Pula: Botswana Journal of African Studies. Vol.14 No.2 (2000) The complexities of female household headship in Botswana Godisang Mookodi University of Botswana, Sociology The concept of female-headed household has emerged as an important analytical category in the examination of poverty in life chances both within the context of policy research and social science scholarship. This paper presents the complexities that arise in the use of the concept offemale-headed household within the context of Botswana. The paper presents some of the criticisms that the concept of headship in itself presents a monolithic and often limited notion of social organisation that fails to take into account complex gendered social interactions that occur within and outside the confines of domestic units. The results of a study conducted in 1996 reveal the complex interplay of cultural-structure and individual agency that are obscured by discrete notions of 'headship '. Introduction The examination of the characteristics of poverty and life chances, focusing on the gender of the head of the household, has been the subject of increasing academic research and debate in recent years in developing countries. Census and household surveys conducted in Botswana reveal that almost half of all households in the country are headed by women, and that a significant proportion of them fall in the lowest income categories. This article reviews and contributes to the debate on household headship in Botswana. The evidence used in the article is from a study that I carried out in 1996 in Botswana. The study examined the implications of household organisation and gender relations of economic production and social reproduction on the life chances of women and their dependants. Research was conducted within a pool of low-income female and male headed/supported households in Manyana, a rural village and Gaborone, comparing similarities and differences in their composition, sources of income and survival strategies employed by women and men within them. The interviews with women and men pointed to the complexity of domestic organisation and the significance of gender hierarchies that are often obscured by focusing on discrete notions of 'headship'. Based on the evidence from this empirical study, as well as that presented by other studies, it will be argued that the utility of the concept within the context of Botswana is diminished by a lack of in-depth account of the culturally-based gendered social relations that shape identity and life chances. Households and ma/wapal Household surveys and other studies of household organisation in Botswana commonly use the Setswana lolwapa as a translation of the word 'household'. In recent years, several social researchers (Driell994; Motts 1994; Gulbrandsen 1996; Garey and Townsend 1996, Townse?d 1997) have pointed to discrepancies in the meaning of the two terms. Most eco~omlc and demo~phic definitions of the household refer to [a person or] persons s~g a commo~ reSidence and resources-especially meals. In contrast, there are two different connotations of the term lolwapa in Tswana language and culture. Firstly, the term 11be p111l'11 oCtile SetswIna word loIwapa. 148 is used to describe a compound or physical yard. Compounds and yards may contain one or several physical dwellings. The second use of the term relates to family membership. This tends to complicate the boundaries of the physical household, as individuals identify members of their /o/wapa as family members who may be present or absent from the residential unit (Gulbrandsen 1996; Townsend 1997). How have researchers produced working definitions of households within the Botswana context? Household survey defmitions of 'the household' focus on physical dwellings occupied by one economic unit described as persons living together, sharing eating and sleeping arrangements within a fifteen day time frame. Garey and Townsend (1996), and Gulbrandsen's (1996) defmitions transcend residential boundaries in order to capture the social and economic interactions between individuals who considered the physical /o/wapa as their home. In contrast, Driel's (1994) working defmition is confined to the resident members of the physical compound. While acknowledging the methodological problems associated with the definition of households, I broadly defined households as individuals who usually share common physical residence and eat together regularly regardless of their family relationship. The term /o/wapa implies different connotations of physical residence and family membership. It became evident that the term t/hogo ya /o/wapa was also a loaded concept within the context of Setswana language and culture. Definitions of female headship Much household research has focused on defmitions of headship that revolve around identifying one individual as the main economic provider and overall decision maker. Researchers questioning these notions of headship have suggested that decision-making is a complex process that involves the participation of various actors in accordance with the different roles and positions within household hierarchies. The review of studies at international level and within the context of Botswana reveals the complexity of intra- household decision making that is a product of socio-cultural and economic processes as well as individual agency. The literature on female-headed households in developing countries illustrates the ongoing struggle to define female headship. Scholars have encountered difficulties in reconciling a) defmitions of headship adopted by policy-oriented national surveys; b) the criteria used by individuals in their identification of household heads, and c) the lived experiences of women and men with regard to resource provision, household management and decision-making within different cultural contexts. Sandra Rosenhouse (1989) and Nancy Folbre (1991) posit that while the use of the head,ship concept in censuses and household surveys partly serves to identify a reference person to whom to relate the various household members, the application of the concept is loaded with implicit assumptions about decision-making processes and resource provision within households. The identification of one household head, whether female or male, is based on unitary models of household organisation which assume that one person is the primary income-earner and decision-maker. This approach under-emphasises the complexity of economic provision under conditions where there are multiple earners and decision-making within households, and socio-economic co-operation between individuals, especially those with limited economic resources. Another key criticism raised by feminist scholars refers to the asymmetry of headship, defining households as female-headed only if no adult male is present, while defining a household as male headed whether or not an adult female is resident (Rosenhouse 1989; Folbre 1991; Kennedy and Peters 1992; Moser 1993; Kabeer 1994). While WID scholarship does not challenge these conceptualisations due to their emphasis on obtaining comparative data and lobbying for women's visibility within 10cal policy-making, some 149 feminist scholars have struggled to refine concepts of female headship in order to provide more accurate descriptions of gender differentiation within domestic units. A number of researchers have analysed the results of surveys conducted at the household level in Asian and Latin American to examine patterns of resource provision, allocation and decision-making to decompose the headship concept. Sandra Rosenhouse's primary concern is with the ambiguity of the concept in national surveys. Rosenhouse analysed the fmdings of an LSMS2 sample survey conducted among about 5110 households in Peru between 1985 and 1986 in order to gain insight into patterns of decision-making, economic support against reported headship. She indicates that when respondents were asked whom they recognised as the head of their households, 83 percent declared males while 17 percent declared females (Rosenhouse 1989: 16). The main criterion for the identification of male headship was marriage (currently in union), while identification of female headship was largely based on widowhood, divorce and separation. Are individuals who are identified as household heads always the principal income earners in their households? Rosenhouse's analysis of data on the employment status of household heads revealed that 7 percent ofall reported heads (male and female) were not in the labour force, and had not been in the twelve months preceding the survey. The findings also showed that within a third (32%) of the households, the head did not bear the primary responsibility for the economic maintenance of the household. The proportion of female heads who were not principal income earners was 44 percent compared to 29 percent of male heads, indicating that a large proportion of households depended on income from other household members (ibid: 23). Data on economic contribution was based on the total hours of income-generating 'market' work of all household members (including goods produced at home excluding housework) contributed by each individual. This criterion was used as a basis for the definition of 'the working head', or that individual who was identified as the major income contributor. A comparison of the two criteria of headship (Le. reported head and working head) resulted in a marked change in the proportions of headship by gender. Based on the economic criteria, the proportion of male heads declined from 83 to 71 percent, while that of female heads increased from 17 to 29 percent (Rosenhouse 1989: 29). A comparison of the working head's relationship to the reported head revealed that children and relatives accounted for 14% of all working heads. While Rosenhouse admits that the 'hours of market work' criterion is limited as a proxy for actual fmancial contribution, she uses it effectively to illustrate the complexity of ~conomic support within households. She also spells out the implications of multiple mcome contribution to decision making within households: The ex~stenceof more than one primary earner within the household is likely to introduce changes m the ~';Yer relations within the household, and therefore in patterns of intra- household ne.g~tlation over resource allocation and control. While it is likely that gender affects bargammg power to some extent, economic contribution may override its effects (Rosenhouse 1989: 41). While .v~ous ~ousehold members (female spouses/partners, children and relatives) may not readily Identify themselves as household heads or be identified as such within the context of ~B!ional censuses and household surv~ys, their contribution of economic re~0';U'ceswlthm households often has direct implications for the allocation of resources W1thmhO~ho~ds. ~ose~ouse's discussion illustrated the limitations of reported headship for the reliable IdentificatIOnof the economic support base of the household. :;:v~tandards Measurement S~cys ~ house~O!d surv~.thllt have been used since the 1980~ to measU1C education, ~~ They. gather mformation on hvmg conditions - measures of inc:omc, expendIture, health, ~l""l'_J" agnculturc, and IC:CCSs to services and ownership of 1SSClS. ISO within households. Rosenhouse's discussion illustrated the limitations of reported headship for the reliable identification of the economic support base of the household. Women who identify themselves as household heads due to the temporary or permanent absence of husbands or cohabiting partners may have limited authority due to their reliance on non-resident male kin for representation in matters relating to their well-being as well as that of their children. Case studies conducted in rural Egypt (Saunders and Mehanna 1993) and rural Bangladesh (Islam 1993) illustrate that while widows identified themselves as the heads of their households, they were reliant on senior male kin representation in public matters, especially those relating to the acquisition of land for agric.ultural production. The forgoing discussion has illustrated some of the difficulties associated with producing universally applicable defmitions of female headship. The studies show that the identified household head is not necessarily the primary income earner, and that often, when other individuals are the primary income earners, they are subjected to the authority of male heads who do not contribute to the daily welfare of household members. The debate on household headship illustrates the need to move beyond 'headship' to examine patterns of resource acquisition and allocation within households. Single mothers and female-headed households Data from national censuses and family health surveys conducted over the last twenty-five years point to declines in marriage, and the growing significance of single motherhood. Several studies (Molokomme 1991; Drie11994; WLSA 1992, 1994) point to the increasing acceptance of pre-marital pregnancy in Botswana. The social sanctions encouraging women to bear children are just as strong as the previous sanctions to marry, with the consequence that women who are barren (whether married or not) are often shunned by their peers and the community. The Botswana Family Health Surveys of 1984 and 1988 showed that the mean age at first birth was between 18 and 20 years. Kann and Mugabe's survey on teenage pregnancies in 1984 showed that approximately half of all nineteen year-old women were mothers (Kann and Mugabe 1988). While most of the studies on household organisation in Botswana concur on the effects of socio-economic change on family forms, social scientists differ in their conceptualisations of female-headed households. Several studies (Izzard 1979, 1985; Kossoudji and Mueller 1983; Driel 1994; Motts 1994) aimed at establishing the prevalence of separate female- headed households on the basis of women's marital status and on the high incidence of extra-marital births. Others (Peters 1983; Molokomme 1991; Townsend 1997) point to the limitations of discrete conceptualisations of female headship due the significance of fluid consensual relationships during the life courses of women and men, as well as the assumption of decision-making roles by non-resident males which limit women's autonomy. Typologies of female headship During the 1980's, a number of scholars (Izzard 1982; Brown 1983; Kossoudji and Mueller 1983; Kerven 1982) pointed to the prevalence of female-headed households in rural Botswana. These scholars developed typologies of headship based on the marital status of women, the ages, as well as temporary and permanent male absence utilising data from various government surveys. The National Migration Study (NMS) was conducted in the late 1970's to document patterns of internal migration between rural areas, and between rural and urban areas. Recognising that migration resulted in the fluidity of household membership over time, the NMS adopted the definitions of household membership: 151 de jure household membership referred to those individuals who were regarded as 'usual' household members; . de facto household membership referred to those individuals who were present at the tune of the study. These classifications were utilised for the definition of household headship. De jure female household beads were defined as those individuals who were identified as the 'usual' or 'pennanent' heads of their households, while de facto female household heads were identified as women who assumed temporary headship in the absence of de jure male household beads. Based on these classifications, twenty six percent of all households in Botswana were permanently headed by women in 1979, while eleven percent were temporarily headed by women (Izzard 1982). The ages and marital status of women were also used to distinguish between the different typologies. In her analysis of NMS data from Gaborone and three rural areas, Izzard (1982) indicated that the majority of the de jure female household beads were over the age of 45 years, and had been previously married. In the national sample however, most of the de jure female household heads were women who had never been married. Kossoudji and Mueller (1983) based their discussion on female headship on data from the Rural Income Distribution Survey (RIDS) which was conducted from 1974 to 1975 by the Central Statistics Office among a representative sample of 1060 households in various rural areas of Botswana. The primary objective of the RIDS was to collect data on income distribution based on detailed accounts of aggregate amounts and sources of household income as well as asset ownership over a twelve month period. Household members were asked to identity the heads of their households among the people who were regular members according to their own perceptions (Kossoudji and Mueller 1983: 836). Fifty five per cent of the households identified men as the heads of their households while forty three per cent designated women. Two percent of the households did not identity the heads of their households. While they did not provide the actual proportions of dejure and de facto female headship by marital status within the RIDS sample, Kossoudji and Mueller pointed to the declining significance of marriage and the prevalence of single motherhood based on data from the 1971 census. They showed that while 56 per cent of women within the 20-24 year age group and 29 per cent of women in the 25-39 year age group were still single, and that a large proportion of women between the ages of 20 years and 39 years had children (Kossoudji and Mueller 1983: 835). They posit that in light of the declining significance of marri~e and the prevalence of single motherhood, these individuals constituted the potential pool of pennanent female household heads. Pauline Peters (1983) initiated a debate on the definition of female headship and headship typologies within policy research in Botswana. She argued that the policy-oriented lite~ture on female-headed households (e.g. Izzard 1979, Kossoudji and Mueller 1983) denv~ from a over-relian~ on typologies which obscured the processes and contexts that detennmed gender and soclo-economic differentiation over time: We cannot analyse the changing sociaL economic or political position of women by reference only to the characteristics of the household head. And many of the dynamic p~ that are crystallised in particular household forms or particular configurations of disadvantage cannot be grasped by taking households as primary units of analysis. To do so leads often to erroneous conclusions because critically important relations within ho~olds and betw~enthem are not taken sufficiently into account (peters 1983: 105). While ackno~ledging the importance of identifying the manager of a domestic unit, she argues that the mtegration of the individual bead and the members of the household into 152 one unit of analysis obscures the complex social and economic interactions within which domestic units, and individuals within them, are embedded. Peters points to the limitations of utilising marital status as a central criterion for defining female headship within the context of Botswana, based on anthropological accounts (Schapera 1933; Comaroff & Roberts 1981) on the lengthy process of marriage and the fluidity of consensual unions in Botswana. Based on this evidence, she argued that female headship may be ephemeral, rather than a permanent phenomenon within women's life cycles; that many women who may be identified as de jure or permanent household heads could later be incorporated into male-headed households through marriage. She also alludes to the fact that the discrete headship typologies largely ignore the social and economic relations that occur between women within households with men who may not necessarily reside within them. Athaliah Molokomme (1991) also points to the limitations of female-headed household concept within Tswana culture, particularly to the significance of non-resident males: A household may appear defacto to be headed by an unmarried adult woman in the sense that there is no adult male who resides in it and exercises decision-making functions on a daily basis. At the same time, some man somewhere such as a father, brother or uncle may de jure head it, in the sense that only he may legally make certain important decisions, such as the capacity to litigate or represent the household in other traditional activities reserved for men (Molokomme 1991: 59). Molokomme's objection to the labelling of such households as female-headed is based on the implication that headship suggests an embodiment of power, which many women do not wield under cultural circumstances. She therefore recommended a cautious application of the term by analysing headship and single motherhood separately. While Oriel (1994) and Motts (1994) acknowledge the complexity ofsocial networks and cultural influences on household organisation, they point to the prominence of separate independent single mother (unmarried, widowed, separated and divorced) domestic units as permanent household types in Botswana. The prominence of matrifocal female-headed households was a central theme of in-depth 3 case studies conducted by Oriel (1994) in paje and Motts (1994) in Kang\ which examined the socio-economic welfare of unmarried mothers and female-headed households within the context of social change. Both of the studies define female headship on the basis of the decision-making and economic management roles that are fulfilled by single mothers in their independent households. These researchers challenge Peters' assertions of the temporality of female headship and the strength of agnatic linkages, arguing that, in the context of change, a large proportion of single mothers maintain independent households on their own, and are the key providers and decision-makers within these domestic domains with little or no support from extended family and fathers of children. Oriel posits that as subsistence agriculture became increasingly subordinated and weakened by the cash economy, a large proportion of wage earning males became less reliant on family-based production which was based on women's and children's labour. Within that context, marriage lost its significance as the economically based regulator of production and procreation within the society. These socio-economic changes had an impact on women's and men's gender ideologies: The relation between gender ideology, gender relations and socio-economic change have had their mutually reinforcing effects. As a result, women seemed to become economically 3A village in the Central District. 4A village in Kwencng District in the South Wcstcm part of the country bordering the Kgalagadi Desert. 153 more dependent on men, whereas men could avoid their responsibilities to support women and children. (Oriel 1994: 212) Oriel argues that within the context of economic change, the ideological significance of motherhood and fatherhood has been altered. While motherhood continues to be a means of gaining status or socio-economic security for women, the significance of fatherhood lies more on 'reputation and virility' rather than in the social importance of fatherhood (Oriel 1994:211). Oriel's definition of household focuses more on the physical homestead (/olwapa) as her unit of analysis rather than applying the concept of household as a unit of consumption and production. Her defmition of headship is based on marital status and 'economic management'. She utilises the de jure and de facto female headship typologies and applies them according to the views and experiences of the fifty senior adult women interviewed during the study. Oriel's findings illustrate the complexity surrounding the use of the concept of household head, as reflected in this table depicting the different categories of headship. Oriel's Definitions of Headship (Paje 1990 -1992) HOUSEHOLD HEAD MARITAL STATUS Self Other Husband Other Total Female Male Married 1 1 19 1 Widow 10 1 1 Sinale 15 1 2 8 Total 62 2 19 3 0 Soume:Drie/1994:150 Oriel indicates that over half of the women interviewed identified themselves as the major decision makers in all economic, educational and social matters relating to their households, and were recognised as such by other household members (Oriel 1994: 149). She classifies these households as de jure headed by women with one exception, the m~ed w?man who was de facto head due to the temporary absence of her husband. The two mtervlewees who referred to 'other women' as the heads of their households were a young unwed mother who established her household next to her mother's but consulted her mother on all 'major de<;isions's,and a married woman who was residing with her husband and children in her mother's homestead who regarded her mother as the head of the household. Three women who lived in 'independent' homesteads referred to non-resident bro~~rs :m~ ~ale kin. as th~ heads of their households as they consulted them on 'maj~r deCISions (I.bld). While Onel acknowledges the incidence of cohabitation, her analYSIS focuses on discrete marital categories. Mott~' study ~1994)among thirty-nine households in Kang also points to the significance of .th~ Isolate~mdependent female-headed household type. Using similar arguments (~o Onel s). regardmg ~e changes in female and male ideologies in relation to economiC ~rod~ctlOnand SOCialreproduction and parenting over time, he developed typologies that Identified ~e female respondents' respective stages in the family life. His six typologies were: .(1). smgle, (~) sing~eand widowed, (3) single and divorced, (4) single and fractured cohab~tatlo?,(5) ~mgleWithnyats; (boyfriend), and (6) cohabiting (Motts 1994: 96). He Identifies.nme ?f the households as nuclear and isolated-lacking economic support from other family units; fourteen were matrifocal or were embedded in multi-generational female-headed households; five had direct social linkages to male headed households sDriel does not aive a 1 . d' .. ... c ear m lcation of tile nature of these 'major dec:isions' in her discUSSion. 154 through birth while maintaining separate lives, while nine enjoyed strong ties with the families (Motts 1994: 91). Both Driel and Motts alluded to the fact that the female heads of households were responsible for making decisions regarding the welfare of their families without male participation. The limitations of both studies stem from their lack of inclusion of male respondents, as well as the lack of comparison of the socio..-economic organisation of male- headed households within the two communities. The ambiguity of HIES classification The Household Income and Expenditure Surveys of 1984/85 and 1993/94 disaggregated household data by the gender of the head. While the data from the latter survey were used in the analysis of poverty by the Botswana Institute of Development Policy Analysis (1996), the HIES 2 did not provide a precise definition of headship. The HIES 2 utilises the de facto approach to household membership, due to the fact that population mobility would create problems in the operationalisation of the concept of 'usual .member', as it would entail detailing exact or appropriate proportions of time spent at each of the places where members reside (Government of Botswana: 1995). There is no clear indication of the basis ofconceptualisation of the 'household head'. The following background information was obtained from several sources including: a paper on methodology that was presented by a CSO official at a seminar held to disseminate Census findings (Government of Botswana 1995), a report on living conditions in Botswana from 1986 to 1994 (Government of Botswana 1996) and the enumerators' manual for HIES2 (Central Statistics Office 1993). The HIES 2 utilised a very broad framework for household headship: Logically, every household must have a head. The head of a household has to be a member of the household. It was the responsibility of the household members to name from amongst members who their head was. In the case of couples, where either the wife or husband is present, one of them is the head. There were exceptional instances where from amongst the members, none, due to age, qualified to assume headship of the household. In such cases, the oldest was appointed head of household (Government of Botswana: 1995: p.32) While the CSO does not provide a detailed defmition of household head, it is evident that the concept is based on the notion that the household head is the 'focal point' of decision making within the household (Government of Botswana 1996: I I). While no reference is made to the head's central role in procuring resources for the household, it is implied in the focus on certain demographic aspects of the household head; e.g., gender, age, educational status, etc. The ambiguous conceptualisation of headship raises questions about its utility as an indicator of economic weltare and household organisation over time. The ambiguous definition of the headship concept within the HIES supports the critiques that have been levelled by Rosenhouse (1989) and others. This ambiguity is reflected in the lack of an indication of the relevance of decision making to economic weltare at the household level. While the disaggregation of household data by the gender of the head has been used as the basis for the identification of gender-based income inequality at the household level, the question that comes to mind is the extent to which the broad conceptualisation of headship provides a reliable basis for the analysis of gender and socio- economic inequality at the household level in Botswana based on the HIES data. This discussion has shown the complexities of defining female headship within the context of Botswana. The diffICulties in the development of discrete typologies of female headship stem from the fluidity of cohabitation, as well as the problems associated with the reliance on de facto identification of female heads within the context of national surveys. While the case studies conducted by Oriel and Motts illustrate the importance of narrowing ISS down the roles of women as mothers and economic providers in their households, both researchers acknowledge the likelihood that headship status is often a temporary stage in the family life cycle-especially in the case of younger women who may either beco~e cobabitees or marry at Iater itages in their lives. Much of the research on female headship has focused on women, with little reference to men. Two studies (Gulbrandsen, 1996 and Townsend, 1997) point to the need to take both female and male experiences into account in order to provide a holistic approach to the study of social organisation. Wbere did all tbe men go? Anthropological studies conducted by 0mulf Gulbrandsen (1996) and Nicholas Townsend (1997) give insight into the men's lived experiences; their attitudes towards sexuality and marriage, as well as their connections and contributions to their natal families, consensual partners and children within the context of socio-economic change. Both of the scholars allude to the inadequacy of the 'residential household' as a unit of analysis within BotSwana, as it fails to capture the socio-economic and cultural dynamics that influence the actions of individuals in fluid domestic arrangements around lolwapa (the agnatic traditional family compound) over time and space. Townsend's anthropological study in the village of Mankgodi6 between 1993 and 1994 fills an important gap in studies of household organisation and social relations by focusing on the lived experiences of men over time. He challenges the approach to the socio- economic welfare of female-headed households based on the residential household as a unit of analysis in social science and policy-oriented research in Botswana, and the notion of male 'irresponsibility' which is claimed by most of the literature (Townsend 1997). Townsend constructed men's life histories focusing specifically on their relationships to various households over their courses. Based on his analysis of the residential arrangements of men of different ages between 1973 and 1993 and the accounts of women and men in the community, Townsend argues that household formation for men extends over a lengthy period of time. Due to changes in marital practices brought about by male migration over much of the twentieth century, he found that men typically established permanent marital homesteads after the age of forty. Before the age of forty, therefore, the male life course was characterised by high physical mobility between urban employment centres and the village. During this time, men establish and maintain social and economic connections with individuals in various households in the urban centres and ruraI areas: Single men, with no social children, who are not living in the village, may appear from one perspective to be single-person households and social isolates, while from another perspective they are members of a lolwapa, attached to others by a variety of competing claims, responsibilities, and relationships both economic and affective (Townsend 1997: 410). Townsend points to the significance of male social parenting; men's contribution to the welfare of their acknowledged biological offspring, as weU as their contribution towards the welfare of other young dependants such as the children of their consensual partners, their siblings and their unmarried sisters' children. Men have claims on their labour and financial resources as sons, brothers, social parents, as nephews and sometimes as sons-in-law. While noting that some men evade many of the claims to their time and income and direct their incom~ an~ en~ on their individual needs, he argues that, due to the ~ial significance of reciprocity and mterdependence, most of the men that he studied attempted to meet their obligations during their life course. ~A smalI village in the JOuth eastern pIUt of the country. 156 Townsend's study illustrates the complexity and significance of social relations and economic transactions over time and space which are not accommodated when analysis focuses on the residential household as the unit of analysis, and women in isolation. What is a household head? My study of household organisation within forty households in Gaborone and Manyana provided the opportunity for a closer examination of the concept of household head or tlhogo ya lolwapa would provide insight into perceptions and experiences relating to the links between income generation and decision making and the social contexts within which these processes take place. The guide for the in-depth interviews included questions that were aimed at gaining insight into decision-making, particularly the definition of household head. I asked women and men to derme the term tlhogo ya lolwapa and to describe the responsibilities that were associated with this role. The discussions showed that the designation of household headship is determined by a complex interplay between cultural norms, economic conditions and gendered individual agency. They also showed that the designation of headship is often determined by significant social relationships that extend beyond the boundaries of the physical residence. The views and experiences of the respondents have been divided into three main spheres of decision-making. The first was culturally sanctioned male authority, the second was social reproduction, while the third was economic provision and income generation. The terms tlhogo ya lolwapa and tlhogo are retained throughout the discussion in order to keep their socio-cultural and linguistic significance. Male culturally sanctioned authority When asked the question 'What does tlhogo ya lolwapa mean?, nineteen women and three men immediately gave the following response: 'Go raya Monna'('It means a man'). Among the women, the response was given by ten self-identified household heads, as well as by five individuals who had identified women as the heads of their households. From these immediate responses, it was evident that the term tlhogo ya lolwapa had specific gender connotations-and was synonymous with 'man or maleness' within the context of cultural beliefs and practices. As mentioned before, men continue to assume responsibility for public political affairs, such as active participation in kgotla meetings. While women are now permitted to attend meetings in the traditional meeting forum, their participation is largely passive. Men dominate discussions and make overall decisions. At the family level, men play a key role in marriage negotiations, funeral rites and other significant cultural practices. The synonymy of tlhogo ya lolwapa and monna (man) is constantly reinforced in these cultural rituals. Kgalalel07, a widowed single mother in Manyana who lived in her younger brother's compound identified him as the head of the household during the screening survey. During the in-depth interview she alluded to the role that her non-resident elder brother played as head of lolwapa (the family), showing the importance of gender and age-based hierarchies in the determination of household headship. She mentioned that when important traditional events such as wedding negotiations and funerals take place in her family and community, her non-resident elder brother would represent her household as the most senior male member of the family: The head of this household is my older brother who lives on the other side of the village. He is the one who is consulted as the elder in the household. He often arbitrates in issues 7 The names used in this article are not the real names of the intcrviewees. IS7 that are relegated to the uncles (bo malome). He is older than both of us [herself and her younger brother). When someone from the kgotla brings an. imp?rtant message, they will want to consult a man, even if he is younger than me, to gIve hIm the message from the elders. He is my kgosi (chief). I cannot supersede him in authority. If he is absent, I can receive the information as a woman only if he doesn't have a wife. If he is married, the message will be delivered to his wife, who in turn will deliver it to my brother. Kgalalelo points to the subordinate status of women in culturally based gender and age hierarchies. During marriage, she had been under the authority and guidance of her husband. She had been forced to return to her natal compound that had been bequeathed to her younger brother following the death of their parents. She effectively fell under the immediate guidance and authority of her younger brother. Her younger brother was in turn superseded in authority by her elder brother in terms of age and by virtue of his having completed the passage into adulthood through marriage. A woman in Manyana who resided in her natal home originally identified her mother as the head of the household during the screening survey. During the in-depth discussion she indicated that her elder brother who lives and works in Gaborone was tlhogo ya lolwapa because he was kgosi, even though he had virtually no input in maintaining the household fmancially. Some respondents alluded to their dependence on tlhogo for guidance, and saw that position as being associated with dealing with and resolving disputes within the homestead, as well as dealing with problems that faced household members from outside. These roles were largely associated with men. Moatlhodi, a self-identified male household head in Gaborone voiced this view: The main thing that I am responsible for in the household as tlhogo ya lolwapa is to ensure that the male responsibilities are taken care of. Care of the children and their mother. I am the one responsible for maintaining discipline in this home. The views expressed by Moatlhodi indicate that there are defmed power relations within consensual relationships and that women occupy subordinate positions in relation to their consensual partners. The culturally defmed gender-specificity of the concept was also reflected in the men's immediate reference to themselves as heads, while many women were ambivalent about the implications of bestowing the title upon themselves within the context of culture. When asked whether a woman was ever regarded as tlhogo ya /olwapa within the context of Tswana culture, many indicated that women could only assume that role in the absence of men; especially if their partners were deceased. Many said that these women would still be required to consult with male relatives when making important decisions that are considered as falling in the male domain, as was clearly illustrated in Kgalalelo's case. A woman in Gabor~ne indicated that while culture was changing, she was still expected to contact her uncle m the event of a death in her nuclear family as the arrangement of funeral rites is the responsibility of men. ' ~ile the vi~~s of respondents illustrated the cultural assumptions reflected in patrIarchal defmlttons of household headship that were alluded to by various scholars (Folbr~ 1991; Ch~t 1~9.1),they also supported the concerns raised by Molokomme (1991) regardmg the apphcablhty of the term 'female-headed household' within the context of Tswana culture. Molokomme's views relate to the cultural framework that continues to promote patriarchal ~elations of power and authority, whereby men are the key players in ma~ers .of cultural unportance such as marriage negotiations and inheritance practices. Whtle smgle w.omenare consi~ered as legal majors in the acquisition of property under the law, they con~mue to be subjected to the authority of male relatives within the cultural context. Marned women are doubly disadvantaged by being subjected to traditional male IS8 authority within marriage, as well as under statutory legal practices that relegate them to the position of minors in relation to their husbands. Against the background of these patriarchal beliefs and practices, however, an increasing number of women are establishing and maintaining 'independent' lives. This view was reflected by some self-identified female heads of households in Gaborone. Mmeisie, a 43 year old woman in Old Naledi gave the following reasons for identifying herself as the head of her household: When you are alone with nobody to help you-like I am, then all responsibilities fall on your shoulders. If there is a problem in this yard, and the police are called, 1 am the one that will have to answer for it. Only me. The police will not ask for my children, or my siblings. They will ask for me. They will ask me what happened. She indicated that she had little contact with relatives in the rural areas. While she may not be recognised as tlhogo ya lolwapa under Tswana custom due to her gender, many of the roles that she performs in her urban household incorporate those traditionally relegated to men. The 'new' autonomy among single women is related to the economic independence that has been partly obtained through wage employment and property-ownership. Mmeisie's views illustrate what Larsson (1989) and Driel (1994) refer to as social freedom and autonomy gained by many women in urban areas from reduced reliance on extended family economic support and influence. It can be argued that a growing proportion of single women who spend the majority of their adult lives and establish homesteads in the urban areas are affected less by kinship influences than those single mothers whose lives remain embedded in their natal families out of economic need. The degree of autonomy that women exercise over their lives and the lives of their dependants is highly contingent on the degree of economic power that they possess. While the autonomy of some of the single women may be adversely affected by economically motivated cohabitation e.g. 'marriage for maize meal', there is a large proportion of women who remain alone by choice or due to the failure of co-resident consensual unions (Drie11994, Motts 1994). Domestic social reproduction For the purposes of this study, domestic social reproduction was defmed as including childbearing, child-rearing and the performance of activities that are related to the daily maintenance of household members as described by Laslett and Brenner (1989) and Fox (1993). These activities are often subsumed under 'housework' within national household surveys and labour statistics, and are not considered to be economic activities. The Labour Force Survey of 1995/96 mentioned the following about female labour force participation rates and housework: Females generally show lower rates of economic activity than males and many are engaged in housework which is still very much work but is not included as an economic activity internationally largely because of the problems of putting monetary values on such activities (Government of Botswana 1996: 3). Respondents pointed to the central significance of social reproduction especially with regard to the maintenance of life. In addition, it was clear that the realm of housework extends far beyond domestic chores to include the procurement of resource~specially food that is consumed by household members. Batswana women have always borne primary responsibility for household food security. This is reflected in Schapera's anthropological accounts on the gender division of labour in agricultural production. Women have always done of the bulk of the work in the agricultural production cycle such as planting, weeding, bird-scaring and harvesting, while men only participated in the initial clearing of the land and the preparation of the land for planting. Colonisation and labour migration significantly altered the relations of production 159 and gender systems by designating Tswana men as 'breadwinners' w,ithin ~e ca~italist wage labour system. While an increasing number of women are breadwinners m theIr own right, they continue to bear primary responsibility for social reproduction which includes the maintenance offood security. Several respondents remarked that tlhogo ya lolwapa is the person who takes responsibility for the daily survival and welfare of members of the household. The qualification of the terms 'survival' and 'welfare' varied according to the gender of the respondents, with men alluding more to general welfare, while women provided details regarding home-management tasks ranging from catering for their childrens' nutritional and educational needs to food acquisition and preparation. Men subsumed 'welfare' under their culturally sanctioned authority. Three men in Gaborone alluded to their responsibility for the overall material welfare of their families, indicating that they were responsible for 'bringing home the food'. The views expressed by women reflected their pre-occupation with making ends meet on a daily basis. Most of the female heads of households indicated that they undertook these responsibilities single-handedly, while the women who were married and cohabiting indicated that they consulted their husbands and partners as they were partly responsible for meeting the financial requirements for household expenses, Provision of resources The procurement of resources includes the production and/or acquisition of cash, goods and food for consumption within the household, as well as the provision of basic necessities such as shelter. While some of the respondents alluded to the links between culturally sanctioned authority and resource provision, it appeared that the link between social reproduction and resource provision was stronger. The link between culturally sanctioned authority and resource provision was expressed in the comments made by couples, and women who were single due to the disintegration of resident consensual unions, or the deaths of consensual partners. The latter link was evident in the views of women who had never cohabited; some of who had established their own households, and those who resided in the households of their single mothers. The discussions also illustrated the complexity of provision of resources and ownership of assets between and within households, and the likelihood that households headed by women would rely on informal networks that sometimes had direct implications for their personal power and autonomy. Several widows referred to the fact that their late partners had been the primary income earners in their households. A widow in Manyana mentioned that if her husband were still ~li~e, he would assume responsibility for supporting the household fmancially. She mdlcated that she was the head of her household because she was a widow, and had assumed primary responsibility for catering for the needs of her children. While seeming to be certain about her role, she felt that her migrant son could be regarded as tlhogo due to his regular contribution to her welfare as well as that of his siblings. Several widows in Manyana indicated that they relied on their non-resident adult children for the upkeep of their homes. An elderly widow who had originally identified herself as the head of her househo.ldin Manyana indicated that she depended on her son to pr?vide food.for her,.and saw hun as the main decision-maker. Another elderly widow mentioned that It was difficult to determine who the head of her household was as her sons assisted her fmancially. She finally said that she thought it was her eldest s~n, as he was the one that she appealed to for financial assistance. Another widow who lived with her daughter and several ~dchildren said it was difficult to determine who the head of her household was, as her resident daughter and non-resident sons contributed to her welfare. 160 The complex interface between economic provision and headship was also reflected in the households of cohabiting and married couples. While men assumed the role of breadwinner and principal decision-maker, it was evident that women played a prominent role in food provision. This was reflected in the case of Mmantho and Mmoshe, a couple in Manyana. While Mmantho had originally indicated that her household relied on her partner's income from casual work, she later indicated that she had been feeding her family of ten with maize and sorghum that she jointly produced with her mother. She indicated that the little income that Mmoshe made was used to supplement the food supply, pay for the children's educational expenses, clothing needs, and to purchase fuel for lighting. Ownership and control over major assets are important aspects of headship. Several respondents referred to tlhogo as monnga lolwapa (the owner of the home). The patterns of home ownership were largely determined by partnership and marital status in both locations. The proprietary consequences of marriage in Botswana subject married women to the marital power of their husbands, who are appointed the legal administrators of their marital estates. Husbands effectively have the right to acquire and sell property without reference to their wives (WLSA 1994). These proprietary consequences -of marriage have dire implications for women. A female respondent in Gaborone indicated that her husband had deserted her. At the time of the interview, she was residing in the marital home with her seven'-children and four grandchildren. She stated that her residential plot was registered in her husband's name, and that he periodically returned to the home threatening to evict the family and to bum the house down. She regretted the fact that the plot had been registered in his .ame, as they had built the home together. Her attempts to apply for a SHHA8 plot were turned down on account of her marital status. She indicated that she could not afford legal advice. While most of the single women in Gaborone were the legally registered owners of their SHHA plots, some who had previously cohabited indicated that their homes were legally registered under the names of deceased partners. Neo's and Sekgopi's partners died after long periods of cohabitation. While Neo's partner had ceded his SHHA plot to her in a written will, Sekgopi's partner had not left a will. Sekgopi said that her partner's family tried to take possession of her house, threatening to evict her family of ten. After lengthy negotiations with her partner's family, they agreed to allow her to stay in the house with the couple's children. Her family continues to be liable to eviction due to the fact that she has not established legal ownership. She also had outstanding service loan arrears with the SHHA. If evicted, she would have no legal avenues of defence under customary and statutory law because she was a cohabitee. The social workers and SHHA officers in Gaborone reported that they were faced with many property disputes regarding the ownership of SHHA plots following the termination of relationships of cohabitation. They indicated that often, single women would apply for SHHA plots while single, and cede ownership to their partners during the course of their relationships on the premise that these men were assisting with the payment of service levies. When these relationships end (usually due to the establishment of new relationships by men), the women and their children were subject to eviction from the residence. Another area of complexity in the ownership of housing is reflected in the lives of single mothers who reside within the households of their natal families. Kgalalelo, a widowed mother of five in Manyana was compelled to return to her deceased parent's home when her marital homestead had physically collapsed. She indicated that the home was left to her single younger brother who is a migrant worker in South Africa. Kgalalelo is unemployed, and does not have the financial means to build a home for her family. She said that while I Self Help Housing Agency low-income housing scheme. 161 she was occupying and looking after the home for her younger brother, she was concerned about the welfare of her family once her brother returned and established a family in the compound. Motlatsi resided with her children in her grandmother's compound that had been inherited by her single mother. She also indicated that her brother stood to inherit the homestead when her mother died, and was also concerned that she would have to move out of the homestead with her children. She indicated that while she planned to establish her own homestead, she spent most of the proceeds from her business (vegetable production) on food and educational expenses for her children, and was unable to save money for building materials. While these women can apply for residential plots through the local land boards, they do not have the financial resources for construction as their limited resources are allocated to children's expenses. These women and their children continue to rely on relatives for shelter. Consultation There were gender differences in decision-making roles and responsibilities. Single mothers dealt with all matters pertaining to the welfare of their dependants on a daily basis; fmancial matters pertaining to assets such as houses and other property, as well as buying food and clothing for children. In the households of couples, it was evident that there was a defmed gender division of labour. Women cared for children, and took care of domestic chores, while their partners were held responsible for the financial upkeep of the household. It appeared that matters pertaining to property and major expenditure were the man's responsibility. The processes of consultation between partners varied. Most couples indicated that they consult/advise each other on matters pertaining to fixed property (e.g., alterations to buildings). Observations from the discussion on headship The forgoing discussion has illustrated the complexity of household headship and decision- making which occurs within and between households. The predominance of patriarchal patterns of male authority over women within and between households continues to be reproduced through cultural nonns and legal traditions. The dependence of women on men has largely been reinforced by gender patterns of wage employment and the creation of the male breadwinner ideology over time. The prominence of single women as decision-makers in their households is largely due to the abs~nce of senior male adults in the household, and the economic 'independence' that they gam fr~m ~~ing their own income. While many of these women are subjected to male a~thonty Withinthe context of traditional rites, and many identified men as the heads of therr households, they assumed primary responsibility for the daily provision of resources and the welfare of their dependants. The contribution of wives, female cohabitees and other household members to economic provisi~n and social reproduction largely goes unnoticed within conceptualisations of headship. ~e delineation of the different spheres of headship pointed to the close interface ~twee~ socla~r~p~oduction and economic provision through processes of mothering. ~e dlsc';18slonof mdlVldualexperiences, choices and survival strategies illustrate that the daily s~l~al of household members is largely determined by the gendered identities and diViSionsof labour within the context of limited choices. Conclusion The lessons .learned from the body of literature that focuses on the roles that individ~ un~ertake with regard to economic production and social reproduction within domestic units stem from th. ab'l' . err Iity to provide a more holistic picture of the interface betWeen 162 structural factors (e.g., culture and the economy), patterns of social and economic co- operation at the domestic level, and the shaping of individual ideology over time. The strength of this type of analysis rests in the ability to look beneath common conceptualisations such as the residential household and headship in order to examine the different configurations of gender relations and socio-economic welfare. Notes Acknowledgement: My PhD study was funded by the University of Botswana! University of Manitoba Link Project, and the University of Botswana. Thanks go to my supervisor, Professor Nancy Howell, my thesis committee members Professors Charles Jones and Lome Tepperman at the University of Toronto. Many thanks for Professor Jay O'Brien for his comments and suggestions on this paper. References Brown, Barbara B (1983) 'The Impact of Male Labour Migration on Women in Botswana' African Affairs 28 (328) 367-388. Chant, Sylvia (1991) Women and Survival in Mexican Cities: Perspectives on Gender. Labour Maruts and Low-Income Households New York: Manchester University Press. Chant, Sylvia (1997a) 'Women-Headed Households: The Poorest of the Poor? Perspectives From Mexico, Costa Rica and the Philippines' IDS Bulletin.28 (3) 26-48. Comaroff, John L. & Simon Roberts (1977)'Marriage and Extra-Marital Sexuality: the Dialectics of Legal Change Among the Kgatla' Journal of African Law 21(1) Folbre, Nancy (1991) 'Women On Their Own: Global Patterns of Female Headship' Washington DC: International Centre For Research on Women. Fox, Bonnie (ed) (1993) Family Patterns and Gender Relations Toronto: Oxford University Press. Garey, Anita & Nicholas Townsend (1996) 'Kinship, Courtship and Child Maintenance Law in Botswana' Journal of Family and Economic Issues 17 (2) 189-203. Government of Botswana (1995) 1991 Population and Housing Census Dissemination Seminar 1-4 May. 1995 Gaborone: Central Statistics Office. Government of Botswana (1996) Stats Brief No 96/5. Gaborone: Central Statistics Office. Gulbrandsen, 0rnulf (1996) Poverty in the Midst of Plenty: Socia-Economic Marginalization, Ecological Deterioration and Political Stability in Tswana Society Bergen: Norse Publications. Islam, Mahmuda (1993) 'Female-Headed Households in Rural Bangladesh: A Survey' In Joan Mencher and Anne Okongwu (eds) Where Did All The Men Go?: Female-Headed/Supported Households in Cross-Cultural Perspective Boulder: Westview Press: 233-242. Izzard, Wendy (1982) 'The Impact of Migration on the Roles of Women' Gaborone: Central Statistics Office. Izzard, Wendy (1985) 'Migrants and Mothers: Case Studies From Botswana' Journal of Southern African Studies 11 (2) 258 - 280. Kabeer, Naila (1993) 'Revisiting the Links Between Gender and Poverty' IDS Bulletin 28 (3) 1-13. Kennedy, Eileen & Pauline Peters (1992) 'Household Food Security and Nutrition: The Interaction of Income and Gender of Household Head' World Development 20 (8) 1077-86. Kerven, Carol (1982) 'The Effects of Migration on Agricultural Production in National Migration Study Vol. 3. Gaborone: Central Statistics Office: 527-653. Kossoudji, Sherrie & Eva Mueller (1983) 'The Economic and Demographic Status of Rural Households in Rural Botswana' Economic Development and Change 3 (4): 831-859. Laslett, Barbara & Johanna Brenner (1989) 'Gender and Social Reproduction: Historical Perspectives' Annual Review of Sociology 15: 381-404. Larsson, Anita (1989) Women Householders and Housing Strategies: The Case of Gaborone, Botswana Stockholm: The National Swedish Institute For Building Research. Molokomme, Athaliah (1992) 'Children of the Fence: Maintenance of Extra-Marital Children Under Law and Practice in Botswana' PhD Thesis. Leiden, University of Leiden. Moser, Caroline (1993) Gender Planning and Development: Theory, Practice and Training London: Routledge. Motts, Nigel (1994) 'A Case Study of The Social and Political Economic Determinants of Female- headed Household Poverty in Kang, Botswana' Master of Science Thesis. University of Guelph. 163 Narayan, Deepa; Sarah Koch-Schulte, Raj Patel, Anne Rademacher & Kai Schaffi (2000) Voices of the Poor: Can Anyone Hear Us? New York: Oxford University Press. O'Laughlin, Bridget (1996) 'Gender Dimensions in Household Survey Research' Population and Development Teaching Texts Institute of Social Studies Population and Development Programme. The Hague, The Netherlands. Peters, Pauline (1983) 'Gender, Development Cycles and Historical Process: A Critique of Recent Research on Women in Botswana' Journal of Southern African Studies 10 (I): 10-122. Rosenhouse, Sandra (1989) 'IdentifYing the Poor: Is 'Headship' A Useful Concept?' World Bank Living Standards Measurement Study Working Paper No. 58. Washington DC: The International Bank for Reconstruction and DevelopmentfThe World Bank. Schapera, Isaac (1933) 'Pre-Marital Pregnancy and Native Opinion: A Note on Social Change' Africa 6: I: 59-89. Schapera, Isaac & John L. Comaroff (1991) The Tswana (revised edition) London: Kegan Paul International. Syson, Lucy (1972) 'Unmarried Mothers in Botswana' Botswana Notes and Records 5: 41. Townsend, Nicholas (1997) 'Men, Migration, and Households in Botswana: An Exploration of Connections Over Time and Space' Journal of Southern African Studies 23 (3) 405-420. Tsie, Baleti (1995) The Political Economy of Botswana in SADCC Harare: SAPES Books. Wood, Lucie Saunders and Sohair Mehanna (1993) 'Women-Headed Households from the Perspective of an Egyptian Village'. In Joan Mencher & Anne Okongwu (eds) Where Did All The Men Go?: Female-Headed/Supported Households in Cross-Cultural Perspective Boulder: Westview Press 193-202. Women and Law in Southern Africa Research Project (WLSA) Botswana (1992) Maintenance Laws and Practices in Botswana Gaborone: Printing and Publishing Company. Women and Law in Southern Africa Research Project (WLSA) Botswana (1996) Women Marriage and Inheritance Gaborone: Printing and Publishing. 164