Women's autonomy in the context of Rural Pakistan.

Sathar, Zeba Ayesha ; Kazi, Shahnaz

The paper explores the elements that constitute women's autonomy in rural Pakistan. Hitherto most research on women's status in Pakistan has either been restricted to proxy measures of women's status generally or to the urban areas. Community or region, each of which has distinctive features, have an overriding influence on this subject. Northern Punjabi women have lower economic autonomy but greater mobility and decision-making authority than women in Southern Punjab. Gender systems at the village level are also important predictors of women's autonomy.

Economic class has a weak and ambivalent influence on women's autonomy in rural Punjab. Class influences both education and employment of women, these remains the routes to empowerment in rural settings. While most women in rural areas contribute economically, the majority works on the household farm or within the household economic unit. These women do not derive any additional autonomy as a result of this contribution. Paid employment, though offset by other restrictions on poor women, offers greater potential for women's autonomy. Education, on the other hand, has a lesser influence on female autonomy in the rural Punjabi context.

INTRODUCTION

South Asia as a region has been slow in experiencing demographic change and is also known for the sharp inequalities in the autonomy and power of men and women. Several researchers have hypothesised that increased gender inequalities and women's autonomy are strongly associated with social and demographic behaviour in South Asia [Caldwell et al. (1982); Dyson and Moore (1983); Morgan and Niraula (1995)]. The Indian states of Kerala and Tamil Nadu and countries like Sri Lanka with higher levels of women's autonomy have demonstrated rapid speed and a much earlier onset of demographic transition than in the rest of South Asia.

Women's autonomy has been studied in Pakistan in the context of its inhibiting influence on fertility and infant child mortality [Sathar et al. (1988)]. However, research on gender aspects of demographic change has relied on conventional measures of women's status such as education, spousal age difference, and employment [Sathar et al. (1988)], or has looked at son preference as a factor influencing the demand for additional children and infant and child mortality [Rukanuddin (1982); Mahmood and Ringheim (1993)]. While the link between female education and fertility or child mortality emerges strongly in most studies, the employment-demographic link is tenuous and is conditioned by type of employment [Sathar and Kazi (1990)]. Furthermore, education and employment among women are both strongly linked with household socio-economic conditions and, therefore, are not totally independent determinants of women's status [United Nations (1993)].

There is a noticeable gap in the literature generally, and in developing countries in particular, in assessing the numerous aspects of women's status. It is mainly due to large variations in the conceptualisation of autonomy. In other South Asian settings, women's autonomy has been linked to education [Jeffery and Basu (1996)], patriarchal structures [Cain (1979)], and marriage endogamy versus exogamy and dowry practices [Dyson and Moore (1983)]. The confounding influence of each context in laying out parameters of gender and social relations seems to be very important in other countries of South Asia [Morgan and Niraula (1995); Taj et al. (1995); Balk (1994)] and needs to be assessed for Pakistan. This country is seen by most western scholars as being influenced most strongly by the fact that it is almost entirely an Islamic society and is therefore viewed as a monolithic community. This assumption tends to lead to the homogeneous treatment of Pakistani women as powerless entities, unable to exercise their lives' decisions and reproductive rights.

In this paper we argue that the situation is indeed much more complex. While Pakistan has an almost homogeneously Muslim population, women's status is heterogeneous in terms of individual and community traits. This study aims to conceptualise women's autonomy in relation to that of men, but also in relative ranking with women from different rural communities in Pakistan. The basis of the sample selection is the hypothesis that gender norms are expected to differ by agro-ecological variations because of their different familial systems and modes of production. Furthermore, because women's empowerment and status are multi-dimensional, several measures have to be utilised to gauge women's status in various settings. Greater power or autonomy in one dimension and in one community is not expected to amount to the same in another. It is a first-step in operationalising terms such as the empowerment of women. At the moment these terms are amorphous and are perfunctorily but not meaningfully being included in programmes, largely as a result of the recent International Conference on Population and Development held at Cairo and the previous conference on women held at Beijing.

DATA

Data were collected in 1993-94 with the specific purpose of investigating--more comprehensively than before--the relationship between women's status and fertility in five Asian countries. In Pakistan, the ten communities chosen for the sample were located in the most populous province of Punjab. They were selected to represent the full range of agro-climatic, ethnic, and linguistic conditions in that province. The villages were roughly of the same size and were purposely selected to be located at considerable distance from any major roads. One peri-urban community was also chosen for purposes of comparison. For each village, about 100 currently-married women between the ages of 15 and 40 were randomly selected for detailed interviews, which lasted over an hour. About 50 percent of husbands were also interviewed. In addition, focus group interviews were conducted with both men and women in all ten communities to establish gender-related norms. The core of the data for this particular paper comes from the 1036 women's questionnaires completed in the course of the survey.

The women's questionnaire comprises a screening roster that collected household information and was used also to establish the number of eligible currently-married women in the household. Only one out of the eligible respondents was chosen per household using a Kish chart. The beginning of the interview collected background information. Sections followed this on birth history, contraceptive knowledge and use, and employment (including details of class of worker, seasonal and regular employment, paid and unpaid). Three special sections were included to cover all aspects of women's status that we could conceive of: intra-family relationships, marriage history and details of dowry, mobility, and gender issues. The last sections of the interview collected information on husbands (mainly socio-economic characteristics) and household possessions. We also draw on the initial information used to select the communities as well as community-level information collected through a specially developed questionnaire. Details about the differences in communities selected are described later.

METHODS

The variables used in the analysis fall into four main areas: community, women's status, socio-economic, and household characteristics. The ten villages are distinguishable by socio-economic development, cultural norms, agro-climatic zones, etc. They are grouped into five sub-regions on the basis of geographical location into barani (rainfed), semi-rainfed, peri-urban, Central Punjab, and Southern Punjab. Figure 1 presents the location of these villages and their grouping. Women's status for the purposes of this paper is further disaggregated into five areas: mobility, decision-making, access to resources, economic autonomy, and the relationship of fear of and communication with husbands. Socio-economic characteristics comprise the education of both spouses. Income, husband's occupation, and landholding are initially considered. The major questions being explored are:

* What are the essential components of women's autonomy in the context of rural Pakistan?

* What is the relative influence of community and socio-economic status on women's autonomy?

[FIGURE 1 OMITTED]

Various aspects of women's status are treated as an independent variable determined by a combination of socio-economic and community-level factors. In previous work, which sets out the analytical framework for this study, Mason (1984) has defined different aspects of women's autonomy. In the South Asian context, it would seem to be important to measure women's mobility, their access to resources, their access to paid employment, and most importantly, their ability to make decisions within the household and those concerning critical aspects of their life. Indices are compiled of what are considered some crucial aspects of autonomy of women: mobility, access to resources, decision-making inside and outside home spheres, economic autonomy, domestic violence, and interspousal communication. These indices are compiled as sums of scores on each of the constituent variables if the respondent herself could decide or do the following.

1. Mobility. Whether the respondent can go alone to the market, to the field, to the next village, to the health centre, to visit relatives. This index varies from zero to five.

2. Economic Autonomy. Whether the respondent can decide to work for money outside the home, whether she says she can survive on her own without the husband, and whether she controls her own income. This index varies from zero to three.

3. Access. Whether she can decide to buy a shalwar-kameez, whether she can decide on how household income is spent, and whether she can handle household money. This index varies from zero to three.

4. Decision-making. Whether the respondent is the major decision-maker about food to be purchased, children's education (as to where they should be sent and for how long), how many children to have, what to do when the child is sick, whether to make a major household purchase, whether to buy animals, whether women should work outside the home. This index varies from zero to eight.

5. Fear of Husbands. Whether women fear their husband, whether husbands have ever beaten them. This index varies from zero to two. This index, unlike the others, operates in an inverse direction, increasing to indicate a diminution of women's autonomy.

6. Communication with Spouse. Whether the respondent speaks to her husband about community affairs, about money matters, and about the number of children to have. This index varies from zero to three.

We examine some basic determinants of women's autonomy along with the role of communities. We use multiple linear regressions to disentangle their independent effects. Most explanatory variables are dummies except for education and age, which are continuous variables. Age, family structure, education of the respondent and husband, and the respondents' economic activity and their residence in communities at different stages of development are the explanatory factors.

FINDINGS

The Settings

Punjab province in Pakistan is the most developed in the country in terms of agricultural productivity, road structure, sanitation, communications, and availability of facilities. However, the rural areas of the province are by no means homogeneous in terms of communication, links with the outside world, educational levels, and exposure to media or income, and, we expect, women's status. All these factors are likely to vary across the ten rural communities located in ten different administrative districts of Punjab. For the purposes of this analysis, we shall regroup these communities into five major regions: Northern Barani (Rainfed), North Western mixed, Central, Peri-Urban, and Southern. This division is based on agro-ecological conditions [PARC (1980)] and their varying impact on farming systems and employment patterns, which directly impact on family and gender systems.

Barani Northern Villages

The three villages located in the rainfed or Barani region have been drawn from Attock, Gujar Khan, and Chakwal districts. They have certain features which, to some extent, are representative of the barani areas of Pakistan. Agriculture in the rainfed region is constrained by the uncertainty of water supply and is mainly geared to subsistence production, with the main crops being wheat, maize, and millet. Since farming does not offer an adequate and reliable source of income, the population has traditionally depended on alternate sources of livelihood, mainly in the armed forces and jobs in nearby urban centres. This region is known for the predominance of employment in the army and police. Due to the requirements of their work, a significant number of the husbands in the barani sites live away from home. The proportion of husbands living away is 37 percent (Table 1). The absence of male heads of household has important implications for the situation of women in these villages. In general, with the men away, women take over a substantial portion of the work in managing agricultural production. Women's participation in work on their own farm in barani villages is significantly greater than that found in the irrigated villages (69 percent as compared to 40 percent in Central Punjab). The extent of women's involvement in farm work outside the home is conditioned by the prevailing patterns of male employment as well as the cultural traditions of the areas. In general, these areas are better known for their earlier traditions for setting up of schools, and educational levels are historically better for these districts than for other parts of Punjab. The greater importance of education in the rainfed regions is reflected in the fact that primary schools for boys in these areas were established in the 1920s, decades before the advent of boys' schooling in the irrigated villages. The educational level of males in the barani regions is considerably higher than the average, with a significant number who have completed ten years of schooling, which is a prerequisite for entry into army and other government employment (Table 2).

North-west Semi-irrigated Areas

The two villages from this 'mixed' area, which is partially barani and partially irrigated, are drawn from Mianwali and Sargodha districts. The topography of this area is a mixture of fairly fertile land mixed with arid and saline areas. Many features of this area resemble the barani areas but agricultural output is greater. Because of this the major occupations are less tilted towards non-farm sector jobs, even though they are more prominent than in the irrigated areas (Table 1). We found our two villages located in this region to be quite conservative in social values and less exposed to outside communications as compared to the barani villages. The Mianwali village stood out in our sample as the most conservative in terms of the observance of purdah among women; the other in Sargodha was known widely for religious schools. Work participation among women was much lower than in the barani areas, as were the levels of husbands living away from home. The educational attainment of both men and women is lower than in barani areas, but still higher than in the irrigated areas (Table 2).

Central Punjab

The two villages selected in this region were from Faisalabad and Sahiwal areas. The employment and income pattern of inhabitants in the canal-irrigated areas located in Central Punjab is quite different from the rainfed

areas. Farming is market-oriented and cash crops are grown, including cotton in the Southern Punjab and sugarcane in central Punjab. While agriculture is a lucrative source of income, access to land varies across the sites. The landless residents of the irrigated regions are mainly employed as labourers in factories or as casual workers in the agriculture and construction sectors. In contrast to the barani regions, male employment in army/police and white-collar jobs is quite limited (Table 1). The pattern of female employment (Table 2) also indicates distinct differences between the barani and irrigated regions. While women's involvement in agricultural production on their own farms is markedly higher in the barani villages, wage work for cash income is more predominant in the irrigated regions.

Education, highly valued in the barani context as a means to better non-agricultural employment opportunities, is not so directly relevant to the means of livelihood in the canal-irrigated agriculturally prosperous areas of Central Punjab. Levels of schooling among men are much lower in irrigated areas. Female education is generally low: on average more than 80 percent of the respondents had received no education. However, when we examine the trends in education, then more women from the younger cohorts had education as compared to older cohorts in Central Punjab in particular, reflecting a recently growing demand for schooling in these areas [Kazi and Sathar (1996)].

Peri-urban

At the upper end of the spectrum, the most developed site in the sample was the peri-urban community of Rakh Kikrian. This community mainly comprised recent rural migrants who were working in the city of Gujranwala. There was barely any agricultural activity, with only 3 percent of the husbands working as farmers and only 13 percent of the households owning land (Table 1). The residents of the peri-urban site were not better educated but had easy access to the facilities located in the major city of Gujranwala, including secondary schools, hospitals, etc. Exposure to mass media was relatively widespread: more than 43 percent of the respondents had watched television in the last week. Housing conditions were more typical of a low-income urban neighbourhood than a rural site. Other settings which ranked high in terms of various indicators of development, particularly the spread of education, the availability in the village of girls' schooling beyond the primary level, exposure to mass media, and the economic status of the household, were located in barani areas and in central Punjab.

Southern Punjab

The least 'developed' and most agricultural area was definitely South Punjab. The two villages chosen from this area were from Multan and Bhawalpur districts. The largest proportions of uneducated women were concentrated in Southern Punjab, followed by the North Western semi-irrigated areas (Table 2). The most prominent agricultural crop is cotton growing in the villages of Southern Punjab. Cotton picking is almost entirely done by women labourers. In addition, female farm labour is also commonly used for harvesting of chilies and vegetables which are mainly grown in irrigated areas. Other main sources of cash income for women in Southern Punjab (and also Central Punjab) is the embroidery of dopattas, shoes, making of footballs, etc., for which there is a large market in the nearby major urban centres. Ali Kharak, in Bhawalpur, was the least developed and most insulated village in the sample. Although located on a metalled road with some minimum basic social infrastructure (primary schools and a non-functional health centre), it was the only site without any electricity.

There are also considerable differences between sites in the level of development as reflected in access to basic facilities and exposure to outside influence. Each of the sites had government primary schools for boys and girls; however, availability of secondary schools and health centres was limited to a few selected sites. Samote in barani areas was the only village with a girls' secondary school, while government-run health centres were found in five of the villages. Family Planning clinics were only available in the peri-urban site of Rakh Kikrian and in one of the villages of Central Punjab. To some extent, the location of both schools and family planning clinics reflects a substantially developed demand for their services. And we can safely conclude that Southern Punjab is least served by those facilities, followed by the barani areas, Central Punjab, and then the peri-urban area due to its proximity to an urban centre.

The level of exposure of the rural communities to the world outside, which affects attitudes and relative openness to modern ideas, also varied considerably. The extent of exposure depends on a number of factors including road access and proximity to large cities, employment in urban centres, as well as access to sources of mass media, particularly radio and television. The barani villages were particularly disadvantaged because they were not connected by a proper road, transport facilities were inadequate and irregular, and travel time to the nearest city took as long as 3 hours. The proportion of respondents in these villages who had watched television in the last week was also below the average for the total sample. This may be offset by employment of male residents in urban areas, and was more commonly found in these villages, which would be expected to lead to greater receptiveness to urban values and lifestyles. Exposure to television was the highest in the peri-urban areas, followed by Central Punjab.

Components of Women's Autonomy in Rural Pakistan

As already raised in the introduction, the issues of gender and women's autonomy and their status have to be defined in the context of a particular setting. In the case of rural Pakistan, it was expected that in order to capture the full extent to which women had power or to assess the freedom they had to take decisions, we had to look at several different components. The first was their ability to move about freely without chaperones and without permission of other gatekeepers such as husbands and mothers-in-law. We use answers to such questions as whether a woman can go alone without permission to a health centre, to a local market, and to the next village to assess her ability to participate in life outside the home. The underlying assumption is that when women are unable to leave their homes, their sphere of activity is restricted to the homestead. This index assesses the extent of flexibility in movement from a value of zero for not leaving the home for any of these pretexts to five if they are totally mobile to go almost anywhere, it is apparent that women are more likely to be able to go to the fields alone in Punjab than they are to be able to access a market or health centre (Table 3). This does have implications for women's access to major services.

Access to resources is to a large extent pre-determined by resources available to the family. However, this measure also relates to the differential share of the same household resources by men and women in the same households. We use measures which are less likely to be biased in favour of wealthier households by using the question of whether a woman can purchase herself a shalwar-kameez (local dress) and whether she has a say in household finances. Since clothing has to be purchased as a basic commodity and all households have some budget however meager, we utilise this measure to assess as to what extent women have access to resources available to the household. The proportion that can buy a piece of clothing independently varies between 32 and 55 percent, whereas the proportion that has a say in household expenses is higher, between 53 and 67 percent.

Decision-making within the household is another critical aspect of participation and power sharing between household members. At one extreme are women who do not have any say even in the most trivial of decisions such as choice of what to cook, and at the other end are women who make the majority of decisions in the household. To a large extent, this measure is likely to be influenced by household living arrangements, presence of husband in the household or a mother-in-law or father-in-law, and by lifecycle situation, such as whether a woman is a new bride or the senior woman with grown- up children. It is clear that women are more likely to make decisions related to children and within the home rather than decisions in the sphere outside the home and particularly those involving money matters (Table 3). For instance, about 65 percent of women have a say in the number of children to have but about 16 percent have a say in a major household purchase.

Some measure of economic independence was thought to be an essential component of women's empowerment. Evidence from elsewhere but particularly from Bangladesh shows that a major factor in social change has been the involvement of women in paid work especially in the non-farm sector. While a wave similar to that of the Bangladesh garment industry had not come to rural Pakistan, we can still gauge whether work outside the home and work which is remunerated enhance women's empowerment. Measures used to construct this index are based on whether a woman feels she can survive on her own without her husband, whether she feels she can make the decision to work outside the home herself, and whether she controls her own income. This measure is generally low for women as they are generally dependent on men. Between 31 and 41 percent women can decide to work outside the home, and about 61 percent feel they can survive without their husbands financially if they need to.

We look next at an increasingly important area of women's position, that of a relationship of fear of husbands and of domestic violence. Here we look at answers from women about whether they often are afraid to disagree with their husbands and whether they are ever beaten or beaten regularly. In this most sensitive of areas may lie a keen indicator of the inequality or the degree of repression of women and deference to men they are living with and dependent on. It is fairly alarming that the majority of women are afraid of their husbands and about one-third have been beaten by husbands. Domestic violence in actual terms affects one-third of women and potential fear of husbands is felt by the majority of Punjabi women.

Lastly, we look at the dimension of communication between spouses. Elsewhere it has been argued that this is an important aspect of the relationship between the genders. Here our assessment is based on women's responses to whether they can talk to their husbands about various matters, ranging from community affairs, money matters, and the number of children they want to have. Communication is generally quite good, with above 55 percent of women able to talk to their husbands about most matters.

In the following sections we focus on whether (1) the measures of women's status vary across these distinctly different regions, (2) whether they vary in the same direction, and (3) whether they vary according to the levels of development or outside exposure.

Women's Autonomy: Variations across Communities

We begin by assessing whether in fact there are important variations in various measures of women's autonomy and gender inequality across the communities and whether they necessarily move in the same directions. In Table 4 it is clear that there are significant differences across the five regions in these indices. This table presents the six measures of women's status: mobility, access to resources, decision-making, economic autonomy, fear of husbands, and communication with spouse. While some regions rank higher in terms of certain indices, they do not rank particularly high on others, thus confirming that women in each setting are not 'empowered' in all departments necessarily.

Mobility is the highest in the peri-urban and Central areas, where women are able to go to more than two out of the five places used to compile this index. Mobility is the lowest in the South of Punjab and in the North West semi-irrigated areas, a fact corroborated by the much greater prevalence of 'purdah' in these regions. In these regions, women are barely able to go to one place alone. To a large extent, mobility of women is associated with their participation in life outside the home and is a strong indicator of their freedom to access important places and spheres otherwise restricted and outside their reach or control. Thus women in the peri-urban and Central Punjab areas are at an advantage in this important sphere.

Access to resources follows quite a similar pattern to decision-making but here women in barani areas are the ones with the greatest access to resources. As mentioned earlier, males in these communities are employed outside the villages, which seems to give women more authority to make transactions. Access to resources is the lowest in Southern Punjab and North Western Punjab, much along the lines of mobility and decision-making.

To some extent, the same pattern is seen with decision-making but here the peri-urban women have a clear edge followed by women in the Northern barani areas, with the other three areas lying at similar levels. The barani women's position is probably relatively higher in this index purely because of the absence of husbands from home in these areas, leaving women to make decisions or at least make more decisions in their absence purely for practical reasons. Decision-making in financial decisions such as purchase of major household goods and livestock is the highest in barani villages but is distinctly lower in decisions within the household such as purchase of food and marriage of children. Women in Central Punjab and in the peri-urban areas are more likely to make decisions on both the inside and outside spheres while decision-making by women is totally restricted in Southern Punjab.

Perception of economic autonomy is most certainly the highest in barani areas and in Southern Punjab, which are also regions where labour force participation of women is the highest. In Southern Punjab this is because of greater opportunities for cash income and because work on farms is more lucrative. In barani areas women are substituting men in subsistence agriculture and are less likely to be paid, and therefore less autonomous economically [Sathar and Kazi (1997)].

The indicator of freedom from fear of or threat from husbands presents a twister to the pattern with a much less fortunate situation of women from peri-urban areas. These women have a very high score on this indicator. It has been posited that the break away from rural and traditional links unfortunately amounts to some negative fallouts such as the weakening of social censure and increases in domestic violence. Wife-beating is a phenomenon which seems to be more prominent in urban areas. In contrast, women in Northern Punjab (both barani and semi-irrigated) are relatively much more free of this threat while women in Central and Southern Punjab score somewhere in between.

Communication between spouses presents yet another dimension of women's autonomy in their ability to express themselves freely with their spouses. Communication seems much more free and intimate in Southern and Central Punjab than in Northern Punjab. This is probably because of the frequent absence of spouses in the latter areas.

Women's autonomy measures in the areas of mobility, economic and non-economic decision-making, and access to resources vary notably across the sub-regions. Furthermore, what contributes to greater autonomy in one direction does not necessarily concur with other factors to allow us to conclude that women in one setting are definitely more autonomous than others. Overall, women in the peri-urban, Northern barani areas, and Central Punjab setting do rank higher on the autonomy measures than women in semi-irrigated areas, and particularly those in Southern Punjab who (despite opportunities for paid work) are unable to compete in other spheres. However, the wider prevalence of domestic violence in the peri-urban site throws a question-mark on its otherwise uniformly highest ranking in terms of other indicators.

Women's Autonomy and Socio-economic Status

As proposed in some earlier work, it is necessary to establish that women's autonomy is indeed distinguishable from socio-economic status [Mason (1984)]. Education of women is the measure most widely used a measure of their relative status and autonomy [Jejeebhoy (1995); Jeffery and Basu (1996)]. In the Pakistani context, women's education has been found to be a strong explanatory factor for differences in contraceptive use, infant mortality, and children's schooling levels [Sathar et al. (1988); Sathar (1987); Mahmood and Ringheim (1993)]. However, its effect was found to be stronger in the urban areas and would be important to explore in this rural sample [Sathar and Mason (1993)]. Respondent's education is strongly positively associated with access to resources but is very weakly associated with mobility. Education is hardly associated at all with the perception of economic autonomy and decision-making inside or outside the home. This is the first indication that female education, perhaps because of its low overall attainment levels in rural areas, is not a strong indicator of women's status as was the case in urban Pakistan [Sathar and Mason (1993)]. While strongly associated with economic class, female education in rural areas in 1993 was not a widespread phenomenon, nor was it associated with greater opportunities for paid employment except in the case of a handful of school teachers.

In the case of urban women, those employed in the modern sector had higher status, whereas in rural Pakistan, the respondent's paid employment is associated positively with all indices. Working for income in the last year is associated with increased mobility and a greater role in both outside and inside decision-making. However, there is no association between working for income and access to resource, which once again seems a class-related issue with only well-off women being able to answer this positively; just as in the case of urban employment mere categorisation of women by employment status does not seem important [Sathar and Kazi (1989)]. A better classification of women's employment appears to be their classification by whether their work is paid or unpaid and whether it is done at home or outside the home. This classification has strong associations with the measures of autonomy. In particular, mobility of those who do no economic work or work inside the home or do unpaid work is much lower than those who do paid work outside the home. in contrast, access to resources of women doing no economic work is the highest, indicating that it is related to household wealth rather than income generated by women themselves. Access to resources is the lowest, however, for unpaid family workers. Economic autonomy, not surprisingly, is the highest among women when they are paid, whether inside or outside the home. So is the decision-making higher among women who are paid even if they work from within their homes. The contradiction once more is in that levels of fear are also higher among women who are doing paid work. Fortunately, this is accompanied by a greater level of communication between spouses.

Class (as measured by income and size of landholding) is expected to be associated with women's autonomy measures. Size of landholding is considered particularly relevant in these rural settings. In general, mobility of women is inversely related to economic class. This is an expected finding as women (controlling for community norms for women's movement and purdah) from poorer households are generally more able to move freely outside the household. A typical observed pattern is of greater seclusion of women as households become wealthier as this is regarded as a 'desirable' characteristic of households with a higher status. This inverse relationship is seen with total income quintiles but not with the size of landholding. Access to resources, surprisingly, is not associated with economic status. Decision-making has a negative association with household economic status, but economic autonomy is greater among poorer women probably because they are more likely to be engaged in economic activity to supplement household income. While communication between spouses is hardly related to economic status, the fear of husbands is lower among households with a higher economic status, as is the prevalence of domestic violence.

We also explored more the relationships between some other good proxies for women's status, i.e., and our direct measures of various aspects of women's autonomy. Amongst them are ages of women, residence near natal kin, and family structure, which are also expected to influence women's autonomy, as documented in other research in South Asian contexts [Vlassoff (1992); Caldwell et al. (1982); Dyson and Moore (1983)]. Age represents the life-cycle position of women in their families. It is usually a good indicator of whether a woman has just joined a new household as a bride, or has established herself as the senior daughter-in-law or is the mother-in-law with married sons. Not surprisingly, age has a strong positive association with all of the autonomy indicators. Older women are much more mobile, have greater access to resources in the family, likely to be economically autonomous, and to make decisions both in the inside and outside spheres. Though this was probably an expected finding, the strength and consistency of the association with all indices of women's autonomy is unexpected. It also confirms that our indices are measuring authentic features of women's lives which would make them 'freer' and more independent as they grow older.

Family structure, which can be argued to be partially though not entirely a function of age, also has a strong association with the women's autonomy measures. The presence of in-laws in the household clearly inhibits them in all these aspects of their lives. Household living arrangements are worth investigating in depth as they clearly lie at the heart of gender and social systems which impact on women's autonomy. Women living in nuclear households (as opposed to those with brothers- and sisters-in-law or parents-in-law) are much more mobile, have greater access to resources, and are able to make more decisions both in the inside and outside home spheres. While family structures are usually reflective of life-cycle phases, with older couples breaking away to form their own household and younger couples living with parents or older siblings, migration away from natal homes also leads to greater prevalence of nuclear arrangements [Kazi and Sathar (1996)].

Exogamous and endogamous marriages (reflected in the variable of whether a woman is still living in her place of birth after marriage) have been established to have profound implications in the Indian literature on women's autonomy [Vlassoff(1992)]. Those women living nearer their place of origin were found in India to have closer links with their natal family which leads to support and greater autonomy or freedom. However, in Pakistan, there is almost entirely a prevalence of endogamy and consanguineous marriages or marriages with relatives, which leads to virtually village endogamy. Since the majority of women in our sample were married to cousins or relatives and were living in the village in which they were born, we found that residence in natal village had no impact on autonomy measures in the Punjabi context. On the other hand, the wide prevalence of "wata sata" or exchange marriages may be strongly associated with whether the woman is living after marriage in the same village where she was born or whether she has moved in from elsewhere. There are other fallouts of such marriages, noted by many of our respondents whose marriages were undergoing negative spin-off pressures because of problems in their siblings' marriage.

There is clearly a complex relationship between socio-economic status and women's autonomy measures, in Table 5 we present the analysis of the determinants of different components of women's autonomy by applying linear regressions to each of them and including the different contexts as dummies, along with the respondent's and husband's education, age, family structure, and work inside and outside the home as independent variables. Age and family structure appear to influence almost all the aspects of women's autonomy and their influence is positive even in enhancing fear of husbands. Respondent's education does not seem to be as strong an explanatory factor except for access to resources and for reducing fear of husbands and for increasing communication with spouse. Husband's education is of no statistical significance in explaining variations in autonomy but, surprisingly, its coefficient is often negative rather than positive. Income per capita also has hardly any influence on women's autonomy. Women's work, on the other hand, does appear to be strongly associated with women's autonomy, positively in the case of autonomy, negatively in the case of access to resources, positively with economic independence from husband, and positively associated with fear of husbands.

The dummies for the regions do retain statistical significance, indicating that community effects are strong in setting norms regarding mobility and economic opportunities for women and levels of education of both men and women. Thus many of the parameters of what women can or cannot do are set by communities which reflect the agrarian and developmental setting in which they are living. Therefore, the context is most important in setting out the possibilities for women's autonomy. Women in barani areas may enjoy making certain household decisions but do not enjoy economic autonomy due to their limited work opportunities. Mobility is much higher in Central Punjab, a factor which emerges as important in accessing markets and as well as sources of health care and family planning. Age and family structure and education and employment (to a large extent determined by economic class) also influence the women's position in the household directly, but their influence is conditioned by the context.

CONCLUSIONS

Hitherto most research on women's status in Pakistan has either been restricted to proxy measures of women's status [Satharetal. (1988)] or to urban Pakistan [Sathar and Mason (1993)]. The current paper explores the area to see what constitutes women's autonomy in rural settings. Little social change has been documented in the rural versus urban areas particularly in fertility [Population Council (1995)]. The importance of community or regions, each of which has distinctive features, is an overriding influence. Anomalies of certain villages, such as cultural differences, may alter a particular village's standing but, otherwise, Northern Punjabi women have lower economic autonomy but greater mobility and decision-making authority than women in Southern Punjab. Thus sub-regional differences are important in setting the stage, although individual villages may still behave distinctly.

Gender systems at the village level are important predictors of women's autonomy. Our findings in Pakistan concur with other recent literature on South Asia [Balk (1994); Morgan and Niraula (1995)]. Other factors such as exposure to urban influence and communication channels, which also operate at the community level, are as important as women's autonomy; since they can also bring modern influences to bear on gender systems.

Economic class has a weak and ambivalent influence on women's autonomy. While it influences both the education and employment of women, these in contrast seem to be the routes to empowerment in rural settings. Average levels of schooling are extremely low in rural Punjab: they vary from uneducated proportions of 90 percent to 75 percent. The context is important because in the barani settings the levels of education, both of men and particularly of women, are quite high. But the opportunities of paid employment are few and far between for women who mainly fill in the agricultural roles of men, who tend to seek off-farm employment outside the village. This confers on them some additional decision-making power, but it also is likely to bind them to living in non-nuclear arrangements which restrict their authority within the household. Thus education is not necessarily a means to greater female autonomy in the rural Punjabi context.

On the other hand, Central and Southern Punjab have lower educational levels of women, but greater opportunities particularly in the latter area of paid employment. While these opportunities are mostly availed of by the poorest strata of women working in agriculture, the case of paid employment within the home seems to be a venue of economic empowerment of women of a slightly higher socio-economic status. While most women in rural areas do contribute economically, a majority of them do work for the household farm or within the context of the household as the economic unit. These women do not derive any additional autonomy as a result of this contribution. Paid employment, although offset by the restriction of poor women due to poverty in the household, has a greater potential for autonomy within the household.

REFERENCES

Balk, D. (1984) Individual and Community Aspects of Women's Status and Fertility in Rural Bangladesh. Population Studies 48:1 21-45.

Cain, M. et al. (1979) Class, Patriarchy and Women's Work in Bangladesh. Population and Development Review 5:3 405-438.

Caldwell, J. et al. (1982) The Causes of Demographic Change in Rural South India: A Micro Approach. Population and Development Review 8:4 689-727.

Dyson, T., and M. Moore (1983) On Kinship Structure, Female Autonomy and Demographic Behaviour in India. Population and Development Review 9:1 35-60.

Jeffery, R., and A. Basu (1996) (eds) Girls' Schooling, Women's Autonomy and Fertility Change in South Asia. New Delhi: Sage Publications.

Jejeebhoy, S. (1995) Women's Education, Autonomy and Reproductive Behaviour: Experience from Developing Countries. Oxford: Clarendon Press, UK.

Kazi, Shahnaz, and Z. A. Sathar (1996) Gender and Development: Searching for Explanations for Fertility Changes in Rural Pakistan. Comparative Perspectives on Fertility Transition in South Asia 1-28.

Mahmood, N., and K. Ringheim (1993) Desired Fertility in Pakistan: What is the Influence of Husbands? Paper presented at the Population Association of America Conference, Cincinnati, Ohio.

Mason, K. (1984) The Status of Women. Rockefeller Foundation, New York.

Mason, K. et al. (1995) Determinants of Women's Power and Autonomy in Five Asian Countries. Paper presented at the Population Association of America Conference.

Mason, K. (1996) A Comparative Study of the Status of Women in Five Asian Countries. Paper presented at the International Symposium: Life and Earth in the 21st Century, Tokyo, Japan, 4 March.

Morgan, P., and B. Niraula (1995) Gender Inequality and Fertility in Two Nepali Villages. Population and Development Review 21:3 541-561.

Pakistan Agricultural Research Council (PARC) (1980) Agro-Ecological Regions of Pakistan. Islamabad.

Population Council (1995) Pakistan Contraceptive Prevalence Survey 1994-95. Islamabad.

Rukanuddin, R. (1982) Infant-child Mortality and Son Preference as Factors Influencing Fertility in Pakistan. The Pakistan Development Review 21:4 297-328.

Sathar, Z. A. (1987) Seeking Explanations for High Levels of Infant Mortality in Pakistan. The Pakistan Development Review 26:1 55-70.

Sathar, Z. A. et al. (1988) Women's Status and Fertility Change in Pakistan. Population and Development Review 14:3 415-432.

Sathar, Z. A., and S. Kazi (1989) Female Employment and Fertility: Further Investigation of an Ambivalent Association. The Pakistan Development Review 28:3 175-193.

Sathar, Z. A., and S. Kazi (1990) Women, Work and Reproduction in Karachi. International Family Planning Perspectives 16:2 66-9,80.

Sathar, Z. A., and K. Mason (1993) How Female Education Affects Reproductive Behaviour in Urban Pakistan? Asia and Pacific Forum 6:4 93-103.

Sathar, Z. A., and S. Kazi (1997) Women's Autonomy, Livelihood and Fertility in Rural Punjab. Pakistan Institute of Development Economics, Islamabad.

Smith, H. et al. (1995) The Impact of Women's Status on the Unmet Need for Contraception in Five Asian Countries. Paper presented at the Population Association of America Annual Meeting, San Francisco.

Taj, A. et al. (1995) Fertility, Patriarchy and Development in India. Population and Development Review 21:2.

United Nations (1993) Women's Status and Fertility Change in Pakistan: The Most Recent Evidence. United Nations Population Division, New York.

Vlassoff, C. (1992) Progress and Stagnation: Changes in Fertility and Women's Position in an Indian Village. Population Studies 46:2 195-212.

Zeba Ayesha Sathar is Chief of Research at the Pakistan Institute of Development Economics, Islamabad, Shahnaz Kazi is Chief of Research at the Pakistan Institute of Development Economics, Islamabad, and also works for the World Bank. Table 1 Characteristics of Households by Five Regions Semi- Peri- Barani irrigated urban Landless Households % of Households which do not Own Land 24.7 58.9 87.3 Mean Size of Landholding 3.7 8.7 7.3 Husband's Occupation Army/Police/Professional 26.1 38.4 8.0 Business/Commercial 8.6 6.3 11.0 Wage Labourers 38.4 35.2 52.0 Farmers 20.5 12.1 3.0 Misc. 34.8 8.4 24.0 Migrants (Outside Country) 3.0 0.5 2.0 Households where Husbands Are Away from Home 34.0 36.2 40.0 Central Southern Punjab Punjab Landless Households % of Households which do not Own Land 64.0 60.3 Mean Size of Landholding 7.5 11.0 Husband's Occupation Army/Police/Professional 9.9 3.9 Business/Commercial 10.4 8.2 Wage Labourers 35.2 50.2 Farmers 22.4 24.7 Misc. 20.3 12.1 Migrants (Outside Country) 1.0 0.7 Households where Husbands Are Away from Home 43.2 21.4 Table 2 Characteristics of Respondents and Husbands Semi- Peri- Batani irrigated urban Husband's Education None 27.4 29.2 47.1 1-5 Years 18.4 13.4 19.6 6-10 Years 28.1 21.8 12.7 11 Years or More 26.0 35.6 20.6 Mean Years of Schooling 5.7 6.0 3.9 Respondent's Education None 72.2 87.1 76.0 1-5 Years 20.5 9.4 15.7 6 Years or More 7.3 3.5 7.8 Mean Years of Schooling Respondent's Work Status Unpaid Work 61.1 41.6 13.7 Paid Work 14.6 2L8 26.5 Respondent's Exposure to Mass Media % Who Listened to the Radio 22.9 10.9 27.5 % Who Watched TV Last Week 26.7 23.8 43.1 Central Southern Punjab Punjab Husband's Education None 32.5 63.9 1-5 Years 30.2 19.1 6-10 Years 19.2 7.5 11 Years or More 28.1 9.5 Mean Years of Schooling 5.3 2.8 Respondent's Education None 70.9 94.2 1-5 Years 21.2 4.1 6 Years or More 7.9 1.7 Mean Years of Schooling Respondent's Work Status Unpaid Work 29.6 16.6 Paid Work 36.9 64.3 Respondent's Exposure to Mass Media % Who Listened to the Radio 22.2 27.5 % Who Watched TV Last Week 34.5 27.5 Table 3 Women's Autonomy Indicators by Region Semi- Peri- Barani irrigated urban Domestic Decision-making Proportion who had a say in: Purchase of Food 67.4 65.3 76.5 Number of Children to Have 63.9 64.4 65.7 Schooling of Children 58.0 49.5 59.8 Marriage of Children 49.7 47.5 57.8 Financial Decision-making Proportion who had a say in: Major Household Purchase 26.0 15.3 16.7 Women's Work Outside Home 39.5 31.2 40.2 Sale and Purchase of Livestock 27.8 18.8 26.5 Access to Household Resources Proportion who: Have a say in Household Expenses 67.2 52.5 64.7 Have Cash to Spend on Household Expenses 68.4 69.8 76.5 Can Buy Clothing Independently 54.9 35.3 34.7 Can Buy Jewelry Independently 24.7 11.4 17.6 Can Buy Gigs for Relatives Independently 35.5 18.3 15.7 Mobility Proportion nho can go alone to: Fields 51.4 33.3 41.2 Market 35.8 26.4 50.0 Health Centres 27.1 17.4 41.2 Next Village 13.5 7.0 20.6 Fear of Husband Proportion who say they are: Afraid to Disagree with Husband 76.7 84.1 86.3 Often Afraid of Husband 25.7 32.0 36.4 Ever Beaten by Husband 28.8 25.9 54.9 Beaten Regularly by Husband 4.9 4.5 13.7 Communication Proportion who say they can and talk to their husband: (a) About What Happens in Community 58.6 55.0 71.6 (b) About How to Spend Money 62.9 68.3 80.4 (c) About Number of Kids to Have 58.5 66.8 88.1 Central Southern Punjab Punjab All Domestic Decision-making Proportion who had a say in: Purchase of Food 73.9 76.3 71.2 Number of Children to Have 67.0 65.1 65.1 Schooling of Children 57.1 44.8 50.4 Marriage of Children 57.9 49.4 51.5 Financial Decision-making Proportion who had a say in: Major Household Purchase 10.3 11.2 16.5 Women's Work Outside Home 39.5 41.5 38.5 Sale and Purchase of Livestock 14.0 17.0 20.8 Access to Household Resources Proportion who: Have a say in Household Expenses 58.1 53.5 59.1 Have Cash to Spend on Household Expenses 73.9 68.0 70.5 Can Buy Clothing Independently 32.0 32.1 39.1 Can Buy Jewelry Independently 15.3 10.0 16.1 Can Buy Gigs for Relatives Independently 23.2 18.3 23.7 Mobility Proportion nho can go alone to: Fields 59.6 29.0 43.2 Market 54.2 20.3 35.3 Health Centres 45.3 17.0 27.8 Next Village 16.3 5.0 11.5 Fear of Husband Proportion who say they are: Afraid to Disagree with Husband 83.2 85.7 82.0 Often Afraid of Husband 35.1 51.2 36.1 Ever Beaten by Husband 35.6 39.9 34.6 Beaten Regularly by Husband 8.4 8.3 7.1 Communication Proportion who say they can and talk to their husband: (a) About What Happens in Community 69.3 72.6 64.5 (b) About How to Spend Money 81.1 81.3 73.5 (c) About Number of Kids to Have 79.9 78.4 71.9 Table 4 Women's Autonomy by Area of Residence and Socio-economic Status Mobility Access (0-5) (0-3) Household Income Quartiles 1 1.75 1.26 2 1.93 1.28 3 1.82 1.25 4 1.72 1.27 5 1.51 1.27 Education of Respondent 0 1.75 1.22 1-5 1.56 1.26 6+ 1.87 1.51 Women's Work No Economic Work 1.53 1.40 Unpaid Work 1.74 1.16 Paid Work Outside Home 2.23 1.27 Paid Work Inside Home 1.47 1.35 Family Structure Nuclear 2.24 1.56 Non-nuclear 1.38 1.03 Landholding Size None 1.73 1.29 <2 1.78 1.32 2-5 1.64 1.11 5-11 1.77 1.20 12+ 1.94 1.20 Husband's Education 0 1.75 1.21 1-5 1.77 1.29 6+ 1.75 1.29 Overall 1.74 1.26 Economic Autonomy Incision-making (0-3) (0-8) Household Income Quartiles 1 .95 1.86 2 .88 1.75 3 .82 1.46 4 .84 1.27 5 .87 1.40 Education of Respondent 0 .89 1.62 1-5 .70 1.06 6+ .83 1.42 Women's Work No Economic Work .63 1.54 Unpaid Work .77 1.32 Paid Work Outside Home 1.32 1.97 Paid Work Inside Home 1.07 1.93 Family Structure Nuclear .95 2.14 Non-nuclear .81 1.14 Landholding Size None .91 1.75 <2 .82 1.42 2-5 .80 1.33 5-11 1.00 1.50 12+ .80 1.03 Husband's Education 0 .94 1.76 1-5 .89 1.90 6+ .82 1.37 Overall .87 1.57 Fear Communication (0-2) (0-3) Household Income Quartiles 1 1.20 2.08 2 1.25 2.03 3 1.17 2.16 4 1.08 2.08 5 1.11 2.12 Education of Respondent 0 1.20 2.07 1-5 1.21 2.10 6+ .93 2.15 Women's Work No Economic Work 1.08 2.06 Unpaid Work 1.12 2.00 Paid Work Outside Home 1.28 2.22 Paid Work Inside Home 1.30 2.27 Family Structure Nuclear 1.20 2.21 Non-nuclear 1.13 1.99 Landholding Size None 1.24 2.17 <2 1.07 2.07 2-5 1.08 1.86 5-11 1.10 1.91 12+ 1.07 2.01 Husband's Education 0 1.23 2.11 1-5 1.29 2.21 6+ 1.09 2.04 Overall 1.16 2.08 Table 5 Determinants of Women's Autonomy Indicators Mobility Access Age .24 *** .21 *** Family Structure (Nuclear = 1) .15 ** .19 ** Years of Female Schooling .01 .09 *** Husband's Years of Schooling .00 .03 Income Per Capita .03 .04 Women's Work Inside the Home .07 * -.09 Women Work Outside the Home .12 *** -.05 ** Area (Omitted South = 1) Barani .14 *** .116 ** Semi-irrigated -.01 -.03 Peri-urban .14 ** -.01 Central .21 -.05 Economic Autonomy Decision-making Age .04 .22 *** Family Structure (Nuclear = 1) .08 ** .23 *** Years of Female Schooling .03 .04 Husband's Years of Schooling -.01 -.04 Income Per Capita .00 .02 Women's Work Inside the Home .14 *** -.04 Women Work Outside the Home .36 *** .03 Area (Omitted South = 1) .0 Barani -.06 .06 Semi-irrigated -.14 ** -.02 Peri-urban .01 .06 Central -.06 * -.03 Fear Communication Age .06 * .14 ** Family Structure (Nuclear = 1) .00 .07 ** Years of Female Schooling -.11 ** .07 ** Husband's Years of Schooling -.05 .05 Income Per Capita -.04 -.02 Women's Work Inside the Home .05 .03 Women Work Outside the Home .09 ** .02 Area (Omitted South = 1) Barani -.08 ** -.27 *** Semi-irrigated -.07 * -.19 ** Peri-urban .11 *** -.02 Central -.01 -.08 ** * Significant at > .1. ** Significant at > .01. *** Significant at >. 001.