Desperate, deceived and disappointed: women’s lives and labour in rural Ethiopia and Uganda

ABSTRACT

Life history interviews from Ethiopia and Uganda, organised around experiences of wage labour, provide rich evidence on the working conditions of many poor, rural women and on what leads them to work for wages. The life histories confirm and illuminate arguments based on large-scale socio-economic surveys carried out in these two countries. Further, findings from the surveys and life histories challenge an influential literature that not only celebrates women’s agency in poor rural areas, but also remains committed to methodological individualism and ideas of choice. Drawing on primary and secondary evidence from Africa (and elsewhere), we insist that violent coercion and catastrophe trump maximising rational selection among alternatives; that the social is fundamental to individual behaviour; and the non-market is key to (labour) market participation. We also provide a brief discussion of the policy implications of this research.

KEYWORDS:

• wage labour
• choice
• violence
• gender

Introduction

Between 2010 and 2013, we used innovative sampling techniques in Ethiopia and Uganda to obtain reliable quantitative data on the characteristics of rural wage workers (Cramer et al. 2014a; Cramer et al. 2014b; Cramer et al. 2016). While analysis of the variables extracted from over 1,700 questionnaires did help us to understand the characteristics of many participants in rural labour markets, the questionnaires did not throw much light on the specific dynamics of how women and children come, often reluctantly, to participate in labour markets, or on the extent to which labour market participation involves choice theoretic types of decision.

Here we argue that some dimensions of the social forces shaping processes and individual actions in rural labour markets only become clear by combining survey questionnaires with a rich description provided by very different, more time consuming, and open-ended interviews of carefully selected respondents. We highlight evidence from ‘life’s work interviews’, whose organising principle is respondents’ experiences of labour markets, collected from a small and purposive sample of the individuals covered by our randomised rural surveys in Ethiopian and Uganda.

Data from the survey questionnaires could be and were used to select those individuals who might provide particularly interesting information by participating in an additional, less tightly structured interview about their life and work experience. Enumerators were carefully trained to identify the ‘best’ potential candidates for a longer, less structured discussion, enabling a purposive selection of more than 100 life’s work interview respondents.

Preliminary analysis of the data in the questionnaires was also helpful in finalising our purposive, stratified selection: we used these analyses to identify respondents with different levels of education, workers in processing factories, migrant labourers, divorced, separated and widowed women, and respondents belonging to different age cohorts. Random selection of respondents for work history interviews might well have resulted in a large number of very similar work histories; for example, too many similar histories provided by mature adult female coffee pickers, while no factory workers, or child labourers were interviewed. Our research confirmed the advantages of embedding or ‘nesting’ life histories in larger surveys (Schatz 2012). This approach also helped to overcome concerns that it is difficult to generalise from life history data (Lewis 2008, 563).

Life histories are rarely used in economic research (Basole and Ramnarain 2016, 145). Yet they can complement other kinds of data; they can help probe assumptions and interpretations; they can enrich understanding of the links between structure and agency, macro and micro (Bourdieu and Accardo 1999). They can ‘challenge received wisdoms by generating nuanced accounts that subvert established knowledge’ (Lewis 2008, 562). Given the importance of narrative and affect in how people actually make decisions (Tuckett and Nikolic 2017), the life history interview is more appropriate than formally structured questionnaires for insights into decision-making and into process, complexity, and contradiction in people’s lives. For example, people are understandably reticent about the intimate details of family relationships (or harassment at work) and how these affect their socio-economic experiences; the stilted performative exercise of a survey questionnaire often fails to bring these relationships to the fore. When experienced ethnographic researchers probe into life stories, they should be able to minimise the ethical risks said to be a feature of most questionnaires investigating gender-based violence (Doss 2021, 15).

Combining qualitative and quantitative evidence in ‘mixed methods’ research has become more common. However, we believe both our focus and our approach to this is distinctive. Few economists have done primary research focussed on the labour market experiences of uneducated rural women in Africa; and very few other social scientists have been able to nest carefully selected life’s work histories within a survey that collected quantitative data on almost 12,000 adults.¹ We hope this contributes to a deeper understanding of the processes determining the poorest women’s participation in labour markets in rural Africa and to more relevant policy responses.

Bargaining, agency, and Economics imperialism: theoretical issues and debates

The life histories we use in this paper were chosen because they provide remarkably articulate accounts of labour market participation, including discussions of the determinants of migration, education, marriage, and reproduction – accounts that are difficult to reconcile with much of the literature on how deprived women in rural Africa survive, but that reinforce findings from our much larger survey results. We attempt, in this article, to search for an explanation for the awkward gap between the life histories of the desperate women we met and the analyses offered in current literature both by many economists and by other social scientists.

Some of the literature that we find unconvincing, and a great deal of recent research on poor women living in the rural areas of low-income countries, has been influenced by a conceptual framework introduced by Amartya Sen in the early 1990s. A recent survey of feminist economists stressed that these researchers have found Sen’s work on capabilities and cooperative conflict particularly fruitful: ‘His work on co-operative conflict highlighted how inequalities of power did not necessarily take the form of overt bargaining but could take the form of silence and apparent consent to the existing arrangements … ’ (Gammage, Kabeer, and van der Meulen Rodgers 2016, 3). Similarly, a number of feminist economists have ‘reinterpreted the bargaining literature to take account of some of the subtle, qualitative resources that subordinate groups can bring to bear … ’ (ibid, 4). These accounts often celebrate the capacity of poor and oppressed rural women to exercise strategic forms of agency, acting as ‘the dynamic promoters of social transformations’ (Sen 2000, 189).

This emphasis on agency (and resilience) has been accompanied by a surge in academic attention on new forms of oppositional action by marginalised groups or individuals, and on social movements and campaigns undertaken by ‘women’s groups organised by development NGOs committed to social change’ (Ray 2020; Dodson 2018, 112–4; Campbell and Mannell 2016, 3; Williams 2017; Kabeer 2011, 500).

One of Sen’s most influential arguments is that ‘decision-making in the family tends to take the form of the pursuit of cooperation … Each of the parties has much to lose if cooperation were to break down, and yet there are various alternative “cooperative solutions”, each of which is better for both the parties than no cooperation at all … ’ (Sen et al. 2003, 323). These ‘solutions’ are said to have achieved positive changes in marital relationships in rural South Asia, ‘but often in silent and implicit ways’ (Kabeer 2011, 525). The belief that modelling cooperative solutions in bargaining games, where the only threat point is the dissolution of the marriage, can provide a useful framework for analysing rural marriage and gender relations ignores the elephant in the marriage. Sen does not mention physical violence when he discusses cooperative conflict within families. Instead, he concludes that: ‘Conflicts between the partially disparate interests within family living are typically resolved through implicitly agreed patterns of behaviour … ’ (Sen 2000, 192, emphasis added). Such a conclusion is at odds with a mountain of evidence on the quotidian prevalence of Intimate Partner Violence in poor rural areas (Devries et al. 2013; Burns et al. 2018).

This problem is recognised by some feminists who acknowledge that ‘much work remains to be done in developing strong and realistic understandings of the possibilities and limitations for women’s agency in coercive situations’ (Campbell and Mannell 2016, 7), or by those who acknowledge that violence is ‘a major denial of agency warranting deeper exploration’ (Hanmer and Klugman 2016, 244).² In this paper we strongly support these arguments for additional work, and we offer some evidence in exploration of the denial of agency, but note that the complex research required appears to have been crowded out by the desire to dig for new case studies celebrating women’s ‘agentic decision making’ and use of the weapons of the weak to navigate through storms of violence (Lentz 2018).³

Non-cooperative bargaining models treating bargaining within marriage as an infinitely repeated two-person game – between partners who are assumed to be entirely selfish – have been used ‘to help explain seemingly irrational behaviour by married women in South Asia’ (Balasubramnian 2013, 612).⁴ This patronising concern to describe the behaviour of poor rural women in terms of their individual ‘rationality’, using the language and conceptual apparatus of decision-making – choice, game theory, navigating between options, ‘managing’ resources in constrained environments etc. – is hard to distinguish from the concerns of, and the language used by, mainstream micro-economists.

Some of the links between mainstream economics and the work of feminist economists have been acknowledged by the President of International Association for Feminist Economics. She welcomes the ‘positive’ fact that: ‘Mainstream economics has embraced the view of the household as a collection of individuals, bargaining with one another over allocations’. She also notes, with less enthusiasm, that ‘This structure fits with the ideological tendency of mainstream economics to think in terms of optimising individuals’. Although she admits that the tools and assumptions of mainstream analysis ‘seem to handicap us’, she remains enthusiastic about some aspects of the work of mainstream economists that, in her view, impose no such handicap: ‘There are some exceptions: within a cooperative bargaining framework, we can consider how changing outside options … may affect household decisions’ (Doss 2021, 16).

It is clear that the work on rural households by Amartya Sen has been inspirational to feminist scholars (Sen et al. 2003, 16). It is also clear that both Sen and many feminist social scientists are aware of the logical flaws and conceptual problems besetting work by mainstream economists, including their work on gender issues; but it is rarely acknowledged that the formal analytics of Sen’s Entitlement Approach ‘are, at core, derived from standard set-theoretic microeconomics within mainstream general equilibrium theory, with some generalisation through access to non-market-related entitlements’ (Fine 2001, 7). This makes Sen’s analytical framework neatly compatible with the broader trend of mainstream economics ‘colonising’ other social sciences. As the examples above from rural South Asia, West Africa and Malawi suggest: ‘noneconomic or nonmarket behaviour is now understood as the rational, i.e., individual optimising behaviour, response to market imperfections. It is appropriate in face of informational, and hence market, imperfections to form social structures … and to engage in what would otherwise appear to be nonrational behaviour, as in customs, trust and norms’ (ibid: 2060). Other feminist scholars such as O’Laughlin (2007) have been more sceptical of the ‘disciplinary legitimacy’ bestowed on feminist economics by game theory and neo-classical economic axioms, and are doubtful that ‘tinkering with women’s market position by exchanging unequal collective rights to productive resources for individual ones will decisively reduce rural poverty in Africa’ (ibid., 21).

The remainder of this article uses nested life’s work histories to clarify how ‘economics imperialism’ (Fine 2002) can narrow analysis and limit our understanding of poverty, migration and employment in rural Africa. Although almost all of the life's work histories collected by the authors, as well as by Deborah Johnston and by Carlos Oya, contain valuable, often moving information, we avoid a mechanistic summary of this rich original resource; instead, we make detailed reference in this article to only two histories (Negussie from Ethiopia and Michelle from Uganda, whose names have been changed), while briefly citing some of our other interviews in both countries. The article first discusses the lives of Negussie and Michelle and some of the themes they suggest; after that we discuss other, related themes, by drawing on different life history material from our sample but in less individual detail. At the end of the article, we touch on the policy relevance of the evidence and arguments presented.

Negussie: tales of the city

When we interviewed Negussie in April 2012 she was living in a new area of half-built housing, by a compost dump for one of the large rose farms, on the margins of Ziway in Ethiopia. She had been living there for a couple of months, with no electricity or water, in a single room with her two-year-old child. She had left the father of the child, an abusive man from whom she did get some paltry maintenance payments that she collected monthly from the nearest office of the army, from which he had retired with a monthly stipend. She was waiting and hoping for another job on one of the large foreign-owned flower farms; in the meantime, she was making and selling injera and beer, but she said she did this out of boredom only because she made virtually no money from it in this sparsely populated edge of the town.

Negussie grew up in northern Ethiopia, in a village in Wollo. She knew people who had left the area. Her own older sister had left, though Negussie did not know where she was. And she had heard enough about life in big cities to have ideas of escape from what, increasingly, was becoming a life of constraining rural drudgery. Young rural women typically live in very repressive social structures: education of girls is often discouraged, their sexuality is denied, their mobility is tightly restricted, they are loaded with domestic duties from a very young age, and they are often forcibly married off or abducted without any say very young (Marshall, Lyytikainen, and Jones 2016, 11; Jones et al. 2020, 9–10). Negussie’s parents never wanted her to leave; but some relatives visiting from Addis Ababa noticed how she liked their clothes and jewellery and these relatives put the idea in her head that she might visit them in the city.

She had been married off to a much older (although not unkind) man, when she was between eight and ten years old.⁵ One day, aged 19 and fed up with her life in the village, Negussie made a break for the city.⁶ She knew where her husband hid their savings of about 400 birr. She stole the money and ran away to Addis, where the relatives who had encouraged her to ‘visit’ when they were in Wollo put her to work as their maid. This family fed her but did not pay her. She had little freedom. And while they had promised they would send her to school – she had not been able to complete the first year of primary school in Wollo – or help her to get a job as a maid in the Middle East, they did nothing about it. Again, her hopes of working in the Middle East are just like those of many other young women in Ethiopia. Very large numbers of Ethiopian women (perhaps as many as 1,500 a day, Jones et al. 2016, vii) want to go to countries in the Middle East to work as maids.⁷

Negussie’s experience of being duped into making a migratory labour market move is a common one. De Regt (2016) writes of one young Ethiopian woman, among others, who migrated to Addis Ababa as an adolescent, lured by an aunt’s promise of getting her educated, who was instead forced into domestic service by the aunt and then, repeatedly, raped by her uncle. Jones et al. (2014) also highlight the role of deception in young women’s entry – in this case again via migration – into labour markets. They point out that the combination of deception by brokers and deeply exploitative employment conditions (in the context of domestic service in the Middle East) means that the difference between migration and trafficking ‘fades into invisibility’.⁸ When migrants’ agency has to be expressed in conditions of such extreme uncertainty (‘imperfect information’), the line between choice and coercion itself fades away.

Eventually Negussie escaped from her life as a maid in Addis and ran away again. By chance she had talked to a man who had discovered that her older sister was in Ziway, one of the centres of the booming flower export industry in Ethiopia. One day, without phoning ahead, she just left for Ziway, There, about 170 km southeast of Addis Ababa, most of the thousands of workers bent over the regimented lines of rose bushes, and most of those sorting and assembling bundles of roses in the Ziway packhouses, are young women.

Negussie worked on and off for some years on the flower farms in Ziway, effectively a company town with the largest concentration of the flower business in Ethiopia. In our quantitative survey we found that 80 per cent of female respondents (working in floriculture, other agriculture, hospitality, and other jobs) in Ziway had, like Negussie, migrated to obtain their current job. Because Negussie was living in a sub-site (West Ziway) where an even higher proportion of respondents (almost 92 per cent) were migrants, her life history might be regarded as providing insights into the experience of many other females in the quantitative survey who migrated in search of employment. These migrants, like Negussie, had made long journeys to get to Ziway; some had travelled more than 500 km.⁹ When they arrived, they were likely to be employed in the most menial, ‘secondary’ types of job in the greenhouses. All of the ‘better’ jobs captured in our quantitative survey of Ziway were held by respondents who had not migrated to obtain their current job.¹⁰

Her sister advised her that the only way she could avoid being harassed by men in Ziway was to marry; she found a partner, a former soldier who had been a guard at a flower farm. Negussie soon left this man because he was violent and abusive. When we spoke to her, she had a young baby and was living without access to water or electricity services on the very edge of town – in an incomplete breezeblock room.

Given the preferences Negussie revealed (more clothes and, hence, more money rather than less; the appeal and apparent freedom of the city over village life with an older husband she did not choose), and with available information (tales of the city, a clue to where her sister was living, knowledge of the work available in the flower sector), it may look as if Negussie made a string of choices between alternatives, choices that could be modelled by an undergraduate economist with the aid of an indifference curve representing the rate at which she was prepared to give up a certain quantum of one good – say, ‘leisure’ in familiar Northern Ethiopian village surroundings or freedom from the disciplines of the wage relation (but equally, the repression and monotony of a woman’s life in rural Ethiopia) – in order to acquire more of another – the cash wage payments that might help her reach her goals. To those impressed by blackboard economics, every twist and turn in Negussie's life will look like maximisation. This may be more than déformation professionnelle; it is rather a circularity of reasoning that deflects analytical attention.

Mainstream economists usually pay little attention to the formation of an individual's preferences, which are assumed to be well-defined, self-known, stable and consistent (Finch and McMaster 2018, 297). Yet one thing that is striking in Negussie’s experience is how her experience of wage employment in a variety of labour markets in Ethiopia has shifted her preferences. Preferences typically are set prior to the actual analytical work of neo-classical economics, yet Negussie’s life history suggests that they ought to be treated as endogenous to economic activities – that the social should assume analytical priority over the individual (Milonakis and Fine 2012, 481). Preference formation is centre stage, part of the action, and not noises off. Moreover, her preferences were neither stable – they changed over time, strongly influenced by her labour market experience and accumulating knowledge – nor consistent: her dominant preference when she told her life history took the form of a fantasy that she could return home to show her parents that she had done reasonably well, which contrasts with her original simple dream of complete escape to city life.¹¹ This change in preferences is a function of a quality unique to labour as opposed to other commodities bought and sold through markets: the capacity for self-conscious reflection (Prasch 2003).

Evidence from India and China also shows that migrant preferences are unstable and contradictory, and impossible to rank ordinally (Giné, Martinez-Bravo, and Vidal-Fernández 2017; Chang 2010, 132). Similarly, the attitudes of female Ethiopian migrants who become sex workers are shot through with contradictions (de Regt 2016): they value the economic independence that sex work offers (by contrast, for example, with work as a domestic servant) but they lament having to leave school and they abhor the frequent abuse they suffer.

It is easier to make sense of Negussie's shifting goals if labour markets (and the behaviour of individuals in labour markets) are understood as radically different from markets for commodities like shoes or cement. Because many Ethiopian women supplying labour are driven by urgent needs (as opposed to mere ‘wants’), women desperately seeking work have structurally weak bargaining power relative to employers, especially in a context of widespread under-employment (Prasch 2003). A day’s labour withheld in Ethiopia is a day’s income lost and potentially life-securing needs unmet. As Joan Robinson noted long ago, ‘the individual … without private means can never be in a position to refuse to work because real wages are too low to be worth the effort. He must earn what he can get or starve altogether’ (1947, 7). Negussie, once she had left her home village, had no choice but to provide her labour to her relations in Addis or, later, to pick up whatever back-breaking work she could in the flower sector in Ziway.

Negussie is just one of many women in our survey who worked as unskilled agricultural labourers and who were particularly poor and vulnerable.¹² Female agricultural casual wage workers in rural Ethiopia and Uganda in our samples are extraordinarily poor even by comparison with the very poorest people captured by other rural surveys. Understanding poverty and the interventions that may reduce poverty in Uganda and Ethiopia requires an analysis of the paltry wages received by large numbers of people in very low labour productivity rural employment – as domestic servants, as seasonal/casual labourers on both small and large farms and on construction sites, or in bars, cafés, and small shops.

Similarly, although migration in search of wage work by young women appears to be on the rise (Chant and MacIlwaine 2015), many of these women continue to be largely invisible in both quantitative and qualitative studies (de Regt 2016, 9; Cramer, Sender, and Oqubay 2020, Chapter 7). Nevertheless, some research on Ethiopian migrants is consistent with our findings, confirming the desperation of women's quest for paid employment and the frequent disappointment, and shame, at how their lives have panned out as casual labourers, as domestic workers or sex workers in Ethiopian towns or as migrant maids to the Middle East (de Regt 2016; Jones et al. 2015). In our own interviews we often came up against the disappointment of the women we met – their distress about sexual harassment on flower farms and the shame that stopped them telling their mothers about it, and the shame that stopped other women, like Negussie, giving up and returning home or even making any contact with home.¹³

Michelle: given headaches by men

Michelle, a woman who did not know her own or her children’s ages, was probably in her mid-30s when interviewed in late 2012 in a village in southwestern Uganda. Most of the major changes in her life, the partial movements in and out of wage employment, the size of her family, where she lived and so on could certainly not neatly be described as the products of utility maximising rational choice (or as ‘fatalist’).¹⁴ An early and profound determinant of the course of Michelle’s life, one that is quite common in rural Uganda, was the separation of her parents when she was about three years old and, consequently, her separation from her mother.

The three children of her parents’ marriage stayed with her father and were brought up by one of the father’s other wives. This stepmother mistreated the children abominably.¹⁵ The drama reached a climax when Michelle’s stepmother branded each of the children on the legs with a red-hot panga blade. The stepmother claimed this was to rid them of jiggers in the children’s legs. One of the father’s sisters found out and was so horrified she whisked away the two youngest children, one of them being Michelle, to live with her.¹⁶

Michelle managed two years of primary school before she became pregnant. Even while at school, she was obliged to help with onerous domestic and agricultural chores, because her Aunt had rescued her from her stepmother. After leaving school, she carried on doing this unpaid work, but also started doing paid casual wage work: pruning, weeding, and mulching in local banana plantations. The quantitative survey confirms that Ugandan women who like Michelle are paid wages to do these tasks have, on average, received significantly less education than all other women in the survey.

Michelle claimed that she was helping her aunt, but ‘helping’ is a loaded word in the world of rural work. Very often, ‘helping’ turns out to be a thinly disguised wage relationship. Thus, a very young girl may appear to an interviewer to look rather different in demeanour or dress from other children in a household. The replies to questions about this young girl suggest that she is a ‘relation’ of some sort who is ‘helping’ in the house in a common arrangement that is effectively a form of domestic service paid in board and lodging. We interviewed one such girl (in Southern Ethiopia), who had also needed rescue by an aunt. She had not been burnt, but she had been suffering from extreme hunger. Eman had left her father when she was about six and, despite her hunger, walked for more than five hours desperately trying to find an aunt she thought could help her.

Her aunt immediately put Eman to work as a domestic servant – carrying water, preparing food for the aunt's family (which she was not allowed to eat) – and was often slapped when she performed these tasks too slowly. It is Eman's job to cook breakfast for her aunt's children before they go to school; she is not allowed to eat this food (or go to school) and this makes her cry.¹⁷ Whenever coffee pickers are required, she is sent to do this work in addition to domestic chores; her aunt took most of her earnings and abused her on those days when she could not find a job picking coffee. Many coffee farmers underpaid or physically abused her. She does not think she could manage to find her way home to visit her parents.

It might not be unwarranted to see in these aunts’ behaviour a compound of care born of family feeling for a hungry or abused child with a roughly rational stratagem to secure access to very cheap labour. The dividing line between charity and more or less bonded labour is not a sharp one.¹⁸ The evidence suggests that Michelle had no real options, during this phase of her life, in terms of alternative ‘choices’ that might have met her needs.

Nor is it wholly clear what kind of maximising leap or choice was involved when Michelle married her husband at a very young age and subsequently bore ten children, having given birth to her first child when she was about 12 years old. Three of her children died by the time they were two years old.¹⁹ Michelle was clear in the interview that she has given birth to many more children than she would have chosen; she would rather have stopped at five, but her preferences counted for nothing.

In 2009, a brother-in-law sent some cattle to be looked after by Michelle and her husband. This brother-in-law was involved in a serious conflict. His maize was cut and his coffee and banana crops were razed. He managed to save his cows by sending them to Michelle, but she had never been clear what sort of compensation she might get for looking after the cows. None of them produced any milk and the brother-in-law came one day and took away one cow to sell. The best that can be said is that she can use the cow dung as fertilizer and even sometimes sells a bit of dung to help pay the school expenses for her younger children. Again, violence, acts of generosity, ruses to protect land and cattle, and ways of snagging a vulnerable woman into a quasi-wage relation with extremely low payments in kind are jumbled up. The distances between kinship, the ‘moral economy’, and capitalist exploitation collapse in these relations.

Michelle needs to do as much casual wage work as she can. She desperately needs cash because her husband will not work for wages, or on any of the small plots where Michelle is permitted by relatives and neighbours to grow food. She said that perhaps there was something wrong with him: sometimes he sat for hours doing nothing, sometimes he locked himself in the house for hours at a time. (Depression or mental disorder, combined with the repertoires of patriarchal power, may be more relevant concepts than threat points and non-cooperative bargaining in explaining the allocation of labour in Michelle's household.) He does seem able to rouse himself now and then to insist on sexual relations, a site of struggle in the household.

Michelle felt she had no choice in the matter of sex and childbearing; whether ‘rational’ or not, she is wary of family planning and has heard rumours about the bad effects of contraceptive devices and injections. Rumours and propaganda about the dangers of contraception have been promoted by American fundamentalist churches and USAID and by the Ugandan President's wife (Human Rights Watch 2005). This partly explains Michelle’s confusion as well as rural Uganda's exceptionally high fertility and low rate of contraceptive use – in absolute terms and compared to Ethiopia – although the violent disapproval of contraception by male partners is also common (Nalwadda et al. 2010, 7).²⁰ In the domain of sexual reproduction Michelle was unable to insist on acting to improve her health and welfare.

The detailed testimony of life’s work interviewees suggests that rational choice and methodological individualism (of whatever variety, Udehn 2002) can only illuminate a small corner of the social reality that researchers and policy makers are interested in. And the evidence shows that the focus ought to switch away from this narrow and uninteresting corner, turning to far greater emphasis on direct and indirect coercion; the constraints imposed by where and to whom people are born; and the institutional and government policy variations that influence the prospects for survival and welfare and that define the extremely limited scope for exerting any agency.

These issues remain unresolved in newer variants of the mainstream household model where, for instance, migration is regarded as a household decision to diversify its income portfolio, as a decision to try to hedge against risk and to ease liquidity constraints (Molini, Pavelesku, and Ranzani 2016, 3). This model of decision-making is taken to be ‘the industry standard’ (Kevane 2004, 37) and certainly ‘good enough’ as a useful benchmark against which to compare real outcomes.²¹

Other analytical work in mainstream micro-economics has moved away from the assumption of a ‘unitary household’ that behaves as if it is pooling or has entirely harmonious multiple individual utility functions. One well-known approach proposes that households engage in a form of Nash cooperative bargaining. In these bargaining models, it may be assumed that resources in the household are allocated efficiently. But there remain analytical anxieties for the conscientious economist confronted with evidence. ‘If the efficient household model cannot adequately account for the intra-household allocation of resources, it appears that it will be necessary to move towards more detailed, culturally and institutionally informed non-cooperative models of the interaction between household members’ (Bardhan and Udry 1999, 17–18). Unfortunately, important features of cultural and institutional contexts cannot easily be quantified and are too often ignored.

For example, a model of intra-household bargaining in rural Ethiopia found that: ‘There does not appear to be a strong systematic association between bargaining variables and many dimensions of household welfare’ (Fafchamps, Kebede, and Quisumbing 2009, 595). One important problem is that these economists have defined bargaining power and women's welfare in such a way as to exclude spousal violence, in contexts where the prevalence of physical violence and threats of violence from intimate partners are known to be remarkably high (Heise and Kotsadam 2015; Abebe Abate, Admassu Wossen, and Tilahun Degfie 2016). Our findings from life’s work interviews suggest that the problem for economics is that usefully simplifying models and ‘heuristic devices’ become locked in – largely thanks to scientific pretentions – and prevent further insight beyond their assumptions and logic.

Family crisis and labour market participation: ‘blood is thicker than theory’²²

The aspirations of poor rural women are frequently thwarted. Their struggles to survive, to secure enough food for their children or to meet the costs of education and the financial shocks of attending health clinics, are prey to the whims of males and the discretion of devious employers, as well as to the predations of political henchmen. In analyses of rural labour conditions, the ugly realities lived by many girls and women are too often masked by euphemistic references to ‘prevailing social norms’ or a ‘moral economy’. These women are especially sensitive to family bereavement and loss, above all to the death or illness of fathers and husbands or simply to being abandoned by them. And these factors again and again account in the life’s work interviews for why, and when, women enter the labour market despite regarding manual labour as demeaning and very much a last resort. These are not ‘contextual’ or peripheral factors secondary to individuals’ specific decisions and ‘choices’ but central features of their actions (Granovetter 1985, 504–505): the social is fundamental to individual behaviour, the non-market is key to (labour) market participation; coercion and catastrophe trump maximising rational selection among alternatives.

The death of a husband can pitch a family into such immediate poverty that a woman and her young children have to seek casual labour where she may previously have resisted this. Our quantitative survey in Ethiopia and Uganda found that women living in ‘female dominated’ households, i.e. in households where more than 75 per cent of adults were female, were particularly likely to end up in the most degrading and despised types of waged jobs, including: weeding, digging, slashing, hoeing, chopping, mulching, ploughing, planting, picking, plucking, harvesting; carrying water, firewood, charcoal, stones and bricks, quarrying stone, or working as a domestic servant, maid, cleaner, or child-minder for a local private household(s).

Although the participation of both young men and women in wage employment in urban areas in northern Uganda has been shown to be precipitated by family strain and breakdown (Mallett, Attim, and Opio 2016, 11), girls and women do seem to be particularly vulnerable. One of the most common reasons cited by young women who had migrated to Addis Ababa as adolescents in de Regt’s (2016) research was abuse, often at home and linked to family upheavals.²³ In-laws and step parents (what Jones et al. 2015, refer to as ‘blended families’) were in de Regt’s research and in our own a particularly common source of abusive upbringing.

Time and again in dozens of life’s work interviews, family crisis (often the health of fathers or husbands) pushed very young women into wage work.²⁴ For example, Namuddu, from southwestern Uganda, suffered two shoves into wage employment. First, when she was 12 years old her alcoholic father was no longer able to support the family.²⁵ More recently, Namuddu had to take up wage employment again when her husband died, and she needed to find income to keep her children in school.

Schooling and working

Young girls want to stay in school but are very often prevented by urgent needs arising from the poverty, divorce, illness, or death of a parent.²⁶ A particularly affecting version of an all-too-common trajectory is that of Beza, an Ethiopian child labourer from Sidama region. She thinks she is 15 and she is enrolled in Grade 3. One of eight children, she takes quite a lot of responsibility for her two younger siblings. Her progress at school has often been interrupted. She repeated Grade 1 once, having been expelled when her father could not afford the exercise books; and Grade 2 twice – she dropped out of Grade 2 when her mother found casual work uprooting and processing enset (false banana) a two hour walk away, so Beza had to stay home and look after the smaller children. She has done wage work on her own though, starting when she was 13 or 14 sorting coffee beans at the Fairtrade certified coffee washing station.²⁷ The guards at the washing station never ask questions about the age of those seeking work: they do usually insist on a 10 birr bribe, but Beza managed to avoid that because the guard was a relative.²⁸ In our Ethiopian coffee sample, 40 per cent of those respondents seeking work within the past 12 months had relied on the help of friends and relatives. The quantitative survey did not, however, provide much information on the sexual favours that guards, foremen and others might demand in exchange for an employment offer or placement in less strenuous work – these important aspects of the real functioning of labour markets could only be uncovered through painstaking life’s work histories. These demands, as well as demands for bribes, are routinely ignored in the labour market research published by mainstream economists such as Blattman and Dercon (2018).

Beza’s family would like their children to migrate. When her older sister left school after Grade 1 they leapt at an offer for her to work as a nanny for the landlord of a relative living in Hawassa. For three years the sister has been doing this job and has never been able to make the 40 km trip home to see her family. She receives no wage for the work (but she does get food) and Beza has heard that the landlord has allowed her sister to start school again. Beza would like a similar position.²⁹ The main appeal is that she would eat better: Beza has not eaten meat for about a year, and she said that she had ‘never been full’ in her life.

The Young Lives Survey in Ethiopia found that children stop attending school and enter the labour market prematurely as a result of parental illness/absence, or following a divorce (Chuta 2014, 5). Not only the life histories, but also our quantitative survey data alert us to the consequences of absence or loss of males in the family in rural Ethiopia and Uganda: in the ‘female-dominated’ households, school dropout rates are significantly higher than in other households.

Conclusion

We argue that a more realistic analysis of rural inequalities and of labour markets is a pre-condition for a discussion of policies relevant for the poorest. It matters that the lives led and work often unwillingly done by these rural women and children are characterised more by violent struggle than by cooperative bargaining, by dissatisfaction and flux rather than equilibrium, by desperate survivalism (euphemistically labelled agency) rather than maximisation, and that the weight of wider social forces, including state interventions to meet or refuse to meet contraceptive or other health needs, household dynamics and structured relations of power dominate the scope for exercising their agency. Arguably, methodological individualism does the very subjectivities of such women a disservice; notions of bargaining equilibria being settled within households draw a veil – of ignorance – over the forced acceptance of other people’s pursuit of pleasure and wealth. And the conceptual heuristic devices of neo-classical economics have shown that they cannot escape their axiomatic foundations and are not fit for the purpose of reaching a more realistic understanding of these women’s lives.

These rural women have fundamental needs and usually cannot hold out for a better price (wage) to come along; they have barely any bargaining power in all the labour markets they can enter. This is a function of the fundamental difference between labour and other marketed commodities; it is compounded by structured gender relations and by widespread repression and abuse of girls and young women (Marshall, Lyytikainen, and Jones 2016; Semahegn and Mengistie 2015; Bantebya, Muhanguzi, and Watson 2014), in contexts where men have almost complete control of both reproductive rights and violence rights.

It will not be easy (and it may not immediately be politically feasible) to combat the distressing labour market outcomes described in these interviews. Collective action to reduce child labour and to improve wages and working conditions in rural areas of Ethiopia and Uganda is probably off the agenda, because ruling parties are repressing independent organisational initiatives (Abbink 2017, 320; Ulandssekretariatet 2016). Trade unions recruiting farm workers receive very little international support. This is reflected in the fact that real wages have been declining in recent years for most of our survey respondents in rural Ethiopia and Uganda, with nominal wages barely rising if at all and food prices rising sharply, as shown in the data from re-surveys and in the life history interviews.³⁰

There are two broad areas where government policy could improve labour market outcomes. First, it is possible to make children, especially girls, stay in school much longer than they typically do. If public investments are made to ensure that children do not drop out from school, there will be direct benefits to young people themselves. Girls and young women need appropriate facilities in schools and they are believed to face dangers when walking long distances to attend secondary schools; there is considerable scope for investment in sanitation and transport, as well as in the monitoring of regulations requiring school attendance. Such investments may result in at least moderately better skills, including verbal ability and numeracy (Woldehanna and Gebremedhin 2015) and, as a result, women may have greater bargaining power to negotiate working conditions in labour markets. There are also indirect benefits. If fewer very young, child and school age girls supply labour to export agriculture (among other activities) there can be upward pressure on real wages. The best evidence of the success of interventions to tighten rural labour markets – resulting in an impressive rise in real agricultural wages over the last two decades – is in rural India (Jacoby and Dasgupta 2015). In rural Ethiopia, there is some evidence suggesting that where real rural wages are very low, then there is a particularly strong correlation between increases in wages and declines in poverty (Bachewe et al. 2016, 19).

Second, there is a case for a sharper policy and public expenditure focus on those areas with the best potential for a rapid rate of growth of wage employment opportunities for women with low levels of schooling: agricultural exports have potential to absorb the female labour supply of the most deprived households, typically greater than do crops grown for domestic consumption (Cramer and Sender 2015; Van den Broek and Maertens 2017; Cramer, Sender, and Oqubay 2020, Chapter 8).

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by The Leverhulme Trust: [Grant Number EM-2015-009 Emeritus Fellowship].

Notes on contributors

John Sender

John Sender Emeritus Professor of Economics at the School of Oriental and African Studies (SOAS), University of London. Past appointments include: Director of the African Studies Centre, University of Cambridge and Visiting Professor of Political Economy, University of the Witwatersrand.

Christopher Cramer

Christopher Cramer Professor of the Political Economy of Development, SOAS, University of London. Distinguished Visiting Professor, University of Johannesburg. Chair of the Scientific Committee of the African Programme on Rethinking Development Economics (APORDE).

Notes

1 Sender, Oya, and Cramer (2006) and Oya and Sender (2009) used a similar research design in a large rural labour market survey in Mozambique. Stevano (2019) did related research on women, choice, and work in another area of Mozambique.

2 Other social scientists avoid the problem by ignoring the local prevalence of Intimate Partner Violence when analysing processes of household decision-making (Bernard et al. 2020; Agence Nationale de la Statistique et de la Démographie (ANSD) [Sénégal], et ICF, 2018: 371).

3 In contrast, there has been relatively little research into how poor rural women benefit from trade unions (Anner 2021).

4 A non-cooperative bargaining model has also been used to explain the ‘inefficient’ or irrational allocation of fertilizers to different household members managing land in West Africa (Haider, Smale, and Theriault 2018). In Malawi, non-cooperative bargaining models have been used to analyse how bargaining can lead to ‘inefficient’ allocations of labour but in this analysis the researchers fail to discuss Intimate Partner Violence (Walther 2018).

5 It is by no means uncommon for girls in rural Ethiopia to be married to men who are 10–20 year older. ‘This age gap can result in a lifetime of constrained agency and violence … (Jones et al. 2020, 9).

6 Atnafu, Oucho, and Zeitlyn (2014) and de Regt (2016) focus on the large number of rural Ethiopians migrating to towns in search of wage employment.

7 We carried out a small survey of women migrating to the Middle East, interviewing them at the airport after they had checked in, but our enumerators regularly found groups of women outside the airport having been denied check-in or having been deported back to Ethiopia – these penniless young women often had no idea how to get home.

8 Research on young female migrants to much smaller towns in Southern Ethiopia suggests that they too are often deceived by brokers (Alemayehu and Fekadu 2019, 66).

9 In Southern Africa, young Tswana women attempting to escape from the violence inflicted by parents and spouses were sometimes sufficiently desperate to walk distances of 300–400 kilometres in search of wage employment (Cockerton 2018, 3).

10 The survey also captured large number of migrants in our other research sites – Jimma and Mubende. In these coffee growing areas the majority of the 'less bad' jobs were also held by non-migrants.

11 The terms ‘fantasy’ and ‘dream’ are not intended to be derogatory. The process of supplying labour is fraught with uncertainty and when rural women plunge into new forms of employment they need a boost from ‘animal spirits’: ‘individuals’ psychological features and emotions play an important role in labor-supply decisions—as happens, for example, with investment decisions’ (Fernández-Huerga, García-Arias, and Salvador 2017, 591–2).

12 The methods used to assess degrees of deprivation in rural Africa are discussed in Sender, Cramer, and Oya 2018.

13 ‘What keeps them in the city is not fear but pride: to return home early is to admit defeat’ (Chang 2010, loc.155). Migrant sex workers in Nazreth, Ethiopia, feel isolated, but are too ashamed to return to visit their villages (van Blerk 2011, 228).

14 Confronted with behaviour like Michelle's that fails to conform to the simple reductionist models, some neo-classical economists resort to ad hoc explanatory narratives, such as the story that African households or individuals hold fatalistic beliefs that prevent them from demanding, for instance, an optimal amount of credit (Bernard, Dercon, and Taffesse 2011).

15 For further examples of mistreatment by stepmothers in rural Uganda see Zalwango et al. 2010 and Bantebya, Muhanguzi, and Watson (2014) and for Ethiopia, Erulka, Medhin, and Negeri 2017.

16 For an Ethiopian example of an intervention by an aunt to enable some schooling see Chuta and Morrow 2015, 7.

17 Near to Eman’s village, other young female domestic workers report that ‘slapping, pulling off their ears, chocking, kicking by foot, punching, beating with shoes, splashing with water, knocking them against a wall and pinching are the most common forms of physical abuses that they frequently encounter’ (Alemayehu and Fekadu 2019, 64).

18 A young Ethiopian boy interviewed in Menagesha had been tricked into migrating from his home hundreds of kilometres away by the promise of waged employment that would enable him to purchase a school uniform and be admitted to Grade III of primary school. His employer neither fed nor paid him. He ran away, fell ill, but was rescued from the street and restored to health by a man for whom he has continued to work for several years. His obligations to this man’s family mean that he cannot bargain for a wage.

19 Epidemiological literature has identified ‘an increased risk of low birthweight, preterm birth, stunting at 2 years, failure to complete secondary schooling, and lower adult height in children of young mothers (≤19 years) compared with mothers aged 20–24 years’ (Saloojee and Coovadia, 2015: e342).

20 There appears to be a strong association between child marriage in Uganda and intimate partner violence (Kidman 2016, 7). Analysis of our quantitative survey data confirms the relatively high prevalence of adolescent pregnancy in rural Uganda and that adolescent mothers are much more likely than other mothers to be deprived and vulnerable (Cramer et al. 2017).

21 Similarly, Bardhan and Udry (1999, 5) suggest that their ‘adherence to the principle of maximisation … should be regarded more as a crude heuristic device than as a definitive statement on human behavioural regularity’.

22 Laurie Lee, As I walked out one midsummer morning, London: Penguin, 2014 edition.

23 Turan et al. (2016) describe the role of intimate partner violence in explaining female migration within Kenya.

24 Families where the father had died or disappeared were, during the industrial revolution in England, a major source of child labour including the very young (Humphries 2013, 413).

25 In Mallett, Attim, and Opio’s (2016) Ugandan survey, 39 per cent of respondents reported growing up with a parent with severe alcohol problems.

26 Orkin (2011) found many children engaged in casual paid employment in the village an hour from Ziway; most of these children combined paid work with school attendance, though often missing or late for school, rather than dropping out. Other studies, also using Young Lives survey data in Ethiopia, have shown the links between paternal death and labour market entry (Himaz 2013, 15). 

27 There is a long and well-documented history of rural wage work in Sidama; and many local people have learned to regard all casual agricultural labourers with contempt (Negash 2017, 46).

28 One national survey in Uganda found that 2.75 million children aged between 5 and 17 were engaged in economic activities, around half of them engaged in what is referred to as ‘hazardous work’, involving long hours in mines, construction sites, agriculture, factories and quarries (Uganda Bureau of Statistics 2010, 136).

29 On the role of older sisters in the recruitment of children as migrant domestic workers, see Alemayehu and Fekadu (2019, 62).

30 We asked a smaller number (401) of the original respondents to repeat the questionnaire/interview. These re-surveys took place one or two years later, varying across the two countries’ samples.