The Ethiopian ”Family Collective” and Child Agency

Any discussion of the “family collective” and the place of children within it requires an understanding of what constitutes a family and what a child is. Since ”family”, “child” and “agency” are all value-laden notions full of meanings, and vary cross-culturally, they need to be unpacked, and this is one of the things I will be doing throughout this lecture. What is a family? Is there one ideal Ethiopian family collective? Who is the ageless, genderless and culture-neutral “child”? Also, how is the agency of children in the plural differentiated, intersected and dissected by religion, ethnicity, stage of childhood, and social maturity, as well as rural and urban locations? To what extent are the actions of children constrained or enhanced by the family collective and vice versa? Clearly, these are very broad and complex questions, and while I do not attempt to answer them all, I try to approach them from the perspective of my own discipline of geography, largely focusing on rural-urban contrasts, differences and linkages. In so doing, I underline some overlaps, contradictions and peculiarities in the ways in which children exercise agency within families by drawing on (and sometimes inferring from) the limited sociological and anthropologi-

I view family collectives in three contexts, namely in extended families, in households and in the intersection of the two, as well as in kinship and social networks of support. To begin with, households and families are contentious terms when discussing the internal functioning of Ethiopian society, largely because societies exist in fluid ways and family relationships are undergoing constant change in their forms and meanings (Ansell & Van Blerk 2004). Although families and households have been used interchangeably, the former are task-oriented residence units whereas the latter are conceived of as kinships groupings with continuing legal, genetic and/or emotional relationships (Boyden 2006). In other words, whereas "households" are units of economic viability consisting of people who coreside, and usually pool their incomes and eat at least one meal together when they are at home (World Bank 2001cited in Boyden 2006, "extended families" are those which are bonded through blood or kinship ties but whose members may -and often do -live apart (Young & Ansell 2003). Extended families may also be split among several households, with membership constantly changing through, for example, individual migration (Ansell & Van Blerk 2004).
Although families vary greatly in form and size, and while nuclear families exist in Ethiopia, extended families are the predominant form in most parts of the country. As Boyden (1994: 45) highlights in the wider context of Africa, traditionally, a woman lives with her husband's family after marriage and their children become the responsibility of the entire family group. The extended family, in which sons live together with their wives and children in one compound, is less common today. However, solidarity with kin and family ties continues, in forms adapted to new social and economic circumstances.
Data provided by the Ethiopian Central Statistics Authority (CSA 2002) suggest an increase in the incidences of the "enlarged nuclear household", a central family unit whose numbers are swelled by distant family members. Migrants may spend long periods residing with relatives, and small-scale enterprises often depend on labour provided by such members of the extended family. Enlarged nuclear households also result from household units temporarily joining to form one unit -in a process known as "joining stoves" -as a way of either coping with the absence of key household members or supporting a weaker household going through a difficult time. From an analytical perspective, as households and families are both dynamic social units that "tend to contract and expand" as members live across space, and as the nature of the relationships binding them perpetually changes, it is useful to acknowledge a third concept, "the household/family nexus" (Ansell & Van Blerk 2004), which I believe is what constitutes the "family collective".
Family collectives have different contexts in different parts of Ethiopia. According to the Norwegian social anthropologist, Jan Brøgger (1984), who carried out fieldwork among the Sidamo in southern Ethiopia, the patrilocal family structure is such that multiple wives may reside in the same compound. 2 Ideally, these co-wives work interdependently as this is advantageous regarding work on subsistence agriculture, cash crops, domestic tasks, and raising children. A related practice in Gedeo, where I carried out my field research for the rural case study, is that of levirate marriage, or widow inheritance, by which a man who takes the wife of his deceased brother is considered to be acting in a socially appropriate way. According to Bevan and Pankrust (1996), widow inheritance has three pivotal functions. First, it ensures the preservation of clan land from being inherited by members of other tribes, as inheritance rights follow the patrilineage. Second, the brother assumes the responsibility of taking care of the wife and children of his deceased brother as an "heir" to his family. Third, and most importantly, levirate marriage ensures that the reproductive power of the widow is "not wasted", especially if the husband dies without having had children, who are seen as both a family resource -as "wealth in people" -and as sources of prestige and social status. 3

Kin and social networks of support
Households and families are neither a unitary phenomenon nor closed social entities, but are intricately intertwined with the wider social networks of support. Perhaps a central concept in family collectives is the one described by the Swedish social anthropologist Eva Poluha. Poluha (2004: 55-59) talks of mahiberawi nuro, in which one must take part in order to be a meaningful member of society: Life in Ethiopia … is dependent on your neighbors and friends; you must live social life or mahiberawi nuro [literally meaning collective life] and do so through membership in various social associations [e.g. solidarity groups, redistributive groups, work groups, religious groups, funeral associations]. When you yourself or a member of your family becomes sick or die your association will help you manage socially, financially and morally … they are your social security.
What this suggests is that the most important functions of family collectives are care and social security, that who belongs in the collective depends on the context, and that obligations are not enforced upon members but rather depend on love, reciprocity and trust. More recently, however, due to deepening poverty, economic inequalities, and environmental and man-made disasters, as well as the impacts of HIV/AIDS, "stylized frac-ture" (Kabeer 2000) in family collectives, 4 and the composition of members and the social contracts governing them, is taking place and producing yet different typologies of households and families.
Family collectives are neither unified nor homogeneous. One of the dangers of talking about the family collective is the tendency to construct it as unitary, cohesive group. Another is the risk of simplifying and romanticizing relationships among the various members who, as individuals, have diverse interests of their own. Family collectives neither always consist of people with common goals nor is there equal power in the access to and distribution of resources or in decision making processes. Instead, there are conflicts and internal power struggles based on age, gender and authority. However, as I have argued elsewhere (Abebe 2007, Abebe & Aase 2007, Abebe & Kjørholt 2008, members have expectations and obligations within the family, which means they have strong reciprocal 5 relationships which entail complex responsibilities towards each other, though these have to be balanced against their own needs and desires. 6 A useful analytical framework in this respect is Sam Punch's (2002) notion of negotiated interdependence in inter-and intra-generational relations, to which I will return shortly. However, first I will highlight how childhood as a phase of life is demarcated by different life events, such as birth, namegiving, circumcision, christening or baptism, commencement of work, and the onset of menarche, among others, and I will explore how these life events give children central or marginal positions, both symbolically and in relation to other members of their families.

Childhood in a life course perspective Life stages in childhood
Many Ethiopian societies acknowledge a number of major developmental milestones within childhood, and some mark these formally with rites of passages during which a number of duties and responsibilities are bestowed on children. I begin with views of family collectives on pregnancy and childbirth -a logical prelude to all further discussions about children. In her book This Place Will Become Home, Laura Hammond (2004) gives a vivid ethnographic account of the life-cycle rituals of returnee refugees from Sudan to Abi Adi, in northern Ethiopia, and how birth and christening are associated with symbols of emplacement and community formation. Like most Ethiopian societies, in Abi Adi a child is often very much longed for and considered as a gift from God. A marriage without children is thought of as troubled and the partners are encouraged to divorce and remarry as soon as possible (Poluha 2008). At birth, the first priority for a traditional birth attendant is to save the mother and then the child. A Christian woman is considered impure after birth and is purified by a religious ritual performed 40 days after the birth of a boy and 80 days after the birth of a girl. According to Hammond (2004), christening marks the child's entry into the church (both as a member of a particular parish and in the wider Christian world), and also its acceptance into the Christian community, the principles of which are to be taught to the child by those who the parents choose to be its godparents.
When boys and girls are baptized they are given a name which corresponds to the saint that is celebrated on the day of baptism. This name can be used in every daily life but is often replaced by another name that is indicative of a wish the parents have for or of the child and which reflects significant events that occurred at the time of the baby's birth. Hence, for example, whereas my church named me Sebehate-selassie, literally meaning the "praise to trinity", my "world name" given to me by my parents is Tatek which in Amharic means Get Ready. I was given the latter name because I was born at a time when Ethiopia underwent the socialist revolution, in the mid-1970s, which was also the time when there was a border conflict with Somalia. On a more personal level, I was given this name because my father was experiencing turmoil in his work life and my arrival meant I had to be prepared for defending him. What this example highlights is not only the generational link between parents and children but also the demands and/or expectations bestowed on a newborn baby by re-lating its naming to the familial, social and political contexts in which it comes into being.
Among the Bishoftu Oromos, boys are circumcised when they are 7 or 8 years old, whereas among both the Oromos and Amharas girls are often circumcised after birth (Nardos 2008). However, the circumcision of girls, though still widespread, is of less religious significance compared to the circumcision of boys and is being more widely questioned in Ethiopian society as a result of campaigns against what are described as harmful traditional practices (Berihun & Aspen 2007). In many places in Ethiopia, babies are usually strapped on the backs of their mothers, grandmothers, aunts, or sisters from when they are a few months old. The child will be breastfed, but when the mother gives birth the new baby will be handed into the care of close relatives until the mother's period of childbed is over and she is able to resume her wifely and motherly duties.
An important sign of maturity in most Ethiopian cultures is when children are said to be able to distinguish good from bad, and right from wrong. This is related to their ability to reason and understand why they are punished, and this may occur some time between the ages of four and seven years old. Studies indicate that children are thought to mature earlier in towns, around the age of four years, compared to in the countryside, and furthermore, girls start to be helpful around the house at a younger age than boys (Poluha 2008). During early childhood the attitude of parents towards children is more one of laissez-faire, partly due to the view that childhood is a state of imperfection. Amhara parents leave their children with little supervision until they are old enough to perform economic tasks but later on train them in an authoritarian fashion to make them grow into adults as quickly as possible, without allowing them a period of playful adolescence (Messing 1985).
Because social maturity raises children's social and economic status and constitutes public affirmation of community membership, it far outweighs the universally used criterion of chronological age. Moreover, the age at which childhood ends is often prescribed by gender. According to Berihun (2004), who studied the dynamics of identity and otherness among the Gumuz of north-western Ethiopia, boys have to wait until after physical maturity before they are recognized as men, while girls enter womanhood at the age of menarche, commonly through marriage which usually is preceded by informal courtship during annual cycle of events.
In many parts of rural Ethiopia, traditional marriages either involve the church or kin. Marriage may also involve abduction -consented and forced -as well as child bride practices (Berihun & Aspen 2007). As I will explain later, these forms of marriages not only represent scenarios in which the collective interests of the family constrain the individual agency of the child but they also suggest that growing up physically is not simply associated with declining dependency but instead may create a situation of diminishing relative autonomy depending on the cultural contexts in which the children find themselves at different times.

Gender, space and time
It is perhaps important to emphasize how gender shapes divergent trajectories, constraints and opportunities among children (boys and girls), though not in the same way between rural and urban areas. As Poluha has forcefully argued in her forthcoming book The Life Worlds of Boys and Girls in Ethiopia: You do grow older, you can move and live elsewhere, you can learn languages, marry someone from another ethnic group or country, but you do not alter your gender, at least not often, and not in Ethiopia.
The gender of boys and girls is distinctly observable in their dress code, hairstyle, patterns of desired behaviour, the spatial and temporal aspects of their lives, the nature of work they engage in, their life aspirations, and their choice of role models. According to Ayele (2008), who carried out research on children in Amhara Region, gender, age, religion, and area of residence all affect the way in which children live and grow up. Ayele argues that life is more burdensome for children in the countryside and that even there, girls tend to be given less time to study than boys. The heaviest tasks are carried out by boys living in the countryside, and the most timeconsuming ones are carried out by the rural girls. This brings me to the discussion of how children's agency is differentiated and how it is negotiated in everyday life within the family collective and beyond.

Children's agency in family collectives
A theoretical overview of children's agency I will start this part of my lecture with a quote from Sinidu, a 15 year old girl whom I interviewed in Addis Ababa: My mother, if she is worried about me when I stay long doing business [working] in the evening, prefers to treat me like a child. But if I happen to disobey, make mistakes or fail to do something according to her expectations, she will say to me, "How come you find this difficult?" "When I was your age, I had married your father". (Abebe 2008: 32) Generational categories such as "childhood" and "adulthood" are culturally variable, also reflecting the significance of power, position and authority (Christiansen et al. 2006). The quote from the 15 year old girl indicates childhood is a shifting category that is imposed, negotiated and renegotiated on the basis of one's relative power in the generational structure. Sinidu's view that she was not a child, despite being fifteen years old, shows how children may view themselves and their childhood differently from how adults view them. Conversely, her mother's implicit assertion that Sinidu was a child -except when she was exercising her parental power -invokes the idea that childhood is a temporary and situationally diverse phase of life. It not only demonstrates how, as a relative social category, it varies according to gender and generational relations but also how children position themselves and are positioned by adults differently in different contexts.
Agency is understood by Robson et al. (2007: 135) as an individual's own capacities, competencies and activities through which they navigate the contexts and positions of their life worlds, fulfilling many economic, social and cultural expectations, while simultaneously charting individual and/or collective choice and possibilities for their daily and future lives. Conceptualizing young people as agents, i.e. competent social beings, means viewing them as "doers" and "thinkers" rather than as "human becomings" (ibid). Thinking and doing are important components of any definition of agency, and there is much evidence that young people are thinkers and doers, as my following discussion will reveal. Continuum of Agency Continuum of children's agency (Adapted from Robson et al. 2007: 144.) In exploring the differentiated nature of children's agency, I use this diagram of "continuum of agency" (figure 1) as an analytical template against which the remaining part of my talk is charted. Children's experiences of agency change depending on who they are with, what they are doing and where they are (Robson et al. 2007). This is because they can be simultaneously powerful and powerless with respect to different aspects of their social worlds. Their everyday lives move back and forth along a continuum of diverse experiences in relation to changing degrees of power and powerlessness (ibid). According to Punch (2007) power relationships are negotiated and renegotiated with different people in different contexts at different times. As the diagram shows, a child may experience a lack of agency in some areas of its life but can exercise it in others, and an individual moves along the continuum accordingly, through time and in relation to decisions that are made. Thus, agency is dynamic. I will identify four differentiated forms of children's agency and give examples of each. These forms, 7 which Robson et al. (2007) have outlined the theory of, range from almost no agency exemplified, in this case, when children are forced to act against their will, through how little agency is exhibited when children (for example, child beggars) act out of necessity in order to improve their own lives, or when children simply reproduce adults' demands. Moving up the scale of agency, there is subtle resistance by children, whereby all attempts to exert agency are secretive, resulting in negotiated interdependence between children and their families. Finally, it is possible to find different ways in which children openly act with adult sanctions and make decisions in search of relative autonomy.

Almost no agency
The first example of "almost no agency" that I provide is that of abduction and exchange marriage, where there seems to be little or no agency on the part of girls and where collective decisions permit a minimum of individual choice. Berihun (2004) carefully documents stories of otherwise victimized girls who are exchanged for marriage, and the socio-cultural contexts of reconciliation and conflict that underpin the practice among the Gumuz of north-western Ethiopia. The practice of exchange marriage requires young boys to have sisters or female relatives to exchange for a bride. However, the boys who want to marry may either have no female relatives readily available, or if there is one, her will to agree marrying the prospective groom on the other side of the exchange may not be guaranteed. Taken at face value, exchange marriage is male-oriented and girls are passive participants in the process. However, Berihun argues, the situation is rather complex when one looks closer at the actual practice, and there are many indirect ways in which girls may exercise agency and constantly challenge the process. For example, a young girl may demand to know the person (relative) who is supposed to exchange her for a wife. This will give her the chance to negotiate in advance and hopefully influence the choice of the family that her brother will be seeking the exchange from, and which she will eventually be marrying into. In other contexts, if the girl is less interested in her family of exchange, she may marry quickly while the male relative either has not yet had time to decide, or already has a wife, or is between marriages and in no hurry to consider an exchange.
However, as Berihun further states, youngsters are not individuals who wait and follow what adults want them to do. Instead, they decide and take actions of their own even when it is against not only the will of the family but also the interests of the wider lineage as well as the normative preferences of the cultural setting. Abduction is used a short cut to avoid exchanges with unknown individuals. Thus a boy may abduct a girl of his choice forcefully, and this will eventually become an issue for the family collectives of both sides to settle. In these different contexts, the agency of girls can be conceptualized as being "backstage" rather than "front stage". While remaining in the background, their agency is exhibited as ranging from having almost no agency and being a victim of forced abduction, to giving consent to be abducted, to demanding to know the boy she is going to be married to, and to exercising subversive agency in negotiating indirectly in the choice of which boy and the process of marriage. In contrast, the agency of a boy responsible for an act of abduction represents that of a public agency, in which they act against the will of the girl and the family collectives.

Agency as reproduction of power
Relationships in a family collective in Ethiopia are usually hierarchical, depending on gender and age. Children are expected to learn and to show adults great respect, and younger children should also respect their older siblings (Poluha 2008). However, men and boys are more respected than women and girls and since physical punishment is quite frequent, men and boys often have the upper hand. Children show their respect through obedience and helpfulness and in return expect to be treated with love and care. Poluha further argues, if respect is lacking, parents and older siblings have a wide variety of sanctions that they can use, ranging from curses to beatings, in order to instil proper behaviour into children.
In my research, the shifting degrees of children's agency is captured through the roles they adopt as beggars, both in the context of enabling their disabled parents to beg, which gives them more generational power over the activity, as well as in their engagement in mobile livelihood strategies to obtain the economic power in order to be at the heart of their household livelihoods. Individually, child beggars pursue different strategies to attract pity, including appearing helpless, victimized, sick, hungry, and not having anyone to turn to at home. In this way, they turn their impoverishment into opportunities for material ends (Abebe 2008). This is an often ignored aspect of the "agency of victims" (Utas 2005), highlighting that even when children are forced by circumstances they are still able to exhibit agency in many indirect ways.
Children's agency is also reflected in the reproduction of adult power. In her book The Power of Continuity: Ethiopia through the Eyes of its Children, Eva Poluha (2004) conceptualizes adult-child relations in Ethio-pia as being the same as one between patrons and clients. Both adult-child and patron-client relationships are established between persons who have something to offer the other, although the value of what is provided by the superordinate is usually considered to be higher than that of the subordinate. Using children as a lens to understand the process of political transformations in Ethiopia, Poluha argues that despite radical government changes from "feudal" to "socialist" to "democratic" regimes, relations between state and people, adults and children, teachers and pupils, and men and women are visibly authoritarian and surprisingly durable.
The main reason why existing forms of power are reproduced is, according to Poluha, that children will grow up in households organized on the basis of gendered hierarchies in which domination of "subordinates" by their superiors is common. Boys are encouraged to be deferent and aggressive but respectful, which is linked with the idea of family protection against enemies, while girls are taught cultural competencies such as being less daring, quiet and submissive, which is linked with the idea of them becoming homemakers (see also Abebe 2008). During their education children learn ways of thinking, although strict gendered and generational hierarchies are found in schools, socializing boys and girls into what Poluha calls the "hierarchy trap." Poluha concludes that because children's agency is restrained by different institutions, they largely reproduce the adult cultural schema further highlighting how dominant forms of power are sustained as legitimate ways of social organization in Ethiopian society.

Agency as negotiated interdependence
Another context of agency is one in which generational power relations are negotiated between children and adults. Studies in southern Ethiopia for example suggest that there is relative autonomy and negotiated interdependence between "seniors" and "juniors". Jon Abbink's (1996) study of the Surma and Fikremariam's (2003) research on the Hammer ethnic groups are typical cases in point here. Both Abbink and Fikremariam confirm that there is a relatively more egalitarian, complementary, inter-and intra-generational relationship between adults and children, and men and women (Hamer & Hamer 1994), and between boys and girls themselves, and in case of the Surma ethnic group, it is the girls who make the ultimate choice of their marriage partner (Abbink 1996).
In their article "The impact of the cash economy on complementary gender relations among the Sidamos", Hamer and Hamer (1994) document how cash cropping and marketing have led to changes in men's and women's aspirations and how the new concept of cash accumulation has become anomalous with the historical complementarities between the sexes. In Sidamo, the present-day Southern Regional State of Ethiopia, the inculcation of mutuality is an important aspect of training and socialization during childhood. The ideology of child training practices also supports complementary relations witnessed in everyday life between men and women engaged in various forms of household production and reproduction. For example, men control cattle that provide milk with which women may choose to prepare butter, and which both historically and in the contemporary world is a scarce commodity. Complementarity is symbolized by women producing a precious commodity, which in turn is also a sign of prestige for men: Butter for women is a sign of invoking their rights to protection from harassment by men. It is also symbolic of their power in that by refusing to produce butter they can withhold prestige for men. Men in turn have to live up to their obligations of respecting and protecting women if they are to be honored with the presentation of a symbol which gives them renowned among other men … Thus, there develops a complementary articulation between symbol and practice. (Hamer & Hamer 1994: 4-5) Mutuality and reciprocity are also central to the relationship between adults and children, and also that the generational power relation between them is negotiable. In his book Humane Development: Participation andChange in Southern Ethiopia, Hamer (1987 1994) provides examples of how accumulation, distribution and utilization of resources are constantly negotiated, and how generational relationships are reflected in a circular interdependence between the elderly and youths (p. 49). As figure 2 illustrates, elderly people have wealth in the form of land and cattle, and recently also cash. They also have authority which is vested in their right to make policies, settle disputes and impose sanctions. The possession of these attributes gives them the means of obligating others, but in turn they must give up most, and ultimately all, of their wealth to their sons in the form of bride wealth, land and cattle. The arrows in the figure show how the self-interests of an elder in maintaining his material well-being and hegemony are balanced by his obligations to redistribute resources and to educate children and youths in an appropriate lifestyle. This favours the self-interests of the youths in obtaining the resources necessary to acquire wealth and reproduce the system. These interdependencies demonstrate the situation of many Ethiopian children who contribute through their labour and financially, and who in time receive land or animals as part of their inheritance to enable them to establish a more independent livelihood. The agency of children in these contexts is translated into ensuring the viability of the family they are a part of by performing their duties to their seniors while at the same time learning the material and social competence needed for their adulthood lives. Interdependent intergenerational relations also capture the dynamics of how kinship, and familial and livelihood systems are woven together into a series of relationships and mediate the ways in which children exercise agency, and how this agency varies in different circumstances (Punch 2007).
Intra-generational relations also influence the ways in which power is distributed within the family collective. In Gedeo, physical strength, birth order and gender shape children's opportunities and constraints regarding access to resources and potential for schooling. Older children and physically stronger boys may be kept at home to help their parents with agricultural work, whereas weaker boys may be sent to school. Boys may engage in selling farm produce, while girls may be responsible for domestic chores, may not receive an inheritance and will tend to have less invested in their education. This is because, as clan exogamy is the rule, girls are treated as part of family collective only until adolescence and are considered as acquiring a "new citizenship" after marriage.
Although children exert their agency in a multitude of ways, these ways are highly differentiated. In most parts of Ethiopia, and especially in middle and later childhood, work is one of the most common expressions of children's agency. Materialist approaches to agency rooted in political economy emphasize the role of children as workers in both productive and reproductive spheres, as is evident from the growing body of literature on the lives of children, work and development (Katz 2004, Nieuwenhuys 1996. Cindi Katz (2004) documents the differentiated agency and relative autonomy of boys and girls who grow crops, tend animals, prepare meals, and look after young siblings in the context of economic and environmental transformations in neighbouring Sudan. She demonstrates how herder boys, for example, despite their work entailing long hours outside the home, have more spatial autonomy than children who work by selling water, or girls who tend to be limited in the domestic sphere. Elsbeth Robson (2004), on the other hand, discusses the caring role of children, particularly girls, in Zimbabwean society. She highlights the fact that households often rely on extended family networks, mentioning that young girls are sent to the city to care for sick relatives -a growing phenomenon, which I have also observed during my research in Ethiopia. In my study, children's agency is contextualized in their role as carers, producers and traders. In Gedeo, children's work is also conceptualized as being "glocalized", where global markets determine the value their work deserves in local places, thereby highlighting not only the geographical contexts of their livelihoods but also the material exploitation such contexts entail (Abebe & Kjørholt 2008).
Children are exploited locally because their work is not considered by wider society as deserving equal status compared to that of adults. From her research in the Keralan fishing communities, Olga Nieuwenhuys (2001) documents the reciprocal relationships between women and chil-dren's work, thus extending Elson's analytical framework which draws attention to the need to account for how general cultural assumptions enter into the evaluation of work. Nieuwenhuys calls for the redefinition of social reproduction in terms of systems of reciprocal obligations that span generations, highlighting the need to study the interplay of different forms of exchange comprising the realm of the economic within the specific cultural setting.
In search for relative autonomy The fourth dimension of agency represents how children seek relative autonomy. Children recognize that their interests may conflict with those of adults and when it does not suit them, they do not always remain within the boundaries of behaviour and activities set by adults. While accepting adult agency and control, and not confronting it directly, they often use various tactics and strategies to change or avoid situations they do not like. In the study area, children's migration is a common way of escaping hardships in times of duress, but also a way of increasing opportunities for more income compared to what they would expect to earn by farming or trading in their home village, as well as a demonstration of a certain level of independence. My study in Gedeo suggests that children make decisions about migration not only for economic reasons but also to seek relative autonomy and to acquire new skills. Geographical mobility by orphans is also a strategy to receive assistance from wealthier members of their extended family households who themselves are spatially dispersed. However, studies on the vulnerability of adolescents in Addis Ababa (Bethlehem, 2005, Van Blerk 2007 highlight that it is the push factors that place constraints on their childhoods in their places of origin, such as severe poverty, parental death, family disruptions, lack of educational opportunities, and fear of early marriage that, in particular, cause girls to migrate to cities.

Conclusions
In conclusion, the extent to which children exercise agency in family collectives depends on the interaction between personal agency -the ability to create and pursue a goal -and structures of opportunities and constraints, which are constituted by institutions (tradition, rules and norms), social relationships with other social actors in material contexts, and ideas (see Bevan and Pankrust 2007). Because children are valued as part of the family collective they are likely to perceive their needs as interdependent with those of other family members rather than taking priority over them. As argued in Abebe and Kjørholt (forthcoming), due to their significance in providing substantial incomes, working children invest their economic power in ways that situate them at the centre of the household economy. They do so by contributing more financially in times when they have sufficient income, but they have to also draw on familial resources in times when they lack money and are unemployed. Interdependent relations within family collectives are also dynamic and evolve during the course of life. As Punch (2007) points out, when children acquire economic power, this tends to increase their social power. As a consequence, relationships between children and families are renegotiated accordingly, highlightingwhere as relative autonomy is not a counterpart to dependency -relative dependence is not necessarily associated with a complete lack of individual agency.