Empowerment and gender norms among Yoruba women and men cassava producers

Abstract

Yoruba women have gained a reputation as empowered traders, but their empowerment in agriculture has received less attention. This study examines the empowerment of Yoruba men and women cassava producers in Nigeria's Southwest geopolitical zone. It combines data from an Abbreviated Women Empowerment in Agriculture Index (A-WEAI) survey with results from focus group discussions and key informant interviews. An additional focus is on how norms that foster balanced household gender relations interact with those that produce inequalities. Yoruba men farmers appeared more empowered than their wives in all the domains measured by the A-WEAI in spite of women's high involvement in processing and trade. The fact that the same domains contributed most to men and women's disempowerment points to structural issues that affect men and women unevenly. The study finds signs of emerging shifts in gender norms: women demanded to be acknowledged as farmers and indicated first moves into land ownership. However, the most asymmetric effect of normative constraints experienced by women as a result of unpaid labor, alongside their own income-generating activities tend to neutralize these gains. Hence, the embeddedness of pockets of potentially symmetric institutions in a larger asymmetric patriarchal system of imbalanced resource access and agency does not produce equitable outcomes. Future research should address men's and women's participation in cassava trade and production and their benefits from it more holistically.

1. Introduction

In an article published in 1971, Sidney Mintz [1], an acclaimed anthropologist, declared: “Probably no people on earth has institutionalized women's rights to engage in trading activity so fully as have the Yoruba (…); Yoruba women not only have a wholly acknowledged right to trade and to use their capital largely as they see fit, but they also dominate the internal market system” (1, p. 260). The author described Yoruba men and women as occupying complementary positions in agriculture and trade, women not only having high levels of control over their own income, but also being expected to make substantial contributions to the upkeep of the household. In his study, Mintz [1] relied on historical sources and the work of Marshall [2]. Although admitting that women might have participated in some farming tasks in the past, Marshall (2, p. 2) writes: “Today women rarely take part in any phase of agriculture even though many of them live in the farmland areas.” Marshall and Mintz [1,2] disseminated an image of empowered Yoruba women traders and processors living and working side by side with men producers. The perception of extraordinary female agency in trade may have contributed to less attention for Yoruba women's engagement in farming. Underlying inequities in value chains as well as their outcomes remained unexplored. Yet, a holistic understanding of these is crucial for designing initiatives that promote equity and empowerment.

This article contributes towards closing this gap. It investigates Yoruba women's involvement in agriculture and enters into a discussion with other studies that have recently investigated the terrain. Data from an Abbreviated Women Empowerment in Agriculture Index (A-WEAI) survey is used, together with qualitative results from focus group discussions (FGDs) and key informant interviews (KIIs), to examine the following two questions.

• 1.How empowered are Yoruba women and men farmers in the A-WEAI domains (input in productive decisions, control over use of income, ownership of assets, access to and decisions on credit, group membership and workload)?
• 2.How do norms that foster balanced household gender relations interact with those that produce inequalities?

The study focuses on the Southwest geopolitical zone of Nigeria, which has a majority Yoruba population. In what follows, the historical debates on gender relations in agriculture in this zone are sketched as well as the discussions surrounding Yoruba women's agency.

Before the 1970s, historians tended to question Yoruba women's participation in agriculture. They conceptualized Yoruba women as “‘traders’, ‘wives’, ‘homemakers’, and ‘farm assistants’, farmers' ‘wives’ but never as farmers”, writes Ojo (3, p. 1). Different explanations have been offered for why historians denied women a role in cultivation. Observations limited to specific agricultural activities and rural locations were generalized and taken as a strict and constant division of labor [4]. Women's obligation to work on their husbands' or other male household members' farms obscured the possibility that they might also have cultivated in their own right [3]. Or more broadly, biased by Western patriarchal norms, colonial and post-colonial authorities and scholars constructed men as rural “breadwinners” and could not find a suitable category in which to fit Yoruba women farmers [5]. Yoruba women's fame as traders persisted [1,2], while detailed case studies only slowly revealed their involvement in agriculture.

The results of these studies direct attention towards the introduction of cash crops, as well as towards economic and political developments in Nigeria and how they interacted with rural gender relations. For instance, with the rise of the palm oil industry and trade in the mid-19th century, as Shields [4] suggests, women reallocated labor to related activities and restricted their own farming, with the exception of Ondo in Eastern Yorubaland, where underdeveloped trade routes led to a continuity of women's cultivation. In neighboring Ekiti, where cocoa farming was introduced in the early 20th century, women in less well-off households had to increasingly work in their husbands' cocoa fields, an obligation that placed limits on their time for individual farming. As a result, they demanded remuneration for their services and in some cases earned access to land, which they could then use for their own purposes [3]. In a long-term study in Oyo, Guyer [6] shows how in a period of twenty years (1968–1988), women became part of an expansion of agriculture, the majority of her women respondents' establishing farms of their own. She explains this development with the growing population and urban demand for food, as well as with transport improvements after the end of the Biafran War and in the oil boom of the mid-1970s. Structural adjustment in the mid-1980s raised the costs of living, especially for health and educational services, and further promoted women's entry into agriculture for income generation. These accounts from different locations suggest shifts in the relations between farming and trade in women's livelihood portfolios. Guyer (7, p. 13) writes that these shifts were eased by the fact that gender is only a secondary characteristic in Yoruba culture, and that there is therefore a “situational flexibility in the gender division of labor”.

Local concepts of gender (as opposed to Western concepts) have also sparked a debate on Yoruba women's agency. Oyewumi [8], taking a central role in this discussion, claims that “farming was not a gender-defined occupation in precolonial Yorubaland” (8, p. 145). The most important principle of social organization was seniority (relative age), while “the fundamental category ‘woman’ – which is foundational in Western gender discourses – simply did not exist in Yorubaland prior to its sustained contact with the West” (8, p. ix). The gender bias and inequalities that are prominent today, she writes, were introduced by European colonization and constituted a process of disempowerment for women. Oyewumi's [8] propositions have met with sharp criticism for being based on an analysis of the Yoruba language alone without sufficient empirical research, for taking the absence of gender demarcations in Yoruba as indicating social reality in the past; and for essentializing both Yoruba and Western culture to establish the opposition that she depicts [9]. Scholars have drawn attention to the complex interactions between local and externally inflicted realities that produce inequalities, and to Yoruba women's ability to maintain agency and create opportunities for their economic and social advancement (for an overview, see [10, 11]). Newer studies on Yoruba gender relations have therefore emphasized the interaction between symmetric and asymmetric gender norms and their effects at the household and community levels [12,13]. In the urban setting of Ibadan, van Staveren and Odebode [12] observe norms that support income earning and contributions to household expenditure by both marital partners. These norms “reflect values of autonomy and equality” (12, p. 920). However, they conclude that in a context that is also characterized by asymmetric institutions (such as women's limited property rights or access to resources), symmetric norms do not necessarily lead to symmetric outcomes.

The interplay between symmetric and asymmetric institutions is relevant not only in the urban context, but also in the context of gender relations in Yoruba agriculture, as Forsythe et al. [14] show in a qualitative study. Taking a value chain approach, they do not juxtapose women as farmers versus women as processors or traders, but ask how gender dynamics at one node shape those at the others. Married couples cultivate cassava on separate plots. Women farmers depend on being allocated land by their husbands (patrilineal inheritance) and face restrictions on the rental land market, where transactions are preferably done among men. In spite of this, women have a high level of control over benefits from land. Male smallholders sell their fresh roots to women processors, often giving priority to their own wives, who – as a result – are ensured sufficient supply. Women's autonomy in commercial activities (including income control) goes hand in hand with men's increasing withdrawal from financial household responsibilities, as Forsythe et al. [14] observe in their sample. In another article, Forsythe et al. [15] evaluate the same results against Kabeer's empowerment concept. Kabeer [16] defines empowerment as the ability to make choices in respect of three interrelated dimensions: resources, agency and achievements. While resources refer to access to and claims on material and social resources, agency relates to decision-making and hidden processes of negotiation. Achievements, finally, means the outcomes for the actors involved. In terms of resources, Forsythe et al. [15] see Yoruba women as having lower levels of access to land, credit and labor (as compared to their husbands), which restricts their capacity to produce. As a result, their agency is limited and they shift their focus to cassava production for household consumption and – when there is a surplus – for processing and market sale. An important outcome may be food security. Overall, the incomes women earn from their activities tend to be smaller than those of their husbands. Therefore, they conclude, market participation should not be simply equated with empowerment. The Women's Empowerment in Agriculture Index (WEAI) framework underlying this study also draws on Kabeer's conceptualization of empowerment, which will allow to compare its results with the findings of Forsythe et al. [14,15]. The analysis will also make use of van Staveren and Odebode's [12] concept of symmetric and asymmetric norms to answer the research questions. For in-depth reviews of empowerment conceptualizations, we refer to Elias et al. [17] and Priya et al. [18].

2. Methodology

2.1. Study approach

This study adopts the A-WEAI framework [24], an abbreviated version of the original WEAI, to investigate the research questions. The framework is multidimensional and examines empowerment in agriculture at the individual level across five domains (5DE): input in productive decisions (production domain), control over use of income (income domain), ownership of assets and access to and decisions on credit (resources domain), group membership (leadership domain), and workload (time domain). The second component of the A-WEAI, the Gender Parity Index (GPI), measures how many women are as empowered as the men they live with in the household. This paper restricts itself to the results generated for the five domains and does not calculate the GPI. Following a mixed-method approach, FGDs and KIIs were combined with the A-WEAI survey tool in order to triangulate, validate and supplement research findings. This creates room to explore whether survey results converge with or diverge from the perspectives of respondents as captured in the qualitative data, and where differences emerge, how these can be explained.

The paper relies on data gathered for a PhD study on gender and agriculture [19]. The dataset relates to empowerment, technology adoption and livelihood status including housing, sanitation, health and household food security. Four other papers have been published from the same set with a focus on women's control in decision-making and how it influences household food security [23], gendered time poverty and its impact on health outcomes [22], factors contributing to time poverty among rural men and women [21] and the role of technology in bridging empowerment gaps between men and women decision-makers in the same households [20]. In this current paper the qualitative data (not used in any of the other publications) are analyzed together with the overall A-WEAI results to assess the interplay of culture, gender and empowerment. Insights are provided not only into the empowerment of rural Yoruba men and women in all A-WEAI domains, but also into gender norms that produce symmetric and/or asymmetric effects on women at the household level.

Having outlined this, the limitations of this study need to be acknowledged. The WEAI framework and tools have been criticized for being designed without sufficient reference to varying local understandings of empowerment (for an overview of and contribution to the debate, see [25]). The A-WEAI results presented in this paper could have benefitted from a better adaptation of the tool to the local context and empowerment concepts. This would have demanded an in-depth qualitative study in preparation of the survey. An attempt to at least partially counterbalance this limitation is the joint discussion of survey and focus group results that were collected within the same study. Local empowerment concepts remain an important area for future research.

2.2. Sampling, data collection and analysis

Between September 2016 and January 2017, the survey was administered to a total of 586 respondents from 360 households in Ogun, Osun and Oyo States of Southwest Nigeria. The choice of 360 households for this study was guided by previous WEAI and A-WEAI studies [26,27] which recommend that a sample size of at least 350 households is large enough for index construction and validation. In addition to this, A-WEAI data collection is resource intensive. Working on large sample sizes may be difficult because of resource constraints [27]. Respondents were randomly selected from the register of the All-Farmers Association (AFAN) in Osun and Oyo States, and the database of the Agriculture Development Program (ADP) in Ogun State. In order to create a gender-sensitive and culturally conducive research set-up, enumerators were assigned to same-sex respondents.

In this paper, a deliberate focus is put on dual adult households in which both husband and wives were interviewed. As a result, the overall sample of 360 households was reduced to a sub-sample of 226 households. Single adult households as well as households in which only one adult was reached were excluded. A comparison of single and dual adult households would have gone beyond the scope of this paper and is – in its narrow sense – not counted as gender analysis [28]. In the 226 dual adult households both husband and wife were interviewed separately. The A-WEAI data collection strategy requires that primary decision makers in dual adult households, husband and wife, should be interviewed separately so as to allow for privacy and prevent other members of the household or spouses from influencing or contributing to the answers provided during the interview [24].

On completion of the survey, six FGDs were conducted in the same communities in Ogun, Osun and Oyo States to explore the underlying factors that support or inhibit women's empowerment. An interview guide was developed to generate discussions around the five domains of the A-WEAI. Three FGDs were each organized with men and women smallholders, resulting in a total of 65 participants (with an average of 11 participants per FGD). Respondents were selected based on their experience of at least five years in agricultural activities, and their residence of at least three years in the study areas, to ensure that they had a good insight into the livelihood context specific to their communities. In addition, 33 key informants were interviewed from the three states. These were community leaders, extension agents, research officers in the Ministry of Agriculture, women leaders, representatives of non-governmental organizations (NGOs) working in agriculture and women empowerment, leaders of farmers' associations and local fabricators of cassava processing machines. The languages of investigation were Yoruba and English. The quantitative data were analyzed using STATA. The study uses descriptive statistics in the sense of frequency counts to determine the participation of the different categories of respondents in the five domains of empowerment. Test of differences was also undertaken using chi-square or t-test for categorical or continuous variable respectively. Audio recordings from FGDs were transcribed and translated from Yoruba to English. Notes from key informant interviews (KIIs) were expanded. Following the qualitative content analysis method, transcriptions and notes were coded with the data analysis software Atlas.ti. Important quotations from research participants are included in the text. Similar mixed methods studies based on the WEAI or A-WEAI were conducted among others by Aziz et al. [29] and Doss et al. [30].

2.3. Study site and sample description

The Nigerian Southwest geopolitical zone has a high population of cassava smallholders, and is one of the top three cassava-producing regions of the country. In smallholder dual adult households, husbands and wives commonly cultivate cassava on separate plots with income retained under individual control [14]. Apart from cassava, other agricultural crops grown in the area include vegetables, as well as staples such as maize and yam. In addition, some farmers produce cash crops such as cocoa, kolanut, oil palm, mango, cashew and citrus. Livestock activities are common, especially rearing of chicken, sheep and goats. These support household food security, either through direct consumption or through income generation. Fishery is rare and piggery is virtually non-existent in the study area, the latter possibly because of a large proportion of Muslim residents. The paper focusses on issues surrounding men and women engagements in economic activities relating to their means of livelihood; hence, issues bordering on participation and decision-making on minor and major household expenditures are omitted. Although the major ethnic group is Yoruba, the rural areas in which the study was conducted are also home to immigrants such as Igedes from Benue State, Ebiras from Kogi State, and Bassas from neighboring countries such as Togo and Benin. Some of them regularly serve as hired labor in Yoruba communities. Apart from publications used in the introduction and discussion of this paper, peer-reviewed evidence on gender relations in agriculture in Osun, Ogun and Oyo states is limited.

Turning to the sample of 226 households, the men are generally older and more educated than their wives (Table 1). Half of the women (50 %) work marginal land sizes of below 1 ha, while 71.4 % of the men have farms of above 1 ha, most in the semi-medium (2–4 ha) or medium (4–10 ha) categories. In terms of value chain activities, the overall involvement of both genders in cassava production, processing and marketing emerged as high with only slight (yet significant) nuances, namely a stronger engagement of men in production and of women in processing. However, only participation as such was measured and not the degree of involvement. The latter could have yielded additional insights. In their production activities, men use more exchange labor or hired labor than their wives. Irrespective of gender, more than half of all respondents have no access to extension services at all, and more than 30 % receive such services only rarely. 56 % of men as compared to 38.5 % of women rely on agriculture as their sole income generating activity. For women, trading ranks high as their other income generating activity (42 % for women versus 7.9 % for men). Table 1 is not based on an A-WEAI module but presents individual level results generated through additional questions.

Table 1.

Sample description.

	Men	Women	Chi-square
Age (years)
Young 18-30	12.1	38.1
Mid age 31-50	58.0	51.8	52.99∗∗∗
Old 51-60	15.6	5.3
Aged >60	14.3	4.9
Highest education (%)
No Formal	18.6	28.6
Qur'anic	3.6	0.9
Adult Education	0.5	0.5	16.27∗∗∗
Primary	38.0	39.3
Secondary	28.1	26.8
Tertiary	11.3	4.0
Farm size (hectares)
Marginal farms land (<1)	28.7	50.0
Small [1,2]	11.7	14.8
Semi-medium [[2], [3], [4]]	22.9	13.3	31.21∗∗∗
Medium [[4], [5], [6], [7], [8], [9], [10]]	24.7	18.6
Large (>10)	12.1	3.3
aCassava value chain activities (%)
Production	99.6	95.1	8.56∗∗∗
Processing	88.9	95.1	5.92∗∗∗
Marketing	96.0	93.8	1.15
aType of labor used in agricultural activities (%)
Family	58.4	62.8	0.89
Exchanged	6.2	2.2	4.32∗∗
Hired	92.9	84.3	8.27∗∗∗
Access to extension (%)
Not at all	50.9	54.9
Rarely	31.3	31.9	1.87
Often	17.9	13.3
Participation in agricultural empowerment projects (%)	26.7	27.4	0.03
Agriculture as only income generating activity(%)	56.0	38.5	13.86∗∗∗
Other income generating activities (%)
Trading	7.9	42.0	69.95∗∗∗
Artisan	16.4	10.2	3.84∗∗
Wage work	6.2	3.1	2.48
Salaried work	6.2	2.7	3.39a
Self employed	14.2	7.9	4.48∗∗

^aMultiple response analyses; ∗p < 0.10; ∗∗p < 0.05; ∗∗∗p < 0.01.

The sample of FGD participants consisted of 32 women and 33 men. Women ranged in age between 24 and 70 years (mean: 43 years) and men between 24 and 68 years (mean: 43 years). Among the women, 24 lived in dual adult and 8 in female adult only households, whereas men all lived with a spouse.

3. Results

This part of the paper presents the quantitative and qualitative findings for each of the five A-WEAI domains. A comparison follows of men's and women's empowerment as well as the contribution of each domain to their empowerment result.

3.1. Input in productive decisions

In the A-WEAI survey tool, input in productive decisions refers to participation in and decision-making on six work activities in the household: food crop farming, cash crop farming, livestock raising, non-farm economic activities (including trade), wage and salary employment, and fishing or fishpond culture. This study excluded fishing or fishpond culture, as it is rare in the study area. Asked about their participation in the activities, only two significant differences emerged between women and men: men had a slightly higher participation in food crop farming (99.1 % men versus 94.7 % women). Their wives indicated more participation in non-farm economic activities (56.2 % women versus 36.7 % men) (Table 2), which may be explained by their prominence in cassava processing and trade. This suggests more diversified sources of income generation activities for women with potential for heavier labor burden. Table 2 indicates a generally higher participation of both men and women in food crop and livestock farming than cash crop production. This suggests a stronger focus on subsistence than on commercial agriculture.

Table 2.

Participation in agriculture and other income generating activities in percent (n = 226).

	Men	Women	Chi-square
Food crop farming	99.1	94.7	7.37∗∗∗
Cash crop farming	46.9	44.7	0.22
Livestock raising	66.8	65.5	0.09
Non-farm economic activities	36.7	56.2	17.21∗∗∗
Wage and salary employment	18.6	15.9	0.56

Table 3 shows the respondents' participation in decision-making on the same activities. Respondents were requested to mention all actors involved in the decisions. The following A-WEAI response options were available and could be selected alone or in combinations: “self”, “spouse”, “other household member” and “other non-household member”. Table 3 summarizes how often the variables “self”, “spouse” and “other household member” appeared in the responses (alone and in combinations) as well as the specific combination of “self and spouse”. The variable “other non-household member” and the combinations “self and non-household member”, “spouse and non-household member” as well as “self, spouse and non-household member” proved negligible and are therefore omitted in this presentation. The table reveals several patterns. Looking at food crops, most men indicated decision-making involving “self”. For both food and cash crop activities, women more often than men reported decisions by their spouse or joint decisions. This contradicts Forsythe et al.’s [14] claim that women have a high level of control over benefits from land. Less gender differences emerged for livestock keeping, except for the fact that women perceived more joint decision-making than men.

Table 3.

Respondents’ participation in decision-making on agricultural activities in percent.

	Men	Women	Chi-square
Food crops (n = 224; 214)
Self	94.2	76.17	28.51∗∗∗
Spouse	31.7	75.2	83.27∗∗∗
Other HH member	7.1	5.6	0.43
Self and spouse	29.9	52.8	19.54∗∗∗
Cash crops (n = 105; 101)
Self	97.1	85.2	9.29∗∗∗
Spouse	24.8	63.4	31.19∗∗∗
Other HH member	2.9	3.0	0.0
Self and spouse	21.9	47.6	11.91∗∗∗
Livestock (n = 149; 146)
Self	87.3	91.1	1.13
Spouse	38.9	50.0	3.66
Other HH member	6.0	6.2	0.0
Self and spouse	24.2	41.8	8.20∗∗∗

Figures for other possible variables or combinations of variables are excluded because they are insignificant. Multiple responses were possible.

Results from the FGDs confirm the above patterns, but also reveal that some of the response variables in the questionnaire provide room for interpretation. This is shown in the statement below:

You know that farming is men’s work; it is the father (senior man in the household) that would bring the decision. Maybe they want to spray the farm, or they want to plant something … it is the father that would say let us spray this farm. Farming is not women’s work. It is modernity that makes to be doing it (FGD Men, Ogun State, 19 October 2016).

This respondent justifies the dominant role of men in decisions relating to agricultural activities with the perception that farming is “by tradition” men's work. Women are assigned authority over processing and marketing certain products, especially gaari (processed cassava). Both men and women said that wives may make suggestions, but that husbands take the decisions.

If the household is the one where the husband and wife are together, it is the husband who normally makes decisions on farming, the husband makes decisions at all times (FGD Women, Ogun State, 19 October 2016).

Women were described as increasingly participating in agricultural production (as compared to the past), and as being prepared to try out new agricultural technologies more often than men. However, sole decision-making by women was seen as more likely to be acceptable in the absence of a partner (i.e. in female adult only households). For dual adult households there seems to be room for interpretation of the variables “self” and “self and spouse”. For instance, looking at food crop production, most men indicated decision-making by “self”. However, it is not clear how far these decisions were taken after having received their wives’ suggestions. On the other hand, more women than men reported joint decisions, maybe involving a discussion of their suggestions. An important aspect that emerged briefly during FGDs, but was not sufficiently covered, is the question of how decisions on agricultural activities are made in polygynous households. This should be followed up in future research through adaptation of the tool (see for instance 31).

3.2. Control over use of income

This domain is assessed, firstly, through respondent's inputs in respect of decision-making on the use of income from food crop farming, cash crop farming, livestock raising, non-farm economic activities and wage and salary. The results of the first assessment (Table 4) show that women were more likely than their husbands to have no input or limited input in decision-making regarding the use of income from various activities. Women's influence was highest in respect of non-farm activities (including trade). While women have higher participation in non-farm economic activities such as trading or wage labor, this does not translate to higher decision-making on the control over income from those activities. This is because other relational and structural factors, such as biased gender relation and discriminatory gender norms tilt the household power dynamics and control of decisions in favor of men. These limit women's mobility, increase domestic workload burden, and restrain their decision-making power especially in the context of rural areas where patriarchy is deeply entrenched. The investigation of different income activities and expenditure levels reveals more symmetric and more asymmetric relations and adds nuances to the picture of gendered decision-making provided by other authors [12,14].

Table 4.

Inputs in decision-making on the use of income generated by respondents in percent.

	No input or input into few decisions		Input into some decisions		Input into most or all decisions		Chi-square
	Men	Women	Men	Women	Men	Women
Food crop farming (n = 80; 168)	3.8	16.8	30.0	69.6	66.3	12.5	75.64∗∗∗
Cash crop farming (n = 30; 67)	3.3	16.4	33.3	70.2	63.3	13.4	25.52∗∗∗
Livestock raising (n = 63; 79)	9.5	6.3	49.2	63.0	41.3	30.4	3.46
Non-farm economic activities (n = 13; 36)	7.7	11.1	15.4	64.4	76.9	19.4	14.27∗∗∗
Wage and salary employment (n = 8; 21)	0.0	33.3	50.0	52.4	50.0	14.3	5.73∗∗

Findings from the FGDs suggest there are variations between households in terms of how decisions on income are made. In monogamous settings, joint control over the use of income was seen as more feasible than in polygynous settings, where conflicts over expenditures could arise between wives. In the latter setting, husband and wives were described as tending to individually control the income that each of them derived from productive activities. In monogamous households, men's stronger decision-making power was justified by the belief that women are generally wasteful and will spend scarce income on “unimportant” items. In negotiations with their husbands, women may refer to this belief and emphasize their thriftiness, as the following quotation illustrates:

We would bring our decisions to the men; the men also would look at the decision. We would tender our decisions and suggestions so as not to spend the income in a wasteful way. Both men and women can then make decisions on how to use the income (FGD Women, Osun State, 17 October 2016).

3.3. Ownership of assets

This dimension concerns ownership of productive resources in the household, such as land, livestock, agricultural equipment, consumer durables, house and others. Through a first question, a list of assets available in the household was constructed. Secondly, respondents were asked whether they solely or jointly own the available assets [20]. Table 5 shows that the number of men who said that they own an asset was generally higher than that of women. This was true for all asset types except for small livestock and poultry. Also, more men than women claimed sole ownership of assets. For small livestock and poultry, men indicated the highest rates of joint ownership with their wives (with no significant gender differences for these two asset types, however). Taking an overall perspective, sole ownership tended to be high for both genders in the study area. This confirms van Staveren and Odebode's [12] observation of men's stronger property rights, but also Aluko's [13] description of (to some extent) independent asset ownership among Yoruba couples, embedded in a context of unequal decision-making.

Table 5.

Ownership of assets in percent.

	Men		Women		Chi-square
	Solely	Jointly	Solely	Jointly
Agricultural land (plots) (n = 148; 112)	70.3	29.7	65.2	34.8	11.57∗∗∗
Small livestock (goats, pigs, sheep) (n = 136; 142)	63.2	36.8	70.4	29.6	1.69
Chicken, ducks, turkeys, pigeons (n = 154; 121)	56.5	43.5	63.6	36.4	2.37
Non-farm business equipment (generator, sewing machine, brewing equipment, fryers) (n = 118; 87)	76.3	23.7	71.3	28.7	7.85∗∗
House or other structures (n = 150; 119)	74.0	26.0	58.0	42.0	15.94∗∗∗
Large consumer durables (refrigerator, TV, sofa) (n = 157; 93)	79.6	20.4	63.4	36.6	38.82∗∗∗
Small consumer durables (radio, cookware) (n = 199; 136)	81.9	18.1	72.8	27.2	58.88∗∗∗
Other land not used for agricultural purposes (residential or commercial land) (n = 145; 105)	79.3	20.7	70.5	29.5	13.62∗∗∗
Means of transportation (bicycle, motorcycle, car) (n = 159; 90)	83.0	17.0	57.8	42.2	64.14∗∗∗

In terms of agricultural land, more men (70.3 %) than women (65.2 %) reported sole ownership, a result that will be questioned below. The FGDs revealed the entanglement of men's land ownership with patrilineal inheritance patterns. The latter tend to limit women's land rights to access only, which is in most cases established through their husbands (and may be lost upon their husbands' death). The size of agricultural plots women are permitted to use tends to be smaller than that of their husbands (see Table 1), with exceptions, as outlined by a FGD participant:

After getting permission from their husband, women have no restriction on the expanse of land for cultivation. In our community, there are women who cultivate a larger expanse of land than men. (…) Men and women have no limitation to the expanse of land they can cultivate as long as they have money to secure the necessary technology (FGD Men, Osun State, 17 October 2016).

Nonetheless, respondents in the same group underlined the husband's important position, even if his wife wants to farm her own lineage's land:

She doesn’t have any control over the land. Even in case she really needs the land, she would have to go to her father and tell him, please I want to eat. And the father would call the husband and hand over the land to the husband not the wife (FGD Men, Osun State, 17 October 2016).

A woman from Oyo State describes how men's control can extend to market relations:

Men must be involved before women can use land. Even if a woman is renting land from outside, it is her husband that would be involved in the negotiation process. Even if the woman is a widow, she would need the help of male members of her dead husband’s family to help her on this (FGD Women, Oyo State, 24 October 2016).

Specific questions tailored to the study context would have produced more nuanced quantitative results.

3.4. Access to and decisions on credit

This domain concerns access to and decisions on the use of credit. It examines whether a respondent in a given household has the ability to take a loan, if he or she desires, and if yes, whether the respondent has done so in the 12 months prior to the survey. It also investigates who has participated in decision-making on borrowing or use of the credit [20,24]. Credit sources in the A-WEAI tool are NGOs, formal lenders (bank or financial institutions), informal lenders, friends or relatives, cooperative societies, and informal savings and credit groups.

With regard to the study context, NGOs, formal and informal lenders, as well as cooperative societies are less common in the rural setting than in major Nigerian towns. Informal groups and friends and relatives emerged as the main sources of credit for smallholders. In the year preceding the study, the result shows that a substantial proportion of men (50.6 %) and women (44.2 %) had received credit in the form of cash from informal groups. In a similar manner, 45.7 percent of men and 40.7 percent of women had received credit in the form of cash from friends. The result also reinforced the prevalence of in-cash rather that in-kind credit in a typical rural setting in Southwest Nigeria, this is because credit was almost exclusively given as cash (Table 6). In-kind borrowing or a combination of cash and in-kind borrowing was negligible and omitted from the Table. Turning to decision-making on credit, both men and women saw decisions on whether to take a loan and on what to do with the money as being predominantly taken by “self”, with men reporting slightly higher (but significant) levels of self-decisions than women (Table 7). For the important source of credit such as friends, significant gender differences emerged concerning decisions on borrowing credits. More men (58.5 %) indicated decision-making by ‘self’ while a lower proportion of women (51.3 %) indicated decision making by ‘self’. This pattern is also observed in decision-making on borrowing credit from a source like NGO where more men (54.7 %) than women (47.1 %) indicated decision-making vested by self. However, both men and women reported higher joint decisions when borrowing from NGO.

Table 6.

Sources of credits among men and women in the study area.

	Men		Women
	Yes, Cash	No	Yes, Cash	No	Chi-square
NGO (n = 168; 165)	14.9	82.7	12.1	86.1	7.56∗∗
Formal (n = 167; 175)	7.8	88.6	9.7	87.4	4.89
Informal lenders (n = 99; 109)	24.2	75.8	23.9	73.4	2.76
Friends (175; 155)	45.7	49.1	40.7	56.1	5.31
Cooperatives (n = 145; 154)	29.7	67.6	17.5	81.2	7.66∗∗
Informal groups (n = 156; 190)	50.6	47.4	44.2	52.1	6.48

∗Other options of credit such as in kind and combination of kind and cash, are either insignificant or negligible and hence omitted from the Table.

Table 7.

Decision-making on borrowing and use of credit.

	Men	Women	Chi-square
Who makes decision on borrowing credit from different sources
NGO
Self	54.7	47.1	4.2a
Other HH member	7.6	0.00	4.0a
Friends/Relatives
Self and spouse	58.5	51.3	5.9a
Decision-making on what to do with credit from different sources
NGO (n = 50; 38)
Self	60.0	44.7	4.0a
Informal lenders (n = 28; 42)
Self	82.1	45.2	0.4
Spouse	10.7	30.9	6.5b
Self and spouse	7.1	19.1	3.7a
Friends/Relatives (n = 139; 123)
Self	58.3	50.4	3.7a
Cooperatives (n = 70; 54)
Self	58.6	46.3	4.5a
Informal Groups (n = 188; 175)
Self	63.6	50.9	1.9
Spouse	16.9	25.1	10.5b
Other HH member	3.4	0.0	4.0a
Self and spouse	16.1	24.0	10.0b

∗Formal Group is not reported because none of the options is significant.

^aFigures reported in the Table are multiple response analyses.

^bAside from “Self”, only significant options are reported for all the credit sources.

Decision-making concerning the use (what to do with) of the credit seems to vary with the different credit sources of credit used. For instance, decision-making vested by ‘self’ seems to be more common in regard to the use of credit from sources like NGO, friends and relatives and cooperatives, with more men alluding to this. More women however reported joint decision making and decision making by spouse on the use of credit from sources like informal lenders and informal groups.

In the FGDs, farmers identified access to credit for agricultural production as one of their biggest constraints. Prices for inputs have risen, they explained, while government initiatives for smallholder credits have yielded no results. Regarding decision-making on the use of credit, both sole and joint processes of decision-making were described. Sole decision-making was seen as more appropriate for men, as this statement by a woman in Osun shows:

The husband can bring the decision on the use of the money, for example … myself … there is no amount of money I take, I would first show my husband, I would say, this is the money I collected, how are we going to use it? (FGD Women, Osun State, 17 October 2016).

In spite of this, the quantitative results confirm higher levels of self- as compared to spouse-decisions for women. This could constitute a tension between norms and social reality.

3.5. Group membership

This dimension measures men's and women's participation in different groups. In the setting of this study these are agricultural groups, credit or microfinance groups, mutual help or insurance groups, trade and business associations, civic groups, religious groups, educational groups, political groups, and neighborhood/village leadership groups. Educational groups include parent-teacher associations, political groups relate to Nigerian political parties, and religious groups are either Christian or Muslim. Business groups are linked to smallholders' productive activities and serve to protect the interests of members, share information on market opportunities, and act as facilitators of credit, transport and inputs. Cassava processors (egbe elero) and transport associations (egbe oloko) were counted as business groups. Ethnic associations (egbe abasa) form part of neighborhood groups [20].

In a first step, respondents were asked whether any of the above groups existed in their community. In a second step, they indicated whether they were active participants in the existing groups. The results in Table 8 show that men more frequently than women engage in agricultural producers' groups, civic and political groups, and neighborhood committees. With respect to farmers' groups, the result supports Lodin et al.’s [32] findings. Table 8 also reveals that among women, there was a higher participation in unspecified groups (including women's groups). For all other group types there were no significant gender differences. Overall, religious groups had the highest level of participation. In the case of processing and transport groups, qualitative data revealed that participation depends on owning the required equipment. A woman from Ogun State and a community leader from Oyo State explain:

Women can participate if she has the processing machine (FGD Women, Ogun State, 17 October 2016).

Women do not usually have machines. It is usually the men. More men have machines (KII Community Leader, Oyo State, 15 November 2016).

Table 8.

Participation in group activities in percent.

Groups	Men	Women	Chi-square
Agricultural producers' group (n = 151; 137)	75.5	58.4	9.55∗∗∗
Credit or microfinance group (n = 149; 122)	69.1	64.8	0.58
Mutual help or insurance group (n = 140; 123)	72.1	66.7	0.93
Trade and business association (n = 118; 111)	47.5	53.2	0.74
Civic group or charitable group (n = 161; 103)	83.9	61.2	17.24∗∗∗
Religious group (n = 193; 190)	89.6	86.3	0.99
Educational group (n = 179; 128)	75.4	74.9	0.02
Political group (n = 196; 155)	65.3	43.2	17.09∗∗∗
Neighborhood/village committee (n = 168; 125)	78.0	59.2	12.02∗∗∗
Others, e.g., women's group (n = 74; 92)	29.7	64.1	19.42∗∗∗

Gender inequalities in income levels and in decision-making on income may limit women's ability to purchase these machines. Women's responsibility for domestic work further limits group participation.

There is no time to attend meetings. By the time we cook, fetch water, take care of the children and also attend to our daily productive activities, it will be night (KII Women Leader, Ogun State, 17 November 2016).

In addition, women are expected to get permission from their husband to join groups, as a woman from Oyo State put it:

If the husband is not cooperating, it can be a hindrance. Men can participate freely, they don’t need to take permission from anyone but women must take permission because you’re still subject to the man (FGD Women, Oyo State, 25 November 2016).

Some husbands' disapproval of their wives' group engagement is based on the belief that women's contact with people of “mixed values” during meetings leads to changes of behavior, especially the development of “disreputable habits”. One could assume that women's lower levels of group participation relate to lower levels of credit access (see Table 6). This, however, is not confirmed by the survey results.

3.6. Workload

The time allocation domain measures the amount of time men and women spend on leisure and work-related activities. Using a detailed time allocation record, respondents were requested to recall the time spent on primary activities in the 24 h prior to the interview. The results are presented in Table 9. For a separate investigation of the results in terms of health outcomes please see Adeyeye [22].

Table 9.

Average time allocation for daily activities (minutes).

Activities	Men	Women	t-test
Sleeping/resting	584.0	555.1	2.11∗∗
Eating/drinking	101.0	102.0	−0.21
Personal care	52.4	36.3	4.47∗∗∗
School	0.5	0.8	−0.38
Work as employed	25.9	14.3	1.44
Own business work	77.2	164.4	−4.42∗∗∗
Farming/livestock	332.3	203.2	6.49∗∗∗
Shopping/getting service	8.2	5.7	0.52
Weaving/textile care	0.0	5.4	−3.54∗∗∗
Cooking	4.8	123.1	−18.66∗∗∗
Domestic work	7.8	73.9	−9.53∗∗∗
Care for children/older people	2.5	9.1	−2.66∗∗∗
Traveling/commuting	19.8	15.3	0.68
Watching TV/radio/reading	38.2	20.9	3.03∗∗∗
Exercising	7.8	0.8	3.52∗∗∗
Social activities/hobbies	70.8	47.8	2.33∗∗
Religious activities	95.6	50.2	6.37∗∗∗
Others	9.1	1.8	1.79∗

Table 9 shows that men spend more time on farming and livestock keeping activities. The lower amount of time spent by women on agricultural production could be associated with smaller plots and herd sizes, the use of less labor-intensive practices, and a lower degree of input application. The time women spend working in their husband's fields is not specified and cannot be assessed. Since the A-WEAI time allocation sheet records the previous 24 h only and does not consider seasonality, all farming-related results would need verification through further research. Women invested on average 164.4 min in running their own business, as compared to 77.2 min reported by men.

Men allocate on average more time to sleeping, entertainment (watching TV, radio, reading) and social and religious activities. Women devote on average 2 h to cooking (123.1 min for women versus 4.8 min for men) and more than 1 h to domestic labor (73.9 min for women versus 7.8 min for men), and more time on care work for children and older people. This is in line with the results concerning group membership, where domestic responsibilities were identified as a challenge for participation. Findings from the qualitative analysis corroborate that women in the investigation area have a workload that does not allow for much leisure. This is due to their involvement in agricultural production, processing and marketing, and their almost exclusive responsibility for childcare and domestic chores, a result that supports Forsythe et al. [14]. In this respect, the FGDs constituted an additional burden for some, as one woman emphasized:

As we are sitting here, we are just sacrificing our time … there is no time … if you call us tomorrow for this discussion, we may not come … I won’t sit down (FGD Women, Oyo State, 25 October 2016).

In terms of women's engagement in multiple income generating activities, several women claimed the right to call themselves farmers (“we are original farmers”). This claim can only be understood against the backdrop of gender perceptions that assign farming to men and neglect women's labor contribution (as outlined in the literature review). This “traditional” perspective is more and more disputed. On top of this, as men and women said, women have increased their engagement in “male” agricultural production over the past few decades, while men are slowly beginning to transgress into “female” marketing. In Table 9 it is not clear whether respondents assigned marketing activities to “farming” or to “running their own business”. If the latter were the case, this would illustrate women's higher time allocation to trading. Marketing is often associated with travels to places as far as Ibadan, Oshogbo or even Lagos (a time investment that is not conspicuous and shows no gendered pattern in Table 9). In one FGD, married women said that their husbands do not keep them from traveling but rather encourage them to do so.

The majority of both men and women rejected domestic workload sharing and saw it as appropriate only for those who are educated or who live in towns.

It is a taboo for men to sweep the floor, even when the women are busy or tired. As farmers, we are always too tired after a day’s work to participate in domestic chores. It does not befit us as men. Our kitchen is outside. How will it look if a man is pounding yam outside and everybody is watching? But for city people, their kitchen is in the house, the men can cook, pound without feeling ashamed, since nobody will see them once they close the door (FGD Men, Osun State, 30 September 2016).

It is only during women's absence from home that husbands can contribute to domestic work, as several men stated. Some married women stressed that when they are at home, they have to conform to their husbands' expectations and the social norms of a “good wife and mother”.

3.7. Comparison of men and women respondents’ empowerment in the domains

After having presented the results for each of the five A-WEAI domains, together with insights from qualitative data, a comparison is made of men and women's overall empowerment (Table 10), and the contribution of each of the indicators to disempowerment (Fig. 1). Also, it will be outlined what the FGD participants themselves described as empowerment and how this relates to the other findings.

Table 10.

Comparison of overall A-WEAI results.

Empowerment Indicators	Men	Women
5DE (1 - M0)	0.98	0.91
Disempowerment score (1 - 5DE)	0.02	0.09
% achieving empowerment (1 - H)	0.92	0.72
% not achieving empowerment (H)	0.08	0.28
Mean 5DE score for not yet empowered (1 - A)	0.72	0.68
Mean disempowerment score (1-5DE) for not yet empowered (A)	0.28	0.32
N (number of observations)	226	226

Fig. 1.

Contribution of each indicator to disempowerment (dual adult household).

In terms of the degree of empowerment (5DE) in the five domains of the A-WEAI, men have a higher degree of empowerment (98 %) than women (91 %). 92 % of the men in the sample emerged as empowered, compared to only 72 % of the women. Women who are not yet empowered (28 % of women in the sample) have on average inadequate achievements in 32 % of the domains. For disempowered men (8 % of the men in the sample), inadequate achievements were slightly lower (28 %). The breakdown of the contribution of each of the six indicators to men and women's disempowerment is presented in Fig. 1. The results show that workload, access to and decision-making on credit and group membership contribute most to both men's and women's disempowerment. For women, their workload adds 45 % to their overall disempowerment, access to and decision-making on credit 20 %, and group membership 15 %. For men, workload ranks first with a 35 % contribution, group membership second with 30 %, and access to and decisions on credit third with 20 %. This means that workload contributes more to women's overall disempowerment than to that of men. On the other hand, access to and decisions on credit and group membership add more to men's disempowerment as compared to that of women. For an investigation of how empowerment gaps relate to technology adoption see Adeyeye [20].

In the FGDs respondents were asked how they would describe an empowered man or woman farmer. Men and women's descriptions differed only slightly. An empowered farmer, both groups imagined, would have access to funds and would be able to hire labor, use improved technologies, purchase inputs, and mechanize production. Women explicitly related their empowerment to increased access to land and the ability to cultivate it. Mechanization and a better infrastructure (such as boreholes for water) would reduce their labor time and burden. While men did not identify a source for the funds, women demanded government support in the form of credits. These emic perceptions of empowerment partly confirm the A-WEAI results, especially in terms of the workload and credits. Women's lack of access to land might have been underestimated in the survey, since only access as such and not the size of the accessed land was considered.

4. Discussion

The A-WEAI results presented above can be summarized to answer the first research question: Yoruba men farmers are more empowered than their wives. The fact that the same domains emerge as most disempowering for the two respondent groups (workload, group membership and access to and decision-making on credit) points to structural issues that affect both women and men, but impact them unevenly. These are the high level of drudgery involved in cassava production, the common lack of credit access (especially where collateral is not given), and the insufficiently tapped potential for collective action [33,34]. Abila [33] shows for Oyo State that small cassava farm sizes constitute a key obstacle to introducing labor-saving technologies (such as mechanized land preparation). Also, the low participation in farmers' groups or cooperatives results in missed opportunities for credits, collective input purchases, produce marketing, and training. If one relates Abila's [33] findings (not sex-disaggregated) to this data, the uneven impacts of the same structural issues become tangible: women are expected to combine arduous farm work on individual plots with multiple other paid and unpaid tasks and – apart from the high labor burden – experience male restrictions on group participation. In (A-)WEAI studies conducted in other contexts, time poverty, lack of credit and low group participation also proved to limit women's ability to make choices [27,35]. Bain et al. [36] see women as “weak winners”, if empowerment initiatives are not accompanied by efforts to substantially reduce female labor burdens.

While the A-WEAI survey yielded measurable results on empowerment and gender relations, qualitative findings draw attention to the norms and institutions that households are embedded in at a broader community level. Gender norms convey who is regarded as the “appropriate” decision maker and how resources are allocated. Van Staveren and Odebode (12, p. 903) plead for a distinction between gender norms “that have similar effects for everyone and institutions that have asymmetric effects, that is, systematically different effects on different groups”. In order to answer the second research question, namely how norms that foster balanced household gender relations interact with those that produce inequalities, symmetric and asymmetric effects from the mixed-methods findings are captured (Table 11) and related to other studies.

Table 11.

Potentially symmetric and asymmetric effects of gender norms.

Potentially contributing to symmetric gender relations	Potentially contributing to asymmetric gender relations
Women can participate freely in non-farm economic activities	Men dominate decision-making on food and cash crops and major household expenditures
Women have no restrictions on market activities and travel for trade	Women's participation in groups' activities may present a perception of “disreputability”; women seek their husband's approval for group participation
	Men have higher levels of asset ownership and control of productive resources (including land)
	Women tend to be solely responsible for child care and domestic work; enforcement through the image of a “good wife and mother”
	Women do cultivate independent plots of land as long as they seek the permission of their husbands or male relatives to access and use the land; women have less access to hired/exchange labor

Women in our sample have higher influence on decision-making in the control of income from non-farm economic activities than all the other economic activities; however, when compared to men, their level of decision-making is still low because more men contribute into ‘Input into most or all decisions’ (Table 4). A limitation observed from the A-WEAI is that it did not spell out what constitutes the three categories of decision-making used in the tool (few decisions, some decisions, or most or all decisions). This limits the interpretation of our findings as it is difficult to understand in detail, for example ‘how’ and ‘in what way’ men and women are contributing to decision regarding the use of income. The A-WEAI tool is not specific about control over income from individual agricultural plots. Questions on this should be added in future when applying the tool in the Yoruba context.

Although differences in control over plot income cannot be carved out from this study's data, Forsythe et al. [14,15] report autonomous decision-making by women and men. What the data reveal is that men dominate decision-making on income from food and cash crops farmed on fields that are on average larger than those of their wives (Table 1). Women's partial financial autonomy includes the obligation to contribute to children's school fees and to household provisioning, for instance by cultivating cassava for household consumption [14]. As women in the FGDs indicated, men tend to encourage female market activities and travel for trade, although Forsythe et al. [15] and Lodin et al. [32] also found contrary cases. Lodin et al. [32] observe that trade is perceived as a “woman's domain”, yet women still tend to seek their husband's approval before leaving to the marketplace – a formal act of deference to men's authority.

While one may read norms of some level of income autonomy and equal household provisioning as potentially contributing to symmetric gender relations, their manipulation allows for the opposite: calling upon women's autonomy, men may gradually withdraw their responsibility for household upkeep and increase pressure on their wives to compensate for it [15]. In such a situation, women's room to maneuver is constrained by the asymmetric effects of other gender norms: Men dominate ownership and control of productive resources [37]. Women depend on their husband or male relatives for access to land. The fact that women engage less in agricultural producers' groups and more in exclusively female groups (Table 8) could be due to the “disreputability” assigned to women who engage in mixed gender groups. Husbands' restrictions on their wives' group membership, as well as women's labor burdens that limit participation, may be seen as having a disempowering effect, especially where access to credits and technologies is promoted through farmers' groups. In Oyo state, Olaosebikan et al. [38] show that women who were trained in the cultivation of bio-fortified cassava ended up with lower scales of production than men, since they could not afford hired labor. Women's constrained access to hired labor as compared to men is also expressed in Table 1. This raises the question how far women can make good for lower agricultural production by engaging in processing and trade. Forsythe et al. [15] and Olaosebikan et al. [38] speak of increasing market competition (that fosters customers' payment on loan), low product prices, exploitation by middlemen, and expensive processing equipment, as factors that limit the profitability of women's activities in this field. The scale of processing may also depend on husbands' household contributions (or the absence of them). Being unsure of their husbands' support, women in Forsythe et al.’s [15] study resorted to uprooting cassava from their own fields little by little to ensure the household was fed. What was left as cassava surplus was processed for income. On the other hand, men tended to sell fresh cassava roots in bulk and received larger amounts of money in one transaction. This could also explain men's higher levels of asset ownership in the data as compared to women (Table 5). The most asymmetric effect, however, may be due to norms that burden women with unpaid labor, such as childcare, domestic chores (Table 9, Fig. 1), and additional tasks in husbands' fields, alongside their own income-generating activities. In a study of agricultural households in Southwestern Nigeria, Pierotti et al. [39] conclude that labor and time use negotiations among husbands and wives are not just about more productivity but about conforming with gendered expectations and hierarchies. One can therefore conclude that van Staveren and Odebode's [12] finding, established in the urban Yoruba context, also applies to gender relations in agriculture: the embeddedness of pockets of potentially symmetric institutions in a larger asymmetric patriarchal system of imbalanced resource access and agency does not produce equitable outcomes. At the same time, signs of potential shifts were observable in this study: women demanded to be acknowledged as farmers and indicated first moves into land ownership (for similar results, see 38).

5. Conclusion

Findings from our study, especially from the qualitative findings, reveal a strong presence of relational and structural issues prevalent among men and women in our study area in the Yoruba communities that restrict women in different areas important to their empowerment. For instance, prevailing beliefs and gendered social norms affect ownership and control of resources, market relations, domestic roles, and to some extent group participation. Thus, our findings reveal a shift towards Bakare-Yusuf [9] – showing that gender relations, power dynamics, and prevailing norms are promoting negative underlying factors to reinforce gender demarcations which dictate the extent of men's and women's empowerment in the different domains and indices of empowerment. Older accounts (for instance those quoted in the introduction; 2, 1) juxtapose agriculture and trade and assign Yoruba women an exclusive and powerful position in the latter. Newer studies take a value chain approach and investigate how gender relations at one node shape women's and men's opportunities at the others. Given that the same women often engage in cassava production, processing, and trade, a joint evaluation of Yoruba women's empowerment in agriculture and trade would be necessary to arrive at valid conclusions. As this study reveals, women farmers are less empowered than their husbands. Despite women's high involvement in processing and trade, it remains questionable whether women's incomes from these activities equal those their husbands earn in production [14,15]. Furthermore, the need to combine multiple unpaid and paid tasks proved to have the most disempowering effect on women in this study. Future research should address men's and women's participation in various cassava value chain nodes and their benefits from it more holistically.

CRediT authorship contribution statement

Olajumoke Adeyeye: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Gundula Fischer: Writing – review & editing, Supervision, Formal analysis.

Data availability

No data associated with the study has been deposited into a publicly available repository. A-WEAI data are available from the first author upon request.

Ethics and consent statement

The PhD study was approved by the Federal University of Agriculture, Abeokuta, Nigeria on April 13, 2016. Informed written consent was obtained from all subjects involved in the study.

Funding statement

The PhD study received no external funding. The authors’ collaborative data analysis for this paper was supported by an Intra-Africa Mobility Grant by the African German Network of Excellence in Science (AGNES).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.