(introduction...)

The argument so far has been that an exclusive focus on poverty outcomes very often means that the processes leading to poverty are either overlooked or otherwise analysed through regressions and correlations of a few variables abstracted from a far more complex scenario wherein a wide range of institutions interact. It also, inevitably, means that gender will be dealt with through a process of disaggregation - either of households (using the gender of the head as the stratifier) or of individuals (differentiating males and females). The question that has not been explicitly addressed, but which is clearly central to our discussion, is: How can the focus of poverty analyses be sharpened and shifted so that the social, economic, cultural and political processes and institutions that are implicated in the creation and perpetuation of poverty become more lucid and central to the enquiry?

An important contribution to poverty analyses has come from the literature on hunger and famines, namely the "entitlements" framework as developed initially by Amartya Sen, and later in collaboration with Jean Dr (Sen, 1981; Dr and Sen, 1989). Entitlement analysis has been useful in directing attention to the processes through which individuals gain access to commodities and other resources (or fail to do so), which is said to depend on their socio-economic position and on the rules that render claims over commodities "legitimate". And in as much as these rules and norms "entitle" people differently and unequally, they draw attention to the likelihood that deprivation will be diversely constituted across a population along the lines of gender, caste, class, etc. (Kabeer, 1997).

Taking on board the important qualifications and criticisms made of this framework, especially of its excessively marketized and "legalistic" view of the rules of entitlement (Gore, 1993; de Waal, 1990), entitlements can now be seen more broadly to encompass not only state-enforced legal rules, but also socially-enforced moral rules which constrain and enable command over commodities. Thus rather than seeing famine as inevitable once certain entitlement shifts occur (such as falling wages in relation to food prices), it is now possible to include rebellions and food riots as forms of collective action that enable the poor to cope with calamity.

In E.P. Thompson's original work, food riots are identified as a form of collective action by which the poor in eighteenth century England, when threatened by exchange entitlement failure in the market-place, ensured that the moral economy of food provisioning, derived from Tudor times, took precedence over legal property rights as rules of entitlement. In that case, entitlement to food depended on acts which were legitimate, but illegal. These illegal acts also had specific rules. The riots were characterized by restraint and discipline, and a key element of them was not theft, but the setting of a just price (Gore, 1993:447)

The understanding that rules and norms are "unruly" and negotiable carries significant implications for both poverty and gender analyses, by providing some analytical space for considering individual and collective expressions of agency and contestation. As we will see below, in some contexts the structural constraints on women may be overwhelming and it may be difficult, or even dangerous, for such expressions of agency to take shape. But elsewhere gender-based norms and rules are both more contested and more malleable, and women are able to express resistance and make resource claims in meaningful ways. The rest of this section will consider gender-based access to resources, focusing in particular on land and labour, which are critical in the "asset portfolio" of the rural and urban poor in many Asian and African countries. But it will be argued that there are important differences in how men and women, in the same contexts, relate to these assets and employ them in their survival strategies.

Gender-Based Entitlements: The Intra-Household Arena

Kynch's (1998) analysis of gender and malnutrition in the north-west Indian village of Palanpur raises questions about how food entitlements (or command over food) are determined, and how men's and women's ascribed roles as producers, provisioners and reproducers affect their claims on each other, and in turn explain particular patterns of nutritional outcomes. Her contribution draws on the existing scholarship on gender-based entitlements (Sen, 1984; 1987; Agarwal, 1992), qualifying it in two significant ways.

First, as was noted earlier, the "gender reversal" in the nutritional status of adults versus their children that is thrown up by her evidence highlights the importance of having a dynamic perspective over the life cycles of men and women, and the well-being risks that men occasionally face. Second, in order to explain these contradictory outcomes, a tripartite division of food-related activities is introduced (producing, provisioning, reproducing) as opposed to the usual bipartite one (producing, reproducing). This helps clarify the complex motivations underpinning both men's and women's interests in provisioning and reproducing - altruistic concern for others, a concern for social status, as well as economic calculation and pursuit of self interest. It thereby avoids the simple dichotomy between female altruism versus male pursuit of self-interest which has been characteristic of some economistic models of the household. It also provides a better understanding of the nature of intra-household claims by drawing attention to, first, the interdependence and complementarities between the three spheres of activity (going beyond the simple "trade-offs"), and second, the ways in which moral claims and obligations can be structured by exclusive gendering of each sphere, and the detrimental impacts of this constraint on household food entitlement.

In terms of provisioning, women in this Indian community are ideologically, economically and socially excluded from the public sphere, and therefore from both some of the energy expenditure that a provisioner bears, and the reciprocity that provisioning offers. The argument of the paper is that while adult women may have been able to maintain an energy balance more easily than adult men who are burdened with energy-demanding agricultural tasks, the most devastating part of the rules of entitlement in this community is the effect on little girls who suffer severe setbacks in their growth process. Future provisioners (boys) are protected more than future reproducers (girls), even though current provisioners (men) may suffer more thinness than the current reproducers (women), probably because of the effort involved in meeting provisioning obligations. This helps to guarantee the continuity of the household, the patriline and its assets.

This argument has resonances with Sudha and Rajan's (1998) reading of the demographic trends in India. They argue that in the context of a patrilineal kinship system where marriages are arranged on principles of dowry and exogamy¹⁵ such as the one characterizing the northern plains of India, family strategies for amassing wealth, and transferring wealth intergenerationally, work through mobilizing kinship networks and manipulating the marriage of their sons and daughters in ways that pose survival threats for their female offspring. Although they provide evidence of women's collective activism and resistance against these practices, the conclusion that can be drawn from their analysis of sex ratios, and also from Kynch's (1998) village-level evidence on daughter disfavour, is that despite such forms of collective action the constraints on women at the household level remain very severe and their status is tied up with conforming to the social norms that deny them (and their daughters) agency.

¹⁵ Dowry is a transfer, in the form of money, gold and consumer goods, from the bride's parents to those of the groom. Exogamy is the practice whereby women are expected to marry into a family that lives in a different village from their natal one.

Gender-based entitlements in this context may fall in the realm of what Bourdieu refers to as "doxa" - that which is accepted as natural and self evident beyond discourse or argumentation - rather than being actively contested and negotiated (Agarwal, 1992). As Kabeer (1998) explains, in these circumstances it is not just difficult, but also dangerous, for individual women, isolated within their families, often cut off from the communities in which they grew up, to challenge the social norms that define them as lesser beings. Collective action in the public arena makes the project of social transformation an act of solidarity rather than individual self interest and is therefore likely to be effective in the long run - but the interlinkages between collective action and transformations at the individual level are complex, diffuse and unpredictable.

For Kynch (1998) too, gaining recognition as either a provisioner or reproducer can be part of individual strategy in building up bargaining strength within the household (what Kabeer refers to as "status"), and part of the argument of her paper is that over-specialization in reproducing can be a very risky strategy. Jackson and Palmer-Jones (1998) probe further the question of why gender divisions of labour take the particular forms they do, focusing on the effort-intensity of work as an important, but generally neglected, dimension. They argue that the energy-intensity of work has significant implications for gendered well-being, is central to women's and men's subjectivities, and therefore relevant to the moral claims they make on each other through processes of intra-household negotiation and bargaining.

Drawing on a number of different disciplines and sub-disciplines that address the issue of work intensity, they show that heavy work is likely to be objectively more burdensome to many women than men, in the sense that relatively small size and a female constitution can mean that the burden of effort intensive work is greater for women (i.e., it feels more burdensome), and an aversion to some forms of such labour may be expected. It may very well be that a woman, faced with such onerous and well-being threatening work, employs economic strategies (hire labour for the task), relational strategies (persuade a daughter-in-law, husband or child to do it) or technological strategies to reduce drudgery. Relationships are resources which can be deployed in this way, and ideologies of work such as discourses about strength and masculinity, or about endurance and femininity, may have real value to women in processes of negotiation and bargaining. The interesting observation made by Kynch (1998), that women in Palanpur were concerned that a metalled road could result in their being "pressured into petty trading" (p. 15, our emphasis) may indeed be indicative of what Jackson and Palmer-Jones (1998) refer to as women's aversion to work, and their strategies for building up "body capital". But even though women in this community may have been successful in reducing drudgery and thereby maintaining energy balance more easily than men, this seems to have been achieved at the cost of childhood under-nutrition, which may have harmed women in other ways.

For Kynch (1998), as well as for Jackson and Palmer-Jones (1998), the household is a central terrain where the gendered rules of entitlement (to work, rest and food) are negotiated and enforced, even though the implications of their analysis extend to other institutional arenas (labour markets in particular). One of the important contributions of Gore's (1993) critique of entitlement analysis was to highlight the fact that these moral rules are by no means limited to domestic institutions (family, household), as Sen's original formulation of "extended entitlements" would have it. "Sen's entitlement analysis", he argues, "marginalises non-governmental sites of rule-making and rule-enforcing which affect entitlement by either downplaying the role of socially enforced moral rules, or compartmentalising them to the domestic sphere. As a consequence, the interplay between the working of socially enforced moral rules and the working of state-enforced legal rules in determining a person's entitlement is ignored" (Gore, 1993:444). This is an important and useful observation not only for poverty analyses, but also for the gender analysis of poverty alleviation policies. It is arguable that some of the more interesting questions that a gender analysis of poverty would raise hinge precisely on this interplay between state-enforced legal rules and socially enforced moral rules, which extend beyond the domestic arena.

Gender-Based Entitlements: The Interplay between Institutions

In a more general sense it has been possible to use the entitlement framework to draw attention to the importance of social institutions in economic life. The entitlement framework fits neatly into what has been termed the "institution-centred" view of poverty, as opposed to the "goods-centred" and "people-centred" approaches (Gore and Figueiredo, 1997). The "institution-centred" approach highlights the importance of institutional arrangements, or the rules that provide the enabling framework within which individuals and groups make choices and go about their daily business of making a living. Social institutions can be understood very broadly to be "the rules of the game ... the humanly devised constraints that shape human interaction" (North, 1990:3), or more usefully, "the operation of tradition, custom, or legal constraint" which tends to create "durable and routinized patterns of behaviour" (Hodgson, 1988:10).

In view of the significant ways in which intra-household relations impact on women's access to resources and well-being, the domestic arena has understandably been the focus of a significant body of feminist scholarship over the past two decades. Since the mid-1980s, however, a rapidly expanding literature has come to explore the ways in which familial and gender norms are drawn upon to construct the terms on which women and men enter, and participate, in public life and in the marketplace (Kabeer, 1994:61). Property institutions, judicial systems, public provision of goods and services, and local and regional labour markets have all been usefully included under this rubric, making it no longer acceptable to see the household as the only institution where gender is present as a salient organizing principle. A similar concern is articulated by Gita Sen (1998): the need to move beyond the assumption that gender power relations at the local level are embedded in conjugal, intra-household relations alone. The structures of power that women confront at the local level, she adds, operate not only within the home, but also in the terrain of communities, local markets and institutions of local government. Walker's (1998) analysis of the South African land reform programme provides a poignant illustration of the gendered nature of these different institutions, as well as the interplay between legal and moral rules that takes place within and across them. It also highlights the difficulties involved in reaching poor women through standard poverty alleviation strategies, when these strategies are implicitly designed to meet the needs of poor male subjects.

Women and land

The South African Constitution and the intricate policy framework for land reform that has been put in place are testimony to the explicit commitment made by the government to gender equality as a long-term goal and to targeting women as a major category of beneficiaries in the short to medium term. However, what Walker's (1998) analysis also reveals are the serious gaps between policy directives and on-the-ground results, and why even a "well-meaning" government is finding it so difficult to implement gender-sensitive, redistributive policies.

Of particular significance to our present discussion are her observations about the role of the judiciary (the Land Claims Court), the institution of chieftaincy (and, in particular, the power of the chiefs to allocate land), and community-level dynamics, which present varying constraints on the process of implementation. The serious tensions between government's commitment to gender equality (reflected in both the Constitution and in the land reform provisions), on the one hand, and its reluctance to alienate neo-traditionalist structures of rural local government (i.e. the institution of chieftaincy), on the other, highlight the complex interplay between state-enforced legal rules and socially enforced - though contested - moral rules, and the contradictions and ambiguities that mark women's positions vis-is these local-level institutions. Walker also warns that what these local power structures mean is that the pursuit of tenure security could in practice end up entrenching existing inequalities in access to land by formalizing what are today informal rights and registering such rights in the name of the household heads only, "thereby fixing women's marginality in a legal trap. ... This is not policy, but the pressures propelling harried officials in this direction are strong" (Walker, 1998:12).

This is a useful observation and hints at two inter-related sets of issues. First is the argument that in social contexts marked by huge differences in educational levels and by differential access to state administration, there is much reason to fear that the adjudication/registration process will be manipulated by the te in its favour, and thereby result in new sources of tenure insecurity for less influential rightholders; both colonial and modern attempts to reform customary tenurial practices in sub-Saharan Africa, it is argued, have fallen into such traps (Platteau, 1995). Emerging evidence from South Africa suggests that those black South Africans most likely to emerge with access to land are strategically placed men with resources derived from non-agricultural sources, as well as political connections (Murray, 1996).

What this also suggests is that where women have some customary usufruct rights, as is the case in many sub-Saharan African countries, there may be fewer arguments for formalizing those rights even if tenure security is defined in very broad terms (as it is in the South African case) and even if women are explicitly targeted as beneficiaries. Second, it is also questionable whether inadequate access to land plays an important part in the creation and perpetuation of women's impoverishment in sub-Saharan Africa. For some women in sub-Saharan Africa, Lockwood and Whitehead (1998) point out, discriminatory inheritance laws and poor land access are significant constraints, but only in a minority of cases is inadequate access to land because of an inability to secure usufruct rights by itself a cause of poverty. "That land rights rarely emerged as a voiced concern of rural women in the PPAs we take as some support for this view" (Lockwood and Whitehead, 1998:27). Walker, however, notes that the cry for land in South Africa is strong, but that land is valued by rural women as a resource in gender-specific ways and that farming does not define their interest in land. Drawing on her micro-level research in one African community in KwaZulu Natal (Cornfields), Walker notes:

Overall, Cornfields women presented themselves as more interested in preserving an agricultural subsistence base than the men. They recognized land as a major household resource in supplying wood, water and thatching grass and were more interested than the men in land for residential purposes. . . . Secure access to land was also tied up with complex notions of 'a rural way of life', which encompassed more than specific kinds of livelihoods and material patterns of living, embracing in addition certain values and ideals about social relationships and manners of human interaction, such as respect for authority and for order, which are perceived to be dangerously absent in urban areas (1998:5-6).

The connections between customary rights, de jure (legal) rights and de facto control are quite complex and varied across different contexts of time and place. Kabeer's (1998) discussion of land rights in Muslim and Hindu communities of India (and Muslim countries like Bangladesh and Iran) raises a number of additional points about the significance of these different rights, bearing in mind the important qualification that gender divisions of labour in this part of the world bear little resemblance to those in sub-Saharan Africa where women's provisioning roles as well as their engagement in production for the market are socially recognized. The relevance of land rights in the South Asian context revolves more significantly around how they can be used by women to make moral claims on others, rather than be used directly for independent cultivation. In the Indian literature the critical measure of women's access to land has tended to be de facto rather than de jure entitlement; on this basis the difference between Hindu and Muslim communities is considered to be inconsequential since in both contexts women are effectively propertyless. However, as Kabeer goes on to argue, there are problems in assuming that de facto ownership can reveal all that there is to know about women's entitlements. Evidence from a number of different Muslim communities (in Bangladesh, Iran and West Bengal) shows that even though Muslim women do waive their land rights to their brothers, they see it as constituting a possible future claim (should their marriages break down), and in some contexts they do in fact press for their property rights. These are potentials which are not easily available to women in communities where such rights have not been recognized by customary law and tradition, even if they have subsequently been brought into existence by legislative action. Kabeer thereby draws attention to the importance of customary construction of rights rather than recently introduced legal ones. She also poses an important question about how changes at the legal level translate into the expansion of women's choices and agency.

The limited conclusion that may be drawn from this brief discussion of women's land rights in South Asia and sub-Saharan Africa is that cultural constructions of women's rights, or customary rights, affect women's attempts to secure a livelihood for themselves and their dependants in ways that are quite distinct across different contexts, and also different from men in the same context. One generalization that can be made is that whatever the nature of the economic system in terms of its productive and exchange relations, women's ability to function as fully acting subjects in relation to property is always less than that of men, and mediated through her relationships with men (Whitehead, 1984). While it is desirable to make women's rights to resources more formal and less conditional on relations with men (or local patriarchal institutions), it also needs to be recognized that ambiguity may have strategic advantages for women: it may arouse less resistance and yet deliver subtle forms of influence and control (Jackson, 1997b). State-enforced legal reforms with an explicit commitment to gender equality can, under some circumstances, strengthen women's position. But whether this potential is realized in practice largely depends on the local and community-level dynamics and institutions through which policy intentions work themselves out, and in particular, on the extent to which poorer women can be effectively organized to engage with and make use of the formal structures and legal opportunities that are being put in place. Moreover, in contexts where customary entitlements - themselves highly variable and changing - recognize women's ownership and/or usufruct rights, it seems important to ask questions about whether these rights need to be formalized, especially in view of the fact that in many sub-Saharan African countries an inability to secure usufruct rights is by itself probably not a significant cause of women's poverty.

Women and labour

A related point emerging from Walker's (1998) contribution with particular resonance for gender-sensitive policy making concerns the nature of constraints on subsistence agriculture, especially as it affects women farmers in South Africa. Drawing on the findings of detailed micro-level research, she argues that lack of land is not the only constraint on agricultural production, confirming some of the observations that have already been made. In addition to the "technical considerations" limiting household subsistence and commercial production (credit, extension), lack of time for labour-intensive working of the land, in particular for weeding, is a critical bottleneck. This reiterates Whitehead's (1984) earlier observation that in many parts of sub-Saharan Africa the usefulness of land rights is limited by the extent of rights in labour, and that for rural women the difficulty of commanding the labour of social superiors means that they can never mount the large exchange work parties which are so important for male farmers.

Labour issues, especially casual labour markets which embrace labour relations between and within households, have been an important theme in the rural political economy of Africa in recent years, and in gender analyses of African agriculture in particular. Their continuing significance is underlined both in Kasente's (1998) analysis of Ugandan agricultural strategies, as well as in Lockwood and Whitehead's (1998) reading of the Poverty Assessments. In the two Ugandan communities studied by Kasente, women's labour constraints are shown to be particularly acute in view of their limited capacity to hire labour (due to cash shortages), which is also leading to a significant reliance on child labour; child labour, it is argued, acts as a substitute for hired labour in women's farming enterprises.¹⁶

¹⁶ The fact that exchange labour is not mentioned in Kasente's paper may be a reflection of the demise of this institution and the growing importance of hired labour arrangements in the Ugandan countryside.

The issue of women's labour constraints in rural sub-Saharan Africa that these contributors highlight is particularly pertinent in view of the current policy consensus on poverty, which vehemently maintains that labour-intensive growth is pro-poor because labour is "the poor's most abundant asset" (World Bank, 1990). This particular policy dogma has even led some analysts to criticize, at a very generalized level, labour-saving high-yield varieties and planting and weed management technologies because, for them, "saving" labour is tantamount to "unemploying" labour (Lipton and Maxwell, 1992).¹⁷ As critics have rightly argued, dismissing labour-saving technologies in a static manner such as this is untenable, for not infrequently labour-saving varieties have been popular with farming women for reducing drudgery, and with women wage workers for stimulating growth in labour markets (Jackson, 1996). If women (and men) smallholders strive to reduce the effort-intensity of their work, then it seems unreasonable to deny them the option of doing so through the adoption of technologies such as labour-saving varieties.¹⁸

¹⁷ Even though in an earlier section of the paper it is admitted that "increased labour intensity may not be the best solution for poor women" (Lipton and Maxwell, 1992:15), the authors offer no clue as to how these two positions can be reconciled

¹⁸ In practice, however, one of the main impediments to the adoption of these technologies has been that they require complementary inputs such as timely irrigation and fertilizers that are, in many countries, beyond the reach of most smallholders.

To argue that the policy prescriptions for poverty reduction through "labour-intensive growth" are problematic, for the reasons noted above, does not mean that the opposite, equally generalized and stereotypical, view of women as "over-worked victims" suffering from "time famine", should be endorsed.¹⁹ In many contexts, such as in Iran and Bangladesh, female underemployment is a serious social issue - not just among the affluent classes but also among some of the poorer urban and rural social strata where poverty is very often associated with "enforced unemployment, particularly among women" (Kabeer, 1998; Razavi, 1997). At a deeper level, what is really at issue is not so much whether women need more work or not, but rather the kinds of work that can raise them and their dependants out of situations of poverty, which in turn draws attention to the content and quality of work. This is how we can make sense of the apparent paradox in Walker's (1998) paper, where she suggests that rural women are overworked, and yet at the same time they voice a clear and explicit demand for more urban jobs which presumably involve less drudgery and offer higher returns to effort.

¹⁹ Some of the contextualized literature on the impacts of structural adjustment policies on women highlights the increasing demands on women's time, as a way of coping with cut-backs in male wages and state provision of services (see Gonzalez de la Rocha, 1998 and references in her paper). But it would be problematic to use these observations as the basis for a generalized and universal description of women's work.

The linkages between poverty and the labour process, which have been the source of long-standing tensions between neo-classical and critical institutionalist approaches, are explored at some length by Razavi (1998), who argues that, as far as labour market issues are concerned, the New Poverty Agenda offers very little that is "new". Both trade liberalization and labour market deregulation, which are the hallmarks of this agenda, have featured prominently as policy conditionalities under structural adjustment and stabilization programmes, and their gender implications remain deeply contested. The fact that this agenda remains wedded to an abstract theory of labour markets that overlooks the insidious ways in which power hierarchies pervade both "formal" and "informal" labour markets inevitably means that it cannot problematize the ways labour market arrangements themselves can perpetuate poverty and discrimination. On the other hand, while power hierarchies (between labour and capital) are central to institutional approaches, the failure to analyse adequately the interplay between class (labour/capital) and gender (male/female) hierarchies across different institutional arenas (labour market, conjugal/familial sphere) means that the distinctiveness of women's experience of work is sometimes missed.

The second, and related, point to note is that while labour constraints are clearly an important issue for some women, there are serious limitations in the way these labour constraints have been conceptualized for policy purposes: first, by ignoring the contextual and institutional parameters of work; second, by de-linking work and well-being; and third, by failing to make adequate linkages between the micro-level gender divisions of labour and the macro-level economic and social processes.

One well-known strand of thinking, which has become near-orthodoxy among development practitioners, comprises the "triple roles" or "multiple roles" frameworks (Moser, 1989; Commonwealth Secretariat, 1989). Arguably these frameworks have been very useful in drawing attention to the multiple demands on women's time, particularly pertinent in the context of drastic cut-backs in social spending and falling real wages - as during the 1980s economic crisis in many parts of Latin America (Gonzalez de la Rocha, 1998). Nevertheless, the analytical emphasis on what women do ("roles") leaves many important issues unaddressed. It says very little, for example, about the kinds of social relationships and institutional contexts within which women carry out their work, which are arguably far more important than the quantity of time allocated to different activities (Kabeer, 1992). Kasente's (1998) observation that women smallholders rely on child labour, while their husbands are more successful in commanding hired labour, is indicative of the kinds of constraints that women farmers face because of their social position in the household and domestic group, even though in terms of labour inputs men and women in this particular setting seem to be making more or less equivalent contributions. Analysing female poverty within a static framework of multiple roles can also have the added disadvantage of being divorced from the complex processes of economic and social transformation that are underway in a particular context; it becomes difficult, therefore, to decipher the implications for gender inequalities of, for example, increasing rural differentiation driven by agricultural growth (Evans, 1994).

Another way of conceptualizing women's time constraints is the "gender and adjustment" model, which makes explicit linkages with the current macro-policy framework in sub-Saharan Africa (Collier, 1989; Palmer, 1991). Using elements of neo-classical economics, it combines the disaggregation of agents by gender with a sectoral disaggregation of activities to show that gender can act as a serious constraint on the mobility of resources (especially labour) between different sectors, thereby frustrating the aims of structural adjustment. It makes the policy exhortation that women's "reproductive burden" be reduced (through public investments in electricity and piped water, etc.) in order to enable them to respond more easily to price incentives, switching their labour from the production of non-tradables and protected tradables, to tradables (the sector that is set to expand under adjustment). Although it would be difficult to dispute the need for social investments to reduce women's reproductive "burden", these particular arguments present some serious analytical and policy blindspots which need to be approached with caution (Elson, 1995; Lockwood, 1992). Two problems in particular are relevant to our discussion.

First, according to this model there are no justifications for reducing women's work burdens, through labour-saving domestic or agricultural technologies, if they use their free time for other purposes, such as rest or even leisure. This style of thinking is not new. The need for domestic water provision, for example, has very often been justified in terms which suggest that women who spend less time on water collection will spend more time on farming. There is a somewhat similar logic behind self-targeting through the labour test, which effectively screens out the less poor and offers effort-intensive work to those ill-equipped to bear the bodily costs of energy-intensive work (the women, and the poor). However, as Jackson and Palmer-Jones (1998) forcefully argue, this kind of policy obsession with extracting work from the poor, which is being validated through the New Poverty Agenda, needs to be viewed with concern since it may not provide much of an escape from poverty; in some contexts and for some groups it may in fact generate "energy traps", undermining their bodily well-being. If women in particular seek to sustain bodily well-being through avoiding heavy work, then poverty-alleviation policies need to consider carefully how they might impact on women's ability to do so. At a more general level, this highlights the tension we have already noted between capability-enhancing employment versus effort-intensive work, or drudgery, that generates poverty through the labour process - a tension that was succinctly captured in the concept of "the working poor".

The other problem with the model is that it fails to problematize the wider policy framework, which is assumed to be inherently benign. It would be quite short-sighted, for example, if this kind of analysis were taken as a suggestion that women's labour bottlenecks should be removed so that women could switch their resources to the production of tradables, without asking further questions about what kinds of market opportunities would be opened up to them, or how this intensified market integration would impact on household food security. In Collier's model, for example, there is no mention of the food security risks (at the household and national levels) that may be involved in relying on export crop production, nor is there any recognition of the long-term price risks involved if agricultural export strategies are pursued in many countries simultaneously - the fatal "fallacy of composition" that characterized agricultural policy prescriptions under traditional SAPs and continues to do so under the New Poverty Agenda.

Kasente (1998) raises some important questions about the tensions between the current policy emphasis in Uganda on "non-traditional export crops" (NTEC) while food security considerations are a priority in women's (and to a lesser extent men's) farming strategies. Fifteen years of political turmoil and institutionalized forms of predation and expropriation have underlined food security concerns and the importance of self provisioning in the Ugandan countryside. This is not to suggest that cash crops and food crops are mutually exclusive categories; in Uganda, beans and maize can be used for both purposes. Rather the question is about how the commercialization of farming will change the ways in which food is secured (from self provisioning to greater reliance on markets, etc.), the possible risks and vulnerabilities that these changes may entail, and the degree to which women's work burdens, well-being and autonomous spheres of activity will be affected.

The issue of food security - and vulnerability - is central here not only because of domestic "market imperfections" exacerbated by minimalist agricultural policies (monopolistic buyers replacing marketing boards, increasing food prices, rising costs of inputs, transport bottlenecks, etc.), but also in view of the fact that both international commodity prices and international food prices are highly volatile (South Centre, 1997). The collapse of international coffee prices in the late 1980s - which put many African economies, including Uganda's, under severe stress - is a recent and vivid reminder of the fact that global markets, in addition to providing the much-emphasized opportunities and outlets, are also capable of generating considerable risks and devastating shocks. It is not at all self-evident that the so-called "non-traditional" export crops (e.g. maize, beans, vanilla) would not be subject to the same risks that characterize the more "traditional" export crops (e.g. coffee, cotton).²⁰ The interesting point emerging from Kasente's (1998) contribution is that women (and to a lesser extent men) farmers are not, in fact, swiftly switching their resources into the production of NTECs such as vanilla, even though they offer an attractive price. And according to Kasente's reading of the situation (based on qualitative evidence), the reasons for this "perverse" supply response on the part of smallholders include concern for household food security (which is particularly evident in the responses of women farmers), as well as women's greater control over the marketing and the revenues from the sale of food crops (compared to pure cash crops like coffee and vanilla). The reasons why female labour is "stuck" in the production of non-tradables in these villages then may be quite different from those identified in Collier's model.

²⁰ As Gabriele Kr (UNCTAD) pointed out, the label "non-traditional" is quite loaded, given that these are fairly traditional international commodities, as well as having been traded across borders in this part of Africa for some time now (personal communication).

The other question that needs to be raised, which takes us beyond the issue of labour constraints in agriculture per se, concerns the strategic importance of farming as a poverty-alleviating mechanism. Here again the papers commissioned by UNRISD provide some interesting clues that need to be probed further. The repressive migrant labour system and the gradual erosion of black peasant production through a combination of factors flowing from state policy, which have been the hallmarks of South Africa's transition from an agrarian to an industrial economy in the twentieth century, have made lasting imprints on rural livelihood strategies. Rural-urban migration has become a way of life and agricultural production, whether for subsistence or for the market, has become a marginal activity. "Agricultural production is, for good reason, seen as a high risk activity ... and most rural women (like most rural men) look to urban jobs as the route to household economic survival and advancement" (Walker, 1998:3). There are serious doubts about whether agriculture can generate the route out of poverty and destitution in the absence of substantial and sustained developmental support from the government, which in the current policy climate seems "difficult" to achieve.

While South Africa is in many ways unique, Walker's (1998) observations about the limitations of agriculture are nevertheless relevant to many other contexts as well. Lockwood and Whitehead's (1998) reading of the survey evidence produced in the Poverty Assessments on Tanzania, Zambia, Uganda and Ghana is that off-farm activities, such as trading, transporting and state employment, have important poverty-alleviating repercussions.²¹ But these are not explored in the policy discussions of the PAs, which simply emphasize raising agricultural productivity through better technology, research and extension ("modernization" of agriculture), and the need for diversification into non-traditional export crops. The preoccupation with agriculture as the sector that can produce labour-intensive growth in sub-Saharan Africa, they suggest, produces a lack of analysis and policy interest in how off-farm activities might become a more important source of income for the poor, or what policies might help an increase in such activities.

²¹ Jayati Ghosh makes a similar point on the basis of Indian survey data for the period 1977-1994. She also makes the important observation that the growth of the off-farm sector was not in any way a spill-over from the agricultural sector (see Ghosh, 1998).

This is not meant to endorse the global policy prescriptions for the agricultural sector, which condemn in advance any attempt by the public sector to intervene in agricultural markets through protective tariffs, quotas and subsidies, as well as transfers of wealth (Barraclough, 1998; South Centre, 1997). The rising input costs and credit shortages noted with concern by some of these authors are testimony to the harmful effects of these policies. Nor is it meant to suggest that the off-farm sector, or the urban labour market, can be the panacea for poor women. Walker (1998), for example, explicitly notes that the industrial sector is highly stratified, not only by race but also by gender, making it difficult for rural women to access urban and industrial jobs - a theme that is also explored at some depth by Gonzalez de la Rocha (1998) and Razavi (1998). Rather Walker's analysis shows the importance for the rural poor of "straddling" different economic options when the returns to farming are so pitiful. This in turn raises two further questions: first, what (if any) are the relations between these different spheres; and second, what is the significance of the gendering of each sphere for issues of subordination and gender equality?

Walker (1998) and Kynch (1998) highlight the interdependence and complementarities between "producing", "provisioning" and "reproducing" in rural South Africa and in rural Uttar Pradesh, respectively. In the urban context, not surprisingly, monetary returns from market-based production take on an even more strategic significance as is clearly reflected in Gonzalez de la Rocha's (1998) analysis based on evidence from urban Mexico. She argues that without incomes coming from wages and salaries, the urban poor's ability to turn to self-provisioning is eroded, leading to a vicious process of cumulative disadvantage. Even support from social networks, she adds, requires investments of time, effort and money, since networks are social constructions that need to be established and maintained; they are not simply "out there" for people to access. When people are so poor that these resources are not available, they no longer form part of a social network and social isolation accompanies labour exclusion. The point emerging very strongly from her contribution is that self-provisioning is not an "autonomous" sphere and it would be irresponsible on the part of decision makers to assume that it can become a viable safety net for the urban poor. Although less obvious, it would be equally misleading to think that self-provisioning is an "autonomous" sphere in those parts of rural sub-Saharan Africa where smallholders predominate. Indicative of their lack of "autonomy" is Lockwood and Whitehead's (1998) observation that food security considerations very often push poor farmers into casual labouring on a daily basis, which effectively diverts their labour from their own production and becomes an important part of the dynamic of impoverishment.

While there is consensus among the authors of the UNRISD commissioned papers on the complementarities between different spheres, there is less agreement as to how the gendering of different spheres impacts on gender relations (i.e. male/female power dynamics and issues of subordination).²² We may agree with Kynch (1998) and Kabeer (1998) that specialization in reproductive work can be very risky for women and young girls because of the low value/status that it carries, based on the evidence that they and Sudha and Rajan (1998) produce on daughter disfavour in north-west India. But the social significance and implications of women taking on a more independent "bread-winning" role are clearly context specific, as well as being open to different interpretations even within the same context.

²² The papers do not, however, explore the nature of the linkages (if any) between agricultural and off-farm activities. See Harriss, 1987 for an analysis of this question in Tamil Nadu (India) and Hart, 1996 for South Africa; both provide a devastating critique of the so-called regional growth linkages whereby the small farm sector is supposed to create the conditions for its own existence by stimulating non-farm jobs within rural regions. A similar conclusion emerges from Ghosh's (1998) analysis of Indian macro-data.

For example, while in South Africa women have not been able to tap into the urban wage labour market and remain predominantly confined to self-provisioning and reproduction, this does not mean that patriarchal relations in rural areas remain "frozen in a rigid mould of 'tradition'" (Walker, 1998:6). On the contrary, Walker argues, gender relations in the rural areas are in considerable flux, with an overall decline in patriarchal authority (also evident in the relationship between patriarchs and the youth) and a concomitant opening-up of social space for contestation between men and women about the terms of male authority and the meaning of "tradition".

Gonzalez de la Rocha's (1998) interpretation of women's entry into wage employment in urban Mexico raises questions about whether women's rapid entry into the wage labour market under conditions of duress, characterized by rising male unemployment, falling real wages and cut-backs in social spending, has been empowering in a feminist sense. A similar point is reiterated by Razavi (1998), who argues that the closing of the gender gap in wages ideally needs to be accomplished in the context of rising, or at least non-falling, wages for men. What these interpretations highlight is the extent to which conflicts of interest between men and women are unlike other conflicts, such as class conflicts, because of the important dimension of "togetherness" that marks gender relations. In a context where the livelihoods of significant numbers of women include a legitimate reliance on male wages, which compensates them for reproductive labour, male unemployment and drastic cut-backs in male wages have enormous significance for women's standard of living.

But the micro-level studies to which Razavi (1998) refers also highlight the possibilities and the "spaces" that entry into the urban labour market has opened up for some groups of women in some contexts, allowing them to re-negotiate the terms of their domestic relationships, and in some cases to walk out of, or not enter into, unsatisfactory relationships. This is not to deny the fact that the increased field of manoeuvring at home may be matched by different patriarchal controls in the factory setting - which, as one study of women's industrial work in Java argued, keeps the Javanese factory daughters "relatively acquiescent, poorly paid and vastly unprotected in industrial jobs that are often dangerous" (Wolf, 1992 cited in Razavi, 1998). This highlights, once again, the need to move beyond the assumption that gender power relations at the local level are embedded in conjugal, intra-household relations alone, and the importance of having an "interactive" view of the dynamics among factory women, factory employment and workers' family and household relations.

The debates on women's entry into wage employment have raised many difficult methodological questions, such as the tension between "objective" criteria (skills, wages, health issues and bodily well-being more generally) and "subjective" criteria (women's perceptions of their work), for assessing the implications of this form of work for both poverty eradication and wider issues of discrimination and subordination (i.e., contestation and negotiation within the home and in the world of work). While reference to some objective criteria of well-being is clearly needed in order to get us away from the utilitarian insistence in taking subjective preferences as the only criteria for making judgements about values and welfare, most of the authors of the UNRISD commissioned papers would also agree that there is a need for women's own perceptions and values to find some space in these discussions if only because they allow us to better understand the "choices" that women make. As the authors of in-depth case studies have pointed out, one of the main weaknesses of structuralist analyses of women's entry into industrial production has been that they have frequently rendered women "faceless and voiceless" (Wolf, 1992:9) by attributing much more personality and animation to capital than to the women it exploits (Ong, 1988:84).

Women's entry into these export-oriented industrial sectors also means that the current tensions between Northern and Southern, and local/national and global perspectives - which are explored at some length in Razavi's (1998) analysis of the debates around the "social clause" - have explicit and direct gender implications, even if the views of women workers are often missing from these deliberations. The fact that feminist analyses of factory work constantly move beyond poverty issues into the wider domain of power and subordination is testimony to the fact that gender issues in development cannot be reduced or "collapsed" into a welfare agenda, and also a reminder of the fact that for women workers, unlike most men workers, the process of becoming a worker and earning a monetary wage can be an empowering one by triggering off a sense of independence and self-worth, as well as facilitating new forms of solidarity (with co-workers) that are independent of kinship and residence.