migrant husband	Widowed/separated	Total
All India	0.34	0.56	0.33	0.35
Age of the woman
20	0.69	0.94	0.86	0.70
30	0.47	0.73	0.70	0.49
40	0.26	0.34	0.33	0.26
49	0.14	0.19	0.13	0.14
Woman’s education
Illiterate	0.29	0.47	0.26	0.29
1–5 grade education	0.32	0.58	0.39	0.33
6–9 grade education	0.42	0.70	0.46	0.43
10 grade-some college	0.49	0.75	0.49	0.49
College graduate	0.46	0.64	0.58	0.47
Place of residence
Rural	0.37	0.58	0.33	0.38
Urban	0.28	0.42	0.34	0.29
Social group
Forward caste	0.39	0.62	0.38	0.40
Other backward classes	0.35	0.59	0.37	0.36
Dalit	0.32	0.48	0.28	0.32
Adivasi	0.32	0.37	0.22	0.31
Muslim	0.29	0.53	0.35	0.30
Christian, Sikh and Jain	0.39	n.a.a	0.35	0.40

^aNot available, cell size less than 50.

To summarize our findings, our results suggest that husbands’ migration has a substantial effect on women’s lives if they do not live in an extended household. When living independently, women are far more likely to make independent decisions regarding day-to-day living as well as longer-term decisions for children’s well-being, and they have greater physical mobility and independence than women living with their husbands. These women also face greater labour demands and are more likely to participate in the labour force. However, many women are likely to be incorporated into extended households when their husbands migrate and they do not experience these challenges as well as liberation from rigid gender rules. Moreover, it is women from the higher social classes who are more likely to be incorporated into extended households when their husbands migrate.

It is important to note that our results do not incorporate some confounding influences. Both migration and residential arrangements may be endogenous. Women who are more confident about managing their own lives may be more likely to encourage their husbands to migrate than those who are not used to making decisions on their own or are in ill-health. Males’ migration decisions may well depend on the availability of other family members to care for the families they leave behind (de Haan 2006). Similarly, women who prefer to be responsible for their own wellbeing and household decisions may be unwilling to live with other family members in the absence of their husbands. These caveats must be kept in mind while interpreting our results.

In evaluating these results, it is also important to note that this study is based on a national survey, which is useful in establishing broad patterns but fails to do justice to the complexities of the migration process and its implication for families. The India Human Development Survey provides a reasonably representative sample of migrants but does not necessarily capture different dimensions of the migration process. In particular it does not do justice to the context of migration in the sampled households, such as length of absence of the husband, nature of migration (seasonal, temporary, permanent), factors motivating the husband’s migration, migration levels within the community and cultural attitudes towards migration. All these factors are likely to have substantial bearing on the relation between solo male migration and empowerment of wives who are left behind. For instance, in regions where men migrate frequently, social institutions may have evolved to support women, reducing a need for them to move in with other family members. Also, length of absence may condition this relationship. In many areas of Uttar Pradesh, many men spend a couple of decades leaving their families behind, returning home for annual visits during festivals or harvest. In other areas, migration may be seasonal during summer months. These different patterns may lead to very different responses on the part of women and families.

It is also important to note that even with such a large sample, the number of women with migrant husbands is too small to permit analysis at a subnational level. Given the diversity of gender relations in India across regions (Jejeebhoy and Sathar 2001; IIPS and Macro International 2007), we have controlled for state-level differences in women’s autonomy and labour force participation by including dummy variables for state of residence in our analysis. However, not being able to interact these with husbands’ migration status is an important shortcoming of this analysis.

Discussion

In spite of the caveats noted above, a focus on broad patterns using large sample surveys serves an important function. Social science models are consistently being challenged by exceptional circumstances and are being modified in response. Research on separate finances for men and women in sub-Saharan Africa led to criticism of neoclassical household economic models and their assumption that the interests of different individuals in the household can be pooled within a single utility function (Folbre 1994). This resulted in reformulations focusing on intra-household distribution considerations (Thomas 1994). Similarly, focus on female-headed households in the United States and Latin America (Buvinic et al.1983) led to a need to move beyond research on male breadwinners and to pay explicit attention to the nature of women’s work (Lloyd 1991; Anker 1998), as well as to a redefinition of the concept of ‘household headship’ in census and other data collection efforts (Presser 1998). Results presented in this paper also have considerable implications for research on the nature of gender and patriarchy in developing countries. It highlights the need to focus on the mediating role of household structures and household members other than the husband in limiting the degree of autonomy and agency of young women.

Our results suggest that any empowering effect associated with husbands’ absence is limited to women who do not live in families with older women, our proxy for an extended household in which the respondent is a junior member. Although considerable attention is devoted to the role of differences in power between men and women, with some notable exceptions (Adams and Castle 1994; Sen et al. 2006) few studies have paid any attention to the role of age and generation in disempowering young women, a shortcoming highlighted by our results. Consequently, whether the husbands are living with the women or living in another state, women do not gain more freedom and autonomy as long as they are living in an extended family.

This diversity is also highlighted in the literature on transnational female migrants engaged in paid domestic and care work. Even though such workers gain a degree of personal and economic independence, migration fundamentally alters their familial experiences and they are also more likely to sacrifice their experiences of motherhood or remain single and childless (Hondagneu-Sotelo 1992; Constable 1999; Lan 2003). In contrast, migration does not necessarily bring about a change in the familial experiences of young migrant men. They continue to fulfil their roles as breadwinners for the family, and marry and have children. This suggests that it is not migration per se but the position of women in the social structure which disadvantages them, whether as solo female migrants or women who are left behind in the places of origin while their husbands migrate.

While not directly geared towards studying the role of social class in shaping gender inequality, these results also suggest that it is the higher social classes that are more reluctant than lower ones to allow women the freedom of living alone. Upper-caste women and literate women are less likely to live alone in their husbands’ absence than lower-caste and illiterate women. These findings echo the observations in urban Egypt (Hoodfar 1996), and suggest that situations where the grip of patriarchy might be relaxed are resisted with greater fervour by the upper social classes than those lower in the class hierarchy.

Finally, our results also point to the unique status of widowed and divorced women. These women seem to stand out even when they are part of an extended family. Unlike married women, widows and divorcees work to support themselves and their children even when they live in an extended family. They are also far more likely to have the freedom to make decisions that concern them or their children. While on the surface this may appear to mark greater opportunities for empowerment, when taken in conjunction with the vulnerabilities of Indian widows noted by other studies (Chen 2000), these highlight the ambiguous position of widows and divorcees in Indian families.