The Women’s Empowerment in Agriculture Index (WEAI) was launched in 2012 as a tool to measure women’s inclusion and agency in the agricultural sector. Funded by the United States’ Feed the Future Initiative, this household survey instrument scores women’s level of decision-making and control, using ten indicators across five domains: production, resources, income, leadership, and time (Alkire et al. Reference Alkire, Meinzen-Dick, Peterman, Quisumbing, Seymour and Vaz2012). In subsequent years, the WEAI was trialed, refined, and revised, leading to both an updated version of the original tool and the Abbreviated-WEAI (A-WEAI), released in 2016, which was designed to address concerns that the original survey was too long for respondents (Malapit et al. Reference Malapit, Pinkstaff, Sproule, Kovarik, Quisumbing and Meinzen-Dick2017). The A-WEAI is anchored by the same five domains but comprises only six indicators, in order to provide a survey instrument that is faster to implement. Given the growing interest of bilateral and multilateral development agencies in the question of women’s empowerment, the tool’s purported ability to provide a universal measurement of empowerment has made the WEAI and its variants increasingly popular with researchers and development organizations; since its launch, nearly one hundred organizations in 53 countries have utilized the tool or an adapted version (Malapit et al. Reference Malapit, Quisumbing, Meinzen-Dick, Seymour, Martinez, Heckert and Yount2019).
As the WEAI’s popularity grew, so too did the criticism leveled against it. Most centers on the assumptions that are drawn about women’s empowerment and autonomy and control over productive resources. S. Akter and colleagues (Reference Akter, Rutsaert, Luis, Htwe, San, Raharjo and Pustika2017) point out that the A-WEAI is derived from data about rural gender relations in some key target countries which may not always transplant well into others. In the same vein, S. Gupta and associates (Reference Gupta, Vemireddy, Singh and Pingali2019) observe that the A-WEAI is not context specific and needs to be adapted to match local circumstances. C. R. Farnworth and colleagues (Reference Farnworth, Rozel, Stirlinga, Chinyophiro, Namakhoma and Morahan2018) argue that the A-WEAI’s conception of empowerment is premised on a male-female dichotomy that neglects the importance of collective and intra-household decision-making (see also Doss et al. Reference Doss, Kieran and Kilic2017). Corey O’Hara and Floraine Clement (Reference O’Hara and Clement2019) note that A-WEAI measures only those forms of empowerment of which respondents are self aware; it neglects the question of whether or not respondents possess critical consciousness. Robin Richardson (Reference Richardson2018) critiques the WEAI for using a universal “cut point” derived from how respondents score on the survey which classifies them as empowered or disempowered, yet does not take into account women’s own perspectives or understanding of their situation.
While the critiques cited above do question the conceptual underpinnings of the tool, they are basically concerned with how shortcomings of the survey can be addressed, whether through adding new variables or indicators to the scoring formula, or by combining the survey with other qualitative methodologies. By contrast, in this commentary we offer a more fundamental critique of the A-WEAI. In our view, the underlying assumption of the survey—that it is possible to obtain an accurate measurement of women’s empowerment using a rapid survey and quantifiable data—reinforces what Andrea Cornwall (Reference Cornwall2018) calls “neoliberal empowerment.” By this, she refers to the way in which women’s empowerment has been appropriated by the development industry and has emerged as a key buzzword with little connection to its orginal transformative meaning. This process has been driven by a widely held belief that enhancing women’s autonomy and decision-making power will automatically lead to increased economic growth and wider societal benefits, despite several critiques that question this instrumental assumption (O’Laughlin Reference O’Laughlin2007; Kabeer & Natali Reference Kabeer and Natali2013; Stevano Reference Stevano2019). By revealing key flaws in the A-WEAI, our goal is to contribute to the critical literature on empowerment which seeks to reclaim the concept and its original emphasis on challenging power relations that uphold or justify social inequality (Batliwala Reference Batliwala2007; Cornwall & Rivas Reference Cornwall and Rivas2015). While this literature provides a rich historical and theoretical critique of neoliberal empowerment, it has yet to provide a fine-grained analysis of the A-WEAI. This is an important gap, because the instrument has proven to be so influential. Continued, uncritical use of A-WEAI may divert attention from other methods and activities that can promote collective action and consciousness-raising needed to foster real empowerment.
The roots of the empowerment concept can be traced to feminist organizing and popular education in the Global South in the 1980s and 1990s. For early pioneers of the concept (Batliwala Reference Batliwala1993; Kabeer Reference Kabeer1994, Reference Kabeer1999), women’s empowerment necessarily entailed challenging hierarchies and cultural norms that justified gender inequality, stressing the importance of collective action in enhancing women’s ability to make strategic life choices, and understanding empowerment as an ongoing and culturally varied process—not a fixed state or measurable outcome. Yet, this emphasis on challenging oppressive norms and attitudes, employing collective action, and sensitivity to specific cultural contexts was diluted as the concept became “mainstreamed” within development institutions and practices and operationalized in quantitative surveys seeking a universal measurement such as the WEAI. Andrea Cornwall (Reference Cornwall2018) uses the term “empowerment lite” to describe the contemporary meaning and use of the concept, in that empowerment has become apolitical and non-threatening, no longer signifying changes in power or hierarchical relations (see Batliwala Reference Batliwala2007). Similarly, Cecilia Sardenberg (Reference Sardenberg2008) calls it “liberal empowerment,” implying that the emphasis is upon individual improvement through entrepreneurialism, particularly with reference to micro-credit. Cornwall and Althea-Maria Rivas argue that the concept rests on a simplified understanding of gender relations that overdetermine other aspects of women’s lives, such that “intimacy, cooperation, mutuality, come to be shrouded in images of the irresponsible and sexually voracious male, and the long suffering and victimized female” (Reference Cornwall and Rivas2015:403).
Our commentary contributes to this critique of neoliberal empowerment by revealing three foundational problems with the A-WEAI. First, the instrument itself relies on categories and codes that lead to distorted representations of intra-household relations. Second, the A-WEAI is unable to account for the diverse circumstances facing farming households across Africa. The final limitation concerns the individualistic nature of the tool and its inability to account for relational dynamics.
Our analysis is based on our experience applying and testing the A-WEAI in Ghana, Kenya, South Africa, and Uganda, as part of a larger research project investigating the impact of new improved crop varieties on intra-household gender relations. Our team purposefully selected five households in each country to test the A-WEAI, for a total of twenty (20) households. The households in each country case study were selected to convey maximum variation, and respondents were recruited through pre-existing connections established by the researchers and their partner organizations. These households reflect demographic, socio-economic, and agricultural characteristics that are similar to the other households in the regions under study. In each country, we observed the A-WEAI being implemented by two enumerators for each household, one for the male and one for the female household members. Enumerators reviewed and practiced using the A-WEAI questions before implementation, using the A-WEAI instruction guide and survey available from IFPRI’s website. The enumerators translated the questions into the relevant local dialect before implementing the survey to ensure consistency in terminology. After the survey was implemented, we debriefed the experience with both enumerators and participants to ask follow-up questions and clarify issues of interpretation, for example, why and when was it difficult to answer a question.
Each country’s case study presented a distinct context that conditioned responses to the A-WEAI; as Andrea Cornwall and Jenny Edwards note, “Context is crucial in making sense about empowerment” (Reference Cornwall and Edwards2010:2). In South Africa, our research was carried out with farm workers and dwellers in the Eastern Cape. These respondents reside on land and in houses owned by their employer, the farm owner. The respondents have access to land for growing subsistence crops and keeping a small number of animals, though these entitlements are often contingent on at least one member of the household maintaining employment status with the farm owner. The Ugandan participants hailed from Nakaseke District in the central Bugandan kingdom. Most of these farmers are over the age of fifty and reside in male-headed households on plots of between five and ten acres, where they cultivate a mix of crops alongside the primary staple, matooke banana. They are generally landowners or reside on family land and polygamy is common. Farmers in Kenya reside in Muranga County, one of five counties in Kenya’s central region. All of these farmers produce mixed crops, including tea, though maize remains the most common staple under production. The land cultivated by farmers in this region is under three acres; farms are both independently and family owned. In Ghana, we selected a farming community in the Wa East district of the Upper West Region. Farmers in this region live in extended family households and farm on family plots. Staple crops include cowpea, maize, and groundnuts. Farmers also keep livestock in addition to their agricultural activities. While farming plots were located further away from the community, it is common to also have food gardening plots nearer to the household. This community, like others in the northern sector of Ghana, is patrilineal.
In this commentary, we examine each module of the A-WEAI, identifying areas of confusion and problematic assumptions. We conclude by offering reflections on creating a more comprehensive and accurate assessment of the relationship between women’s empowerment and agriculture.
Household Decision-Making and Income Generation: Module G2
After the initial process of individual identification, the first substantive section of the A-WEAI, Module G2, lists a series of livelihood activities and asks respondents to report the extent of their involvement in decision-making for each one. Respondents are asked to rank how much input they have in decisions regarding the stated activity (G2.03), to what extent they can make personal decisions (G2.04), and how much input they have regarding income derived from those activities (G2.05).
The first two activities focus on the farming of food crops (Activity A) and cash crops (Activity B). While this distinction between food and cash crops makes sense as a starting point, given the historical pattern in many parts of Africa where men have more responsibility over crops grown for market (Dolan Reference Dolan2001), our experience with the A-WEAI suggests that a hard delineation between these two categories is misleading. In Uganda, for example, gender dynamics vary, depending on the crop and the plot. One female coffee farmer explained that she holds a high degree of personal decision-making power over her own coffee plot but only a small degree of input on her husband’s plot. We also observed that the delineation between food and cash crops was fluid; one husband explained that he only considered coffee a cash crop, while his wife considered matooke banana, maize, groundnuts, and beans cash crops because they sold the excess at the local market. This trend persisted throughout all of our Ugandan households. Men and women understood the distinction between food and cash crops differently, which invalidated the comparability of their answers. Delineating all crops and plots across a single axis (food vs. cash crops) collapses distinct dynamics into a single value that provides a homogenized and distorted representation of women’s participation, decision-making, and control over all crop types.
Female respondents in South Africa often identified themselves as the primary decision-makers for both food and cash crops, including the use of income from crops sold. Husbands typically corroborated this pattern, claiming that they were little involved in crop production and selling, due to the fact they were employed on commercial farms. The survey thus gave an impression of strong female empowerment with respect to these two activities. However, further questioning undertaken after the survey was completed revealed that the food crops grown by these families were grains and vegetables, grown on small garden plots adjacent to their houses. In South Africa, “cash crops” were simply the small surpluses of these crops, representing a relatively minor aspect of the overall household income. Thus, while small-scale agriculture provided some supplemental food and income, it was less central to these farm worker households than the wages earned on the farm or from social grants.
Two other problems relating to decision-making were revealed in Kenya. In two cases, the husband and wife comprised one household amid several related families living on a larger family compound. The land and farm were owned by the husband’s father. Cash and food crop production were a family activity and outputs were shared by the family, though the ultimate decisions were made by the owner of the land and farm, the eldest male. Thus, questions about decision-making and participation were actually a reflection of what was allocated to the household, and not a function of what the individual household members had produced or contributed.
Activity Codes G and H ask these same questions concerning participation, decision-making, and input within major and minor household expenditures made in the last twelve months. One major source of confusion for respondents here was that the category of expenditure failed to differentiate between those made for the individual use versus those made for the household; empowerment in such cases is therefore contingent on the purpose for which the money is being used. Decision-making processes are also more complex in extended and polygynous family households. Asking individuals about decision-making in situations where such decisions would be made collectively was awkward. In South Africa, men explained that they can borrow money without consulting their wives if the purchase is something just for them, but if they are borrowing for a family-related expense, then they consult with their wives. In Ghana, one male respondent reported borrowing money to pay for school fees, with the understanding that this was joint debt borne with his wife. In Kenya, men differentiated between selling a bicycle or motorcycle, which was their property and did not require consultation, as opposed to decisions regarding agricultural inputs, which needed to be made jointly with their spouses. One Ugandan farmer underscored this distinction using the example of selling a cow: if the cow were being sold for school fees—a joint expense—then he would consult with his wife, but if the cow were being sold to buy a bicycle—an individual expense—then he would make this decision on his own. Our experience suggests that confusion between individual and collective senses of responsibility leads to confusion and inconsistency in responses. This paralleled the difficulty in documenting access and ownership to productive capital.
Access to Productive Capital: Module G3 (A)
Module G3 (A) lists a range of assets and asks if anyone in the household owns them (3G.01). It then asks the respondent if the asset is owned solely, jointly, or not at all (G3.02). Generally speaking, most respondents understood the concept of ownership in terms of documentation, the right to sell, and full access. But these three dimensions are not captured by the options presented in the survey. The problems we encountered here were not so much in the definition of the assets (which are more specific than the livelihood activities listed in G2), but rather in the delineation of ownership, specifically around the designation of “joint.” In each country, participants often claimed that their assets were jointly owned. Yet, what they implied by “joint” varied considerably, and the survey fails to capture this nuance and variation.
The designation of joint ownership can overstate the degree to which women can access certain resources. For example, a woman in South Africa stated that she jointly owned the family vehicle with her husband, despite the fact that the husband was the only driver. To illustrate her joint ownership, she recalled how on a recent trip she made to town, her husband gave her money to pay a mechanic for repair work which had been recently completed on the vehicle. That she was given the responsibility to pay the mechanic was, for her, evidence of joint ownership of the car. While this example suggests some degree of communication and cooperation around the management of the vehicle, one could equally interpret it as substantiating a picture of hierarchical gender relations, in which the woman is merely delegated a task (paying the mechanic) by her husband. Labeling these relations as “joint” is problematic because the A-WEAI scores sole ownership and joint ownership the same way; both are counted as meeting the threshold of empowerment. The survey scores this woman as if she were the sole owner of the car, when in reality, ownership means different things for her than what is intended by the survey. This parallels other assets. In Kenya, one household noted joint ownership of a television. But the husband was clear that he had the final say over what was going to be viewed on the TV, and his preference dominated. Thus, the TV was a joint asset, and decision-making about selling that asset was argued to be joint, but use of the asset was hierarchical. In Ghana, respondents expressed discomfort in discussing ownership, because assets were family assets, sometimes passed down from older generations for use of the whole family (which comprises multiple nuclear family units).
While respondents often implied, through the “joint” designation, that assets were owned collectively with the spouse and/or other family members, it sometimes included other authorities. This was particularly evident with reference to land. Farm workers in South Africa often said they owned land jointly, but they were referring here to a sense of co-ownership between themselves and the official landowner (a white farm owner). The “joint” designation not only fails to specify the nature of these relationships, it also disguises how women may be subordinated within them. In the case of South Africa, the jointly held land to which farm workers referred actually entailed a weaker land entitlement for women. As mentioned above, the husbands were the ones employed on the farm. A wife’s access to land and housing was contingent upon the husband remaining employed; in the event of his death or divorce, the wife could be evicted from the property. This weaker attachment to land was conveyed to us in the different responses by one couple in module G3. While both husband and wife stated that their land was owned jointly, only the man said that their house is owned jointly (implying again shared ownership with the white farm owner). For her part, the wife claimed no ownership of the house, stating flatly “it is owned by white farmer, we are just staying there.”
The narrow depiction of land ownership presented similar complications in other countries as well. In Ghana, participants hesitated to comment on the question of land ownership. This is because of the clear delineation of people as either settlers or Tendaanba (landlords/land custodians). Only people from Tendaanba clans can claim “ownership” over land. The Tendaanba, as the original inhabitants and owners of the land, have socio-religious jurisdictions over vast territories and communities, but they are not present and/or involved in the day-to-day administration of these places. Even then, preferred discourse revolves around custodianship. Subsequently, while men could speak about living and working on family lands, the women we interviewed found this question altogether inappropriate. With some clarification, and often apologies, the women spoke about working on family plots that belong to or had belonged to the male spouse’s family. Generally, while participants are unlikely to ever be evicted from their farms or homelands, they cannot sell any land or convert farming land into residential plots without consultation with the Tendaanba. In any case, land granted to settling clans is only for subsistence purposes. In Uganda, respondents explained patiently that different ownership regimes related to different plots. If the husband and wife both contributed to the purchase of the plot, then the ownership would be joint, but if the savings were the man’s alone, he considered himself the sole owner. Enumerators were forced to select a single answer to encapsulate these contradictory arrangements.
Other instances of productive capital listed in Module G3(A) produced results that were similarly contested. In South Africa, male respondents would indicate “joint” ownership of wages, but their wives indicated feeling no ownership over their husband’s wages, although they would attest that they were allowed to buy groceries with it. In Uganda, it was the question over tools that generated the most confusion. Most households owned multiple pangas (machetes), with one that belongs to the husband that no one else is allowed to touch, and others that are owned jointly by all household members. Again, the enumerator is forced to collapse distinct regimes of ownership into a single response.
Ownership regimes are multidimensional, often overlapping, and subject to a range of social norms; as a result, they are difficult to capture in a simple survey (Doss et al. Reference Doss, Kieran and Kilic2017). Augmenting this challenge is the manner in which respondents interpret ownership; as Krista Jacobs and Aslihan Kes (Reference Jacobs and Kes2015) discuss, perceived ownership and joint ownership often vary depending on whom you ask. More to the point, as the examples above illustrate, perceived ownership does not necessarily equate to equal decision-making powers or legal rights. The take-away here is that the A-WEAI survey misses the complex and specific ways in which people exercise control over resources.
Access to Credit: Module G3 (B)
Module G3B asks farmers about their decision-making regarding borrowing over the past twelve months. This tool goes through a range of different lenders, including non-governmental organizations, formal lenders such as banks, informal lenders such as friends and relatives, and formal/informal group-based lending programs. Our experience suggests that many forms of credit are excluded by this list. For example, in South Africa, the primary lender of relevance for the respondents was their employer (a white farm owner). The enumerator chose to list this as an “informal lender” on the survey, since the option of employer is not listed. However, the money these respondents receive is in the form of a cash advance—money paid out to them in advance of payday, which would later be deducted from their salary. Only men reported having access to this form of credit. The survey would thus score women as disempowered for this section, even though men would say that they consult with their wives regarding the use of the income. In Uganda, a follow-up conversation with one female farmer revealed that she had participated in an arrangement whereby she received a heifer and was required to return a pregnant heifer to the lending group as repayment. When asked why she chose not to disclose this within the tool itself, the respondent replied that she understood the question to be focused only on cash loans and not in-kind loans. This limitation is noted explicitly within the tool itself; there is a footnote attached to the list of lending sources instructing enumerators to give “locally relevant examples” (Feed the Future 2015:6). But the rigid nature of the tool and the accompanying time restrictions leave no room to capture these outliers.
Other dynamics are similarly occluded here. In South Africa, some men mentioned that they can borrow money without consulting their family if the purchase is for something just for them, but if they are borrowing for a family-related issue, it requires wider consultation. In Uganda, one farmer reported borrowing money from formal lenders at three different times in the past year. The first time, he did it on his own and did not tell his wife about it; the second time, he decided to tell his wife because the loan was for school fees (a joint expense); and the third time, the husband and wife borrowed together. In this case, the enumerator decided to score the decision-making as undertaken by both the husband and wife, ignoring that the first loan was made without the wife’s knowledge, and thus serving to overestimate the measure of her knowledge and control over these matters.
Time Allocation: Module G4
The time allocation activity (Section G4) seeks to establish how much time is spent on productive and domestic tasks. The respondents are defined as having an adequate workload if the number of hours they spend on work-related tasks (i.e., wage employment, farming, shopping, weaving/sewing, textile care, cooking, domestic work, caring for others, commuting, and traveling) is less than the time poverty line of 10.5 hours (Malapit et al. Reference Malapit, Pinkstaff, Sproule, Kovarik, Quisumbing and Meinzen-Dick2017).
The section on time allocation presented problems in each country. A central issue that cut across each setting was that the respondents, more often women, had difficulty assigning only one activity for each fifteen-minute segment of the day, because they are constantly multi-tasking. For example, the enumerator in South Africa recorded a female farmer in South Africa watching TV for seven hours, but in reality, she completed multiple tasks during that time period, including washing, mending, meal preparation, and caring for children and elders. In Kenya and Ghana, men and women laughed out loud when asked how much time they spent “exercising,” which is one of the time allocation activities. In Kenya, several households were farming on steeply sloped land. This necessitated going up and down routinely throughout the day for farm and household duties, and thus they exercised routinely, but not formally. In Ghana and Kenya, women were more likely to multitask, while men were more likely to be engaged in one activity at a time. Thus, the tool tended to underestimate the number of tasks women are undertaking concurrently. One woman in Ghana talked about being able to take a break for only about an hour during the day (while doing other miscellaneous tasks she does not consider “work”), which illustrates the challenge of breaking down the day into productive and non-productive hours.
At a broader temporal scale, the A-WEAI tool requires respondents to recall their activities over the past twelve months. Farmers struggled with this scope. Most farmers preferred to recall their activities according to a calendar year, forcing the enumerators to amend their records to fit within this more intuitive mode of recollection. Farmers also struggled with the variations contained within this timeframe. In Uganda’s tropical climate, twelve months comprises two wet and two dry seasons. In Kenya, there is both a long and a short rainy season, which are different in length. The measure does not consider variations in activities between dry and wet seasons, and thus the measure of workload and the time allotted for activities is highly dependent on when the enumerator arrived on the farm.
Group Membership: Module G5
In our assessment, this final section of the A-WEAI tool, dedicated to group membership (Section G5), is the most accurate. The section is flexible enough to track the range of groups and associations respondents belonged to. Still, some confusion around what constitutes group membership arose. In Uganda, for example, it was unclear whether membership means that someone pays dues or whether it can include someone who is simply a participant in group activities. In Ghana, farmers felt they were members of a group if their names were written down or if they attended meetings and participated in activities. In South Africa, one participant noted that there is no specific question about political party membership or trade union involvement. This might be because it is a sensitive area of discussion, but as a result the questionnaire potentially misses important ways women exercise agency. In Kenya, political parties were also notably absent from the list of possible farmer group memberships. Also challenging was how to categorize ethnic-specific groups, such as the Council of Kikuyu Elders, which several respondents wanted to be distinct from the other categories. Some minor tweaks to phrasing could be sufficient here, such as expanding the parameters of the question and allowing participants more agency in defining what group membership means in their particular context.
Conclusion
In this commentary, we have outlined several concerns and limitations with the A-WEAI that arose from our experiences testing this tool in four African countries. The first series of limitations relates to the instrument itself and the goal of rapidly administering and completing the survey. The A-WEAI relies on categories and codes that produce distorted representations of intra-household dynamics. We identified a number of these misrepresentations, including the distinction between food versus cash crops, ambiguity around what distinguishes sole versus joint ownership, and assessing control based on a focus on assets versus expenditures. The decision codes themselves present another source of misunderstanding. For example, farmers asked for repeated clarification around the meaning of “to what extent,” forcing enumerators to make difficult and imprecise judgements. Many chose to quantify “extent” in terms of percentages, which collapses the dynamics of both severity and frequency. Restrictive categories lead to restrictive conversations; in our experience the most important gender dynamics were revealed through extended, informal conversations that occurred after the A-WEAI had been completed. There are additional challenges associated with the A-WEAI’s rapid implementation; there is a propensity for rushing (A-WEAI enumerators are instructed to spend a maximum of one hour with each respondent, eliminating the possibility of follow-up questions), an insufficiency with regard to enumerator training (training sessions last one day only), and issues with simultaneous translation (for example, there is no direct translation of “empowerment” in many African languages, forcing enumerators to add their own explanations and elaborations) (see Rizzo et al. Reference Rizzo, Kilama and Wuyts2015 and Cheema et al. Reference Cheema, Mehmood and Khan2018 for similar critiques).
Additionally, the A-WEAI is unable to account for contextual differences and the diverse circumstances facing households across these four countries. The A-WEAI is premised upon comparing responses between one male and one female member of the same household to evaluate gender relations in agriculture. But many of the households we encountered did not fit neatly within this categorization, and none of their decision-making is confined within this single household bubble. For example, follow-up conversations in Uganda and Ghana revealed that a number of participant households were polygamous, which is common within central Uganda and the Wa East district (respectively) where the tool was tested. Thus, the A-WEAI’s emphasis on singling out one male and one female household member completely ignores the important gendered dynamics that shape interactions between senior and younger wives (Nyantakyi-Frimpong Reference Nyantakyi-Frimpong2019). In South Africa, we interviewed farm workers who live on commercial farms full time. They do not have their own land, but they have access to land for growing crops. In Ghana, respondents were mainly subsistence farmers who farm on their own plots and on family farmlands, occasionally engaging in communal farming on a rotational basis. Even if a nuclear family household is physically separated from the main family house, decisions are often made with reference to the larger extended family and the marriage and residential patterns thereof. For instance, a wife of a younger brother would defer to the wife of an older brother in decision-making on the basis of seniority. Many of the Kenyan households who participated were allocated plots of land on family property for the purpose of harvesting cash crops, but with decision-making and revenue shared between several households on the family plot. The goal of the WEAI and its variants is to provide a single aggregate score for a particular region of a country. In doing so, it erases all sorts of context-specific information that is crucial to understanding gender relations in agriculture (Schatz & Williams Reference Schatz and Williams2012; Carletto et al. Reference Carletto, Jolliffe and Banerjee2015).
The final limitation concerns the individualistic nature of the A-WEAI. The scholarship on household bargaining and cooperation has expanded our understanding of the farming household as a unit that incorporates multiple decision-makers (Doss & Quisumbing Reference Doss and Quisumbing2020). But the A-WEAI misses the motivations, structures, and outcomes of household bargaining that are only revealed via interactive exercises (Bernard et al. Reference Bernard, Doss, Hidrobo, Hoel and Kieran2020; Iversen et al. Reference Iversen, Jackson, Kebede, Munro and Verschoor2011; Lecoutere & Jassogne Reference Lecoutere and Jassogne2019). The A-WEAI adopts a collective approach, insofar as it recognizes a single male and a single female decision-maker, but it fails to incorporate any sort of collective exercise where farmers share experiences, learn from one another, or interact. In doing so, the A-WEAI implicitly promotes a model of personhood that is at odds with the more collective values and/or practices and norms that persist in rural Africa. The result is a measure that equates empowerment with individual autonomy but ignores how intra-household relationships determine outcomes, leading to a distorted picture of intra-household dynamics. Many of our respondents were reluctant to overestimate their individual level of control over specific items or issues, relaying that even individual decisions have to be made with the family in mind; they viewed it as morally problematic to act only in one’s self-interest. We believe that relational dynamics, where individuals correct, elaborate, or contradict one another, need to be foregrounded in order to obtain accurate understandings of empowerment in agriculture.
The A-WEAI’s core limitations stem from the assumptions that constitute the tool’s essence: that women’s empowerment in agriculture can be accurately conveyed through a large, rapid, assessment of households using quantifiable data. While this approach may provide intriguing, generalizable snapshots of women’s empowerment in agriculture and can promote opportunities for regional or national comparison, the complexities of women’s day-to-day lives are lost. In general, the A-WEAI reinforces ahistorical and problematic notions of women’s empowerment that ignore other forms of agency, in particular socio-cultural contexts (Parpart et al. Reference Parpart, Rai and Staudt2002; Mahmood Reference Mahmood2011). As Naila Kabeer emphasizes, whether a particular development intervention ends up moving the needle on empowerment depends greatly on “the questions asked, and the interpretations given to the answer, both of which reflect the underlying model of intrahousehold relationships which underpin these evaluations” (Reference Kabeer2001:80).
The most recent efforts to address the limitations addressed here have led to the creation of the newest evolution of the WEAI tool: the project WEAI (pro-WEAI) and the WEAI for value chains (WEAI4VC). While still under development, these latest iterations promise a more multidimensional understanding of empowerment that moves beyond the farm to encompass additional realms such as processing, trading, and marketing (Malapit et al. Reference Malapit, Ragasa, Martinez, Rubin, Seymour and Quisumbing2020). The pro-WEAI was designed to redress some of the limitations we have flagged here by incorporating a set of qualitative protocols alongside the quantitative index designed to capture the nuance of local contexts (GAAP2 2018). While these newest iterations certainly seem like a step in the right direction, we remain skeptical that a quantitative index with qualitative validation will be sufficiently nuanced to capture the intra-household dynamics that shape outcomes in rural development. Tacking on qualitative validation to a quantitative index is simply not enough. We support S. Fraval and associates’ (Reference Fraval, Hammond, Bogard, Ng’endo, Etten, Herrero and Okafor2019) call for a critical evaluation of the reductionist tendencies that plague large-scale agricultural surveys such as the A-WEAI, in order to enhance the credibility and reliability of the resultant datasets. Households can only be understood by spending time with members of that household. Researchers need to develop new strategies to apprehend these complex realities rather than trying to gloss over them. The A-WEAI obscures subtle but profoundly important household dynamics and social relations in Africa that must be understood if women’s empowerment in agriculture is to be gained.
The Women’s Empowerment in Agriculture Index (WEAI) was launched in 2012 as a tool to measure women’s inclusion and agency in the agricultural sector. Funded by the United States’ Feed the Future Initiative, this household survey instrument scores women’s level of decision-making and control, using ten indicators across five domains: production, resources, income, leadership, and time (Alkire et al. Reference Alkire, Meinzen-Dick, Peterman, Quisumbing, Seymour and Vaz2012). In subsequent years, the WEAI was trialed, refined, and revised, leading to both an updated version of the original tool and the Abbreviated-WEAI (A-WEAI), released in 2016, which was designed to address concerns that the original survey was too long for respondents (Malapit et al. Reference Malapit, Pinkstaff, Sproule, Kovarik, Quisumbing and Meinzen-Dick2017). The A-WEAI is anchored by the same five domains but comprises only six indicators, in order to provide a survey instrument that is faster to implement. Given the growing interest of bilateral and multilateral development agencies in the question of women’s empowerment, the tool’s purported ability to provide a universal measurement of empowerment has made the WEAI and its variants increasingly popular with researchers and development organizations; since its launch, nearly one hundred organizations in 53 countries have utilized the tool or an adapted version (Malapit et al. Reference Malapit, Quisumbing, Meinzen-Dick, Seymour, Martinez, Heckert and Yount2019).
As the WEAI’s popularity grew, so too did the criticism leveled against it. Most centers on the assumptions that are drawn about women’s empowerment and autonomy and control over productive resources. S. Akter and colleagues (Reference Akter, Rutsaert, Luis, Htwe, San, Raharjo and Pustika2017) point out that the A-WEAI is derived from data about rural gender relations in some key target countries which may not always transplant well into others. In the same vein, S. Gupta and associates (Reference Gupta, Vemireddy, Singh and Pingali2019) observe that the A-WEAI is not context specific and needs to be adapted to match local circumstances. C. R. Farnworth and colleagues (Reference Farnworth, Rozel, Stirlinga, Chinyophiro, Namakhoma and Morahan2018) argue that the A-WEAI’s conception of empowerment is premised on a male-female dichotomy that neglects the importance of collective and intra-household decision-making (see also Doss et al. Reference Doss, Kieran and Kilic2017). Corey O’Hara and Floraine Clement (Reference O’Hara and Clement2019) note that A-WEAI measures only those forms of empowerment of which respondents are self aware; it neglects the question of whether or not respondents possess critical consciousness. Robin Richardson (Reference Richardson2018) critiques the WEAI for using a universal “cut point” derived from how respondents score on the survey which classifies them as empowered or disempowered, yet does not take into account women’s own perspectives or understanding of their situation.
While the critiques cited above do question the conceptual underpinnings of the tool, they are basically concerned with how shortcomings of the survey can be addressed, whether through adding new variables or indicators to the scoring formula, or by combining the survey with other qualitative methodologies. By contrast, in this commentary we offer a more fundamental critique of the A-WEAI. In our view, the underlying assumption of the survey—that it is possible to obtain an accurate measurement of women’s empowerment using a rapid survey and quantifiable data—reinforces what Andrea Cornwall (Reference Cornwall2018) calls “neoliberal empowerment.” By this, she refers to the way in which women’s empowerment has been appropriated by the development industry and has emerged as a key buzzword with little connection to its orginal transformative meaning. This process has been driven by a widely held belief that enhancing women’s autonomy and decision-making power will automatically lead to increased economic growth and wider societal benefits, despite several critiques that question this instrumental assumption (O’Laughlin Reference O’Laughlin2007; Kabeer & Natali Reference Kabeer and Natali2013; Stevano Reference Stevano2019). By revealing key flaws in the A-WEAI, our goal is to contribute to the critical literature on empowerment which seeks to reclaim the concept and its original emphasis on challenging power relations that uphold or justify social inequality (Batliwala Reference Batliwala2007; Cornwall & Rivas Reference Cornwall and Rivas2015). While this literature provides a rich historical and theoretical critique of neoliberal empowerment, it has yet to provide a fine-grained analysis of the A-WEAI. This is an important gap, because the instrument has proven to be so influential. Continued, uncritical use of A-WEAI may divert attention from other methods and activities that can promote collective action and consciousness-raising needed to foster real empowerment.
The roots of the empowerment concept can be traced to feminist organizing and popular education in the Global South in the 1980s and 1990s. For early pioneers of the concept (Batliwala Reference Batliwala1993; Kabeer Reference Kabeer1994, Reference Kabeer1999), women’s empowerment necessarily entailed challenging hierarchies and cultural norms that justified gender inequality, stressing the importance of collective action in enhancing women’s ability to make strategic life choices, and understanding empowerment as an ongoing and culturally varied process—not a fixed state or measurable outcome. Yet, this emphasis on challenging oppressive norms and attitudes, employing collective action, and sensitivity to specific cultural contexts was diluted as the concept became “mainstreamed” within development institutions and practices and operationalized in quantitative surveys seeking a universal measurement such as the WEAI. Andrea Cornwall (Reference Cornwall2018) uses the term “empowerment lite” to describe the contemporary meaning and use of the concept, in that empowerment has become apolitical and non-threatening, no longer signifying changes in power or hierarchical relations (see Batliwala Reference Batliwala2007). Similarly, Cecilia Sardenberg (Reference Sardenberg2008) calls it “liberal empowerment,” implying that the emphasis is upon individual improvement through entrepreneurialism, particularly with reference to micro-credit. Cornwall and Althea-Maria Rivas argue that the concept rests on a simplified understanding of gender relations that overdetermine other aspects of women’s lives, such that “intimacy, cooperation, mutuality, come to be shrouded in images of the irresponsible and sexually voracious male, and the long suffering and victimized female” (Reference Cornwall and Rivas2015:403).
Our commentary contributes to this critique of neoliberal empowerment by revealing three foundational problems with the A-WEAI. First, the instrument itself relies on categories and codes that lead to distorted representations of intra-household relations. Second, the A-WEAI is unable to account for the diverse circumstances facing farming households across Africa. The final limitation concerns the individualistic nature of the tool and its inability to account for relational dynamics.
Our analysis is based on our experience applying and testing the A-WEAI in Ghana, Kenya, South Africa, and Uganda, as part of a larger research project investigating the impact of new improved crop varieties on intra-household gender relations. Our team purposefully selected five households in each country to test the A-WEAI, for a total of twenty (20) households. The households in each country case study were selected to convey maximum variation, and respondents were recruited through pre-existing connections established by the researchers and their partner organizations. These households reflect demographic, socio-economic, and agricultural characteristics that are similar to the other households in the regions under study. In each country, we observed the A-WEAI being implemented by two enumerators for each household, one for the male and one for the female household members. Enumerators reviewed and practiced using the A-WEAI questions before implementation, using the A-WEAI instruction guide and survey available from IFPRI’s website. The enumerators translated the questions into the relevant local dialect before implementing the survey to ensure consistency in terminology. After the survey was implemented, we debriefed the experience with both enumerators and participants to ask follow-up questions and clarify issues of interpretation, for example, why and when was it difficult to answer a question.
Each country’s case study presented a distinct context that conditioned responses to the A-WEAI; as Andrea Cornwall and Jenny Edwards note, “Context is crucial in making sense about empowerment” (Reference Cornwall and Edwards2010:2). In South Africa, our research was carried out with farm workers and dwellers in the Eastern Cape. These respondents reside on land and in houses owned by their employer, the farm owner. The respondents have access to land for growing subsistence crops and keeping a small number of animals, though these entitlements are often contingent on at least one member of the household maintaining employment status with the farm owner. The Ugandan participants hailed from Nakaseke District in the central Bugandan kingdom. Most of these farmers are over the age of fifty and reside in male-headed households on plots of between five and ten acres, where they cultivate a mix of crops alongside the primary staple, matooke banana. They are generally landowners or reside on family land and polygamy is common. Farmers in Kenya reside in Muranga County, one of five counties in Kenya’s central region. All of these farmers produce mixed crops, including tea, though maize remains the most common staple under production. The land cultivated by farmers in this region is under three acres; farms are both independently and family owned. In Ghana, we selected a farming community in the Wa East district of the Upper West Region. Farmers in this region live in extended family households and farm on family plots. Staple crops include cowpea, maize, and groundnuts. Farmers also keep livestock in addition to their agricultural activities. While farming plots were located further away from the community, it is common to also have food gardening plots nearer to the household. This community, like others in the northern sector of Ghana, is patrilineal.
In this commentary, we examine each module of the A-WEAI, identifying areas of confusion and problematic assumptions. We conclude by offering reflections on creating a more comprehensive and accurate assessment of the relationship between women’s empowerment and agriculture.
Household Decision-Making and Income Generation: Module G2
After the initial process of individual identification, the first substantive section of the A-WEAI, Module G2, lists a series of livelihood activities and asks respondents to report the extent of their involvement in decision-making for each one. Respondents are asked to rank how much input they have in decisions regarding the stated activity (G2.03), to what extent they can make personal decisions (G2.04), and how much input they have regarding income derived from those activities (G2.05).
The first two activities focus on the farming of food crops (Activity A) and cash crops (Activity B). While this distinction between food and cash crops makes sense as a starting point, given the historical pattern in many parts of Africa where men have more responsibility over crops grown for market (Dolan Reference Dolan2001), our experience with the A-WEAI suggests that a hard delineation between these two categories is misleading. In Uganda, for example, gender dynamics vary, depending on the crop and the plot. One female coffee farmer explained that she holds a high degree of personal decision-making power over her own coffee plot but only a small degree of input on her husband’s plot. We also observed that the delineation between food and cash crops was fluid; one husband explained that he only considered coffee a cash crop, while his wife considered matooke banana, maize, groundnuts, and beans cash crops because they sold the excess at the local market. This trend persisted throughout all of our Ugandan households. Men and women understood the distinction between food and cash crops differently, which invalidated the comparability of their answers. Delineating all crops and plots across a single axis (food vs. cash crops) collapses distinct dynamics into a single value that provides a homogenized and distorted representation of women’s participation, decision-making, and control over all crop types.
Female respondents in South Africa often identified themselves as the primary decision-makers for both food and cash crops, including the use of income from crops sold. Husbands typically corroborated this pattern, claiming that they were little involved in crop production and selling, due to the fact they were employed on commercial farms. The survey thus gave an impression of strong female empowerment with respect to these two activities. However, further questioning undertaken after the survey was completed revealed that the food crops grown by these families were grains and vegetables, grown on small garden plots adjacent to their houses. In South Africa, “cash crops” were simply the small surpluses of these crops, representing a relatively minor aspect of the overall household income. Thus, while small-scale agriculture provided some supplemental food and income, it was less central to these farm worker households than the wages earned on the farm or from social grants.
Two other problems relating to decision-making were revealed in Kenya. In two cases, the husband and wife comprised one household amid several related families living on a larger family compound. The land and farm were owned by the husband’s father. Cash and food crop production were a family activity and outputs were shared by the family, though the ultimate decisions were made by the owner of the land and farm, the eldest male. Thus, questions about decision-making and participation were actually a reflection of what was allocated to the household, and not a function of what the individual household members had produced or contributed.
Activity Codes G and H ask these same questions concerning participation, decision-making, and input within major and minor household expenditures made in the last twelve months. One major source of confusion for respondents here was that the category of expenditure failed to differentiate between those made for the individual use versus those made for the household; empowerment in such cases is therefore contingent on the purpose for which the money is being used. Decision-making processes are also more complex in extended and polygynous family households. Asking individuals about decision-making in situations where such decisions would be made collectively was awkward. In South Africa, men explained that they can borrow money without consulting their wives if the purchase is something just for them, but if they are borrowing for a family-related expense, then they consult with their wives. In Ghana, one male respondent reported borrowing money to pay for school fees, with the understanding that this was joint debt borne with his wife. In Kenya, men differentiated between selling a bicycle or motorcycle, which was their property and did not require consultation, as opposed to decisions regarding agricultural inputs, which needed to be made jointly with their spouses. One Ugandan farmer underscored this distinction using the example of selling a cow: if the cow were being sold for school fees—a joint expense—then he would consult with his wife, but if the cow were being sold to buy a bicycle—an individual expense—then he would make this decision on his own. Our experience suggests that confusion between individual and collective senses of responsibility leads to confusion and inconsistency in responses. This paralleled the difficulty in documenting access and ownership to productive capital.
Access to Productive Capital: Module G3 (A)
Module G3 (A) lists a range of assets and asks if anyone in the household owns them (3G.01). It then asks the respondent if the asset is owned solely, jointly, or not at all (G3.02). Generally speaking, most respondents understood the concept of ownership in terms of documentation, the right to sell, and full access. But these three dimensions are not captured by the options presented in the survey. The problems we encountered here were not so much in the definition of the assets (which are more specific than the livelihood activities listed in G2), but rather in the delineation of ownership, specifically around the designation of “joint.” In each country, participants often claimed that their assets were jointly owned. Yet, what they implied by “joint” varied considerably, and the survey fails to capture this nuance and variation.
The designation of joint ownership can overstate the degree to which women can access certain resources. For example, a woman in South Africa stated that she jointly owned the family vehicle with her husband, despite the fact that the husband was the only driver. To illustrate her joint ownership, she recalled how on a recent trip she made to town, her husband gave her money to pay a mechanic for repair work which had been recently completed on the vehicle. That she was given the responsibility to pay the mechanic was, for her, evidence of joint ownership of the car. While this example suggests some degree of communication and cooperation around the management of the vehicle, one could equally interpret it as substantiating a picture of hierarchical gender relations, in which the woman is merely delegated a task (paying the mechanic) by her husband. Labeling these relations as “joint” is problematic because the A-WEAI scores sole ownership and joint ownership the same way; both are counted as meeting the threshold of empowerment. The survey scores this woman as if she were the sole owner of the car, when in reality, ownership means different things for her than what is intended by the survey. This parallels other assets. In Kenya, one household noted joint ownership of a television. But the husband was clear that he had the final say over what was going to be viewed on the TV, and his preference dominated. Thus, the TV was a joint asset, and decision-making about selling that asset was argued to be joint, but use of the asset was hierarchical. In Ghana, respondents expressed discomfort in discussing ownership, because assets were family assets, sometimes passed down from older generations for use of the whole family (which comprises multiple nuclear family units).
While respondents often implied, through the “joint” designation, that assets were owned collectively with the spouse and/or other family members, it sometimes included other authorities. This was particularly evident with reference to land. Farm workers in South Africa often said they owned land jointly, but they were referring here to a sense of co-ownership between themselves and the official landowner (a white farm owner). The “joint” designation not only fails to specify the nature of these relationships, it also disguises how women may be subordinated within them. In the case of South Africa, the jointly held land to which farm workers referred actually entailed a weaker land entitlement for women. As mentioned above, the husbands were the ones employed on the farm. A wife’s access to land and housing was contingent upon the husband remaining employed; in the event of his death or divorce, the wife could be evicted from the property. This weaker attachment to land was conveyed to us in the different responses by one couple in module G3. While both husband and wife stated that their land was owned jointly, only the man said that their house is owned jointly (implying again shared ownership with the white farm owner). For her part, the wife claimed no ownership of the house, stating flatly “it is owned by white farmer, we are just staying there.”
The narrow depiction of land ownership presented similar complications in other countries as well. In Ghana, participants hesitated to comment on the question of land ownership. This is because of the clear delineation of people as either settlers or Tendaanba (landlords/land custodians). Only people from Tendaanba clans can claim “ownership” over land. The Tendaanba, as the original inhabitants and owners of the land, have socio-religious jurisdictions over vast territories and communities, but they are not present and/or involved in the day-to-day administration of these places. Even then, preferred discourse revolves around custodianship. Subsequently, while men could speak about living and working on family lands, the women we interviewed found this question altogether inappropriate. With some clarification, and often apologies, the women spoke about working on family plots that belong to or had belonged to the male spouse’s family. Generally, while participants are unlikely to ever be evicted from their farms or homelands, they cannot sell any land or convert farming land into residential plots without consultation with the Tendaanba. In any case, land granted to settling clans is only for subsistence purposes. In Uganda, respondents explained patiently that different ownership regimes related to different plots. If the husband and wife both contributed to the purchase of the plot, then the ownership would be joint, but if the savings were the man’s alone, he considered himself the sole owner. Enumerators were forced to select a single answer to encapsulate these contradictory arrangements.
Other instances of productive capital listed in Module G3(A) produced results that were similarly contested. In South Africa, male respondents would indicate “joint” ownership of wages, but their wives indicated feeling no ownership over their husband’s wages, although they would attest that they were allowed to buy groceries with it. In Uganda, it was the question over tools that generated the most confusion. Most households owned multiple pangas (machetes), with one that belongs to the husband that no one else is allowed to touch, and others that are owned jointly by all household members. Again, the enumerator is forced to collapse distinct regimes of ownership into a single response.
Ownership regimes are multidimensional, often overlapping, and subject to a range of social norms; as a result, they are difficult to capture in a simple survey (Doss et al. Reference Doss, Kieran and Kilic2017). Augmenting this challenge is the manner in which respondents interpret ownership; as Krista Jacobs and Aslihan Kes (Reference Jacobs and Kes2015) discuss, perceived ownership and joint ownership often vary depending on whom you ask. More to the point, as the examples above illustrate, perceived ownership does not necessarily equate to equal decision-making powers or legal rights. The take-away here is that the A-WEAI survey misses the complex and specific ways in which people exercise control over resources.
Access to Credit: Module G3 (B)
Module G3B asks farmers about their decision-making regarding borrowing over the past twelve months. This tool goes through a range of different lenders, including non-governmental organizations, formal lenders such as banks, informal lenders such as friends and relatives, and formal/informal group-based lending programs. Our experience suggests that many forms of credit are excluded by this list. For example, in South Africa, the primary lender of relevance for the respondents was their employer (a white farm owner). The enumerator chose to list this as an “informal lender” on the survey, since the option of employer is not listed. However, the money these respondents receive is in the form of a cash advance—money paid out to them in advance of payday, which would later be deducted from their salary. Only men reported having access to this form of credit. The survey would thus score women as disempowered for this section, even though men would say that they consult with their wives regarding the use of the income. In Uganda, a follow-up conversation with one female farmer revealed that she had participated in an arrangement whereby she received a heifer and was required to return a pregnant heifer to the lending group as repayment. When asked why she chose not to disclose this within the tool itself, the respondent replied that she understood the question to be focused only on cash loans and not in-kind loans. This limitation is noted explicitly within the tool itself; there is a footnote attached to the list of lending sources instructing enumerators to give “locally relevant examples” (Feed the Future 2015:6). But the rigid nature of the tool and the accompanying time restrictions leave no room to capture these outliers.
Other dynamics are similarly occluded here. In South Africa, some men mentioned that they can borrow money without consulting their family if the purchase is for something just for them, but if they are borrowing for a family-related issue, it requires wider consultation. In Uganda, one farmer reported borrowing money from formal lenders at three different times in the past year. The first time, he did it on his own and did not tell his wife about it; the second time, he decided to tell his wife because the loan was for school fees (a joint expense); and the third time, the husband and wife borrowed together. In this case, the enumerator decided to score the decision-making as undertaken by both the husband and wife, ignoring that the first loan was made without the wife’s knowledge, and thus serving to overestimate the measure of her knowledge and control over these matters.
Time Allocation: Module G4
The time allocation activity (Section G4) seeks to establish how much time is spent on productive and domestic tasks. The respondents are defined as having an adequate workload if the number of hours they spend on work-related tasks (i.e., wage employment, farming, shopping, weaving/sewing, textile care, cooking, domestic work, caring for others, commuting, and traveling) is less than the time poverty line of 10.5 hours (Malapit et al. Reference Malapit, Pinkstaff, Sproule, Kovarik, Quisumbing and Meinzen-Dick2017).
The section on time allocation presented problems in each country. A central issue that cut across each setting was that the respondents, more often women, had difficulty assigning only one activity for each fifteen-minute segment of the day, because they are constantly multi-tasking. For example, the enumerator in South Africa recorded a female farmer in South Africa watching TV for seven hours, but in reality, she completed multiple tasks during that time period, including washing, mending, meal preparation, and caring for children and elders. In Kenya and Ghana, men and women laughed out loud when asked how much time they spent “exercising,” which is one of the time allocation activities. In Kenya, several households were farming on steeply sloped land. This necessitated going up and down routinely throughout the day for farm and household duties, and thus they exercised routinely, but not formally. In Ghana and Kenya, women were more likely to multitask, while men were more likely to be engaged in one activity at a time. Thus, the tool tended to underestimate the number of tasks women are undertaking concurrently. One woman in Ghana talked about being able to take a break for only about an hour during the day (while doing other miscellaneous tasks she does not consider “work”), which illustrates the challenge of breaking down the day into productive and non-productive hours.
At a broader temporal scale, the A-WEAI tool requires respondents to recall their activities over the past twelve months. Farmers struggled with this scope. Most farmers preferred to recall their activities according to a calendar year, forcing the enumerators to amend their records to fit within this more intuitive mode of recollection. Farmers also struggled with the variations contained within this timeframe. In Uganda’s tropical climate, twelve months comprises two wet and two dry seasons. In Kenya, there is both a long and a short rainy season, which are different in length. The measure does not consider variations in activities between dry and wet seasons, and thus the measure of workload and the time allotted for activities is highly dependent on when the enumerator arrived on the farm.
Group Membership: Module G5
In our assessment, this final section of the A-WEAI tool, dedicated to group membership (Section G5), is the most accurate. The section is flexible enough to track the range of groups and associations respondents belonged to. Still, some confusion around what constitutes group membership arose. In Uganda, for example, it was unclear whether membership means that someone pays dues or whether it can include someone who is simply a participant in group activities. In Ghana, farmers felt they were members of a group if their names were written down or if they attended meetings and participated in activities. In South Africa, one participant noted that there is no specific question about political party membership or trade union involvement. This might be because it is a sensitive area of discussion, but as a result the questionnaire potentially misses important ways women exercise agency. In Kenya, political parties were also notably absent from the list of possible farmer group memberships. Also challenging was how to categorize ethnic-specific groups, such as the Council of Kikuyu Elders, which several respondents wanted to be distinct from the other categories. Some minor tweaks to phrasing could be sufficient here, such as expanding the parameters of the question and allowing participants more agency in defining what group membership means in their particular context.
Conclusion
In this commentary, we have outlined several concerns and limitations with the A-WEAI that arose from our experiences testing this tool in four African countries. The first series of limitations relates to the instrument itself and the goal of rapidly administering and completing the survey. The A-WEAI relies on categories and codes that produce distorted representations of intra-household dynamics. We identified a number of these misrepresentations, including the distinction between food versus cash crops, ambiguity around what distinguishes sole versus joint ownership, and assessing control based on a focus on assets versus expenditures. The decision codes themselves present another source of misunderstanding. For example, farmers asked for repeated clarification around the meaning of “to what extent,” forcing enumerators to make difficult and imprecise judgements. Many chose to quantify “extent” in terms of percentages, which collapses the dynamics of both severity and frequency. Restrictive categories lead to restrictive conversations; in our experience the most important gender dynamics were revealed through extended, informal conversations that occurred after the A-WEAI had been completed. There are additional challenges associated with the A-WEAI’s rapid implementation; there is a propensity for rushing (A-WEAI enumerators are instructed to spend a maximum of one hour with each respondent, eliminating the possibility of follow-up questions), an insufficiency with regard to enumerator training (training sessions last one day only), and issues with simultaneous translation (for example, there is no direct translation of “empowerment” in many African languages, forcing enumerators to add their own explanations and elaborations) (see Rizzo et al. Reference Rizzo, Kilama and Wuyts2015 and Cheema et al. Reference Cheema, Mehmood and Khan2018 for similar critiques).
Additionally, the A-WEAI is unable to account for contextual differences and the diverse circumstances facing households across these four countries. The A-WEAI is premised upon comparing responses between one male and one female member of the same household to evaluate gender relations in agriculture. But many of the households we encountered did not fit neatly within this categorization, and none of their decision-making is confined within this single household bubble. For example, follow-up conversations in Uganda and Ghana revealed that a number of participant households were polygamous, which is common within central Uganda and the Wa East district (respectively) where the tool was tested. Thus, the A-WEAI’s emphasis on singling out one male and one female household member completely ignores the important gendered dynamics that shape interactions between senior and younger wives (Nyantakyi-Frimpong Reference Nyantakyi-Frimpong2019). In South Africa, we interviewed farm workers who live on commercial farms full time. They do not have their own land, but they have access to land for growing crops. In Ghana, respondents were mainly subsistence farmers who farm on their own plots and on family farmlands, occasionally engaging in communal farming on a rotational basis. Even if a nuclear family household is physically separated from the main family house, decisions are often made with reference to the larger extended family and the marriage and residential patterns thereof. For instance, a wife of a younger brother would defer to the wife of an older brother in decision-making on the basis of seniority. Many of the Kenyan households who participated were allocated plots of land on family property for the purpose of harvesting cash crops, but with decision-making and revenue shared between several households on the family plot. The goal of the WEAI and its variants is to provide a single aggregate score for a particular region of a country. In doing so, it erases all sorts of context-specific information that is crucial to understanding gender relations in agriculture (Schatz & Williams Reference Schatz and Williams2012; Carletto et al. Reference Carletto, Jolliffe and Banerjee2015).
The final limitation concerns the individualistic nature of the A-WEAI. The scholarship on household bargaining and cooperation has expanded our understanding of the farming household as a unit that incorporates multiple decision-makers (Doss & Quisumbing Reference Doss and Quisumbing2020). But the A-WEAI misses the motivations, structures, and outcomes of household bargaining that are only revealed via interactive exercises (Bernard et al. Reference Bernard, Doss, Hidrobo, Hoel and Kieran2020; Iversen et al. Reference Iversen, Jackson, Kebede, Munro and Verschoor2011; Lecoutere & Jassogne Reference Lecoutere and Jassogne2019). The A-WEAI adopts a collective approach, insofar as it recognizes a single male and a single female decision-maker, but it fails to incorporate any sort of collective exercise where farmers share experiences, learn from one another, or interact. In doing so, the A-WEAI implicitly promotes a model of personhood that is at odds with the more collective values and/or practices and norms that persist in rural Africa. The result is a measure that equates empowerment with individual autonomy but ignores how intra-household relationships determine outcomes, leading to a distorted picture of intra-household dynamics. Many of our respondents were reluctant to overestimate their individual level of control over specific items or issues, relaying that even individual decisions have to be made with the family in mind; they viewed it as morally problematic to act only in one’s self-interest. We believe that relational dynamics, where individuals correct, elaborate, or contradict one another, need to be foregrounded in order to obtain accurate understandings of empowerment in agriculture.
The A-WEAI’s core limitations stem from the assumptions that constitute the tool’s essence: that women’s empowerment in agriculture can be accurately conveyed through a large, rapid, assessment of households using quantifiable data. While this approach may provide intriguing, generalizable snapshots of women’s empowerment in agriculture and can promote opportunities for regional or national comparison, the complexities of women’s day-to-day lives are lost. In general, the A-WEAI reinforces ahistorical and problematic notions of women’s empowerment that ignore other forms of agency, in particular socio-cultural contexts (Parpart et al. Reference Parpart, Rai and Staudt2002; Mahmood Reference Mahmood2011). As Naila Kabeer emphasizes, whether a particular development intervention ends up moving the needle on empowerment depends greatly on “the questions asked, and the interpretations given to the answer, both of which reflect the underlying model of intrahousehold relationships which underpin these evaluations” (Reference Kabeer2001:80).
The most recent efforts to address the limitations addressed here have led to the creation of the newest evolution of the WEAI tool: the project WEAI (pro-WEAI) and the WEAI for value chains (WEAI4VC). While still under development, these latest iterations promise a more multidimensional understanding of empowerment that moves beyond the farm to encompass additional realms such as processing, trading, and marketing (Malapit et al. Reference Malapit, Ragasa, Martinez, Rubin, Seymour and Quisumbing2020). The pro-WEAI was designed to redress some of the limitations we have flagged here by incorporating a set of qualitative protocols alongside the quantitative index designed to capture the nuance of local contexts (GAAP2 2018). While these newest iterations certainly seem like a step in the right direction, we remain skeptical that a quantitative index with qualitative validation will be sufficiently nuanced to capture the intra-household dynamics that shape outcomes in rural development. Tacking on qualitative validation to a quantitative index is simply not enough. We support S. Fraval and associates’ (Reference Fraval, Hammond, Bogard, Ng’endo, Etten, Herrero and Okafor2019) call for a critical evaluation of the reductionist tendencies that plague large-scale agricultural surveys such as the A-WEAI, in order to enhance the credibility and reliability of the resultant datasets. Households can only be understood by spending time with members of that household. Researchers need to develop new strategies to apprehend these complex realities rather than trying to gloss over them. The A-WEAI obscures subtle but profoundly important household dynamics and social relations in Africa that must be understood if women’s empowerment in agriculture is to be gained.
Acknowledgments
Numerous individuals and organizations assisted with the research on which this commentary is based. In Ghana, we thank Francis Sanyare and Pronet-North. In Kenya, we thank Daniel Ndegwa, and Nancy Gathege and SACDEP-Kenya. In South Africa, we thank Lalitha Naidoo and ECARP. In Uganda, we thank Regina Nakyanga, Rodgers Atwooki and Volunteers Effort for Development Concern (VEDCO). We also thank all the farmers who participated in the research. Finally, this project benefited enormously from Alanna Taylor’s excellent work on data collection and management.