Dietary Power and Self-Determination among Female Farmers in Burkina Faso: A Proposal for a Food Consumption Agency Metric

Abstract

Although food security is traditionally defined with four dimensions, there are increasing calls for an additional two: agency and sustainability. However, it is especially challenging to measure agency, a person’s control over their food production and consumption. Measuring women’s agency is especially critical in African regions south of the Sahara, where women play prominent roles in farming and food preparation. This paper explores the feasibility of creating a metric to measure agency within food systems using data related to food security and dietary diversity among female rice farmers in southwestern Burkina Faso. Informed by the literature on agency, autonomy, and food sovereignty, we developed an agency index based upon a subset of questions in the Household Food Insecurity Access Scale (HFIAS). We call this metric the Food Consumption Agency Metric (FCAM). By applying the FCAM, we then sought to understand how agency complements our understanding and measurement of food security. In exploring the relationship between agency and food security more broadly, we found that agency can be quantified in a way that increases its accessibility to policymakers to create more just food systems and expand how we approach food justice.

1. Introduction

Famine, hunger, and food insecurity remain global challenges despite decades of intervention. Current estimates suggest that “between 720 and 811 million people in the world faced hunger in 2020” [1]. Africa bears the brunt of this issue with the highest prevalence of hunger for any world region. As of 2019, an estimated 234 million Africans living south of the Sahara were chronically undernourished, and 250 million people, nearly 20% of the population, experienced hunger [2]. Although these figures fail to distinguish between total caloric versus nutritional deficits, there is no question that food insecurity remains a pressing issue for the continent. With these grim statistics on the rise, philanthropic and international aid measures have attempted to address the problem.

Food security means having “physical, social, and economic access to sufficient, safe, and nutritious food to meet dietary needs for a productive and healthy life [at all times]” [3] (p. 5). A person is food-secure when they live in neither hunger nor fear of hunger [4]. To better combat food insecurity, scholars and policymakers frequently break it down into its component parts or dimensions. One can think of the dimensions of food security—availability, access, utilization, stability, sustainability, and agency—as legs of a stool. All of the legs must be sturdy if policymakers are going to work on them individually and on food security as a whole, lest people fall off a wobbly stool with weak legs. Agency is increasingly understood as a critical dimension of secure food systems [5] and refers to individuals and groups’ power to shape what and how they farm as well as their control over the types of food they consume [6] (p. XV). Those with limited agency within a food system may experience malnutrition in different forms, from acute food insecurity to micronutrient deficiencies to obesity. The fundamental right to food is outlined in the UN charter, suggesting that it is essential that people have the power to operationalize and manage their food production and consumption [6,7,8]. Furthermore, beyond its inherent value, understanding different actors’ roles and agency over food systems increases our ability to create more equitable food systems [9,10]. It has been shown that the more control one has over their food system, the better their dietary outcomes are, especially when considered under political ecology frameworks [1,11].

Agency shapes the contours of various food systems and human–environment interactions; it permeates almost every dimension of food production and consumption. Picking what to grow and how to grow it as well as determining what to eat, where it comes from, and when we get to eat it are all essential aspects of agency. The HLPE defines agency as “the capacity of individuals or groups to make their own decisions about what foods they eat, what foods they produce, how that food is produced, processed and distributed within food systems, and their ability to engage in processes that shape food system policies and governance” [6]. As such, it is essential that individuals have the power to operationalize and manage their food production and consumption. Agency over food is therefore an integral component of food security, an idea forwarded further by the food sovereignty movement [12].

Food sovereignty is the right to healthy and culturally appropriate food produced sustainably by agricultural systems determined by the farmers themselves. The aim is to place the hopes and needs of those who produce, process, and consume the food at the heart of food systems rather than the demands of corporations and markets [13]. Often, this movement seeks to repoliticize the “how to feed the world” debate by calling for actor-oriented and informed perspectives and methods. In doing this, the aim is to overcome the observed essentialization and neglect of individual people [12]. Such reforms will often call for increased control of food production and consumption and establish alternative food networks to support local needs in the struggle for food justice and sovereignty [14]. Furthermore, the food sovereignty movement has offered solutions that marry the larger political structures’ offerings with local needs and desires, again highlighting how increasing food sovereignty can increase agency via cooperation with larger entities [12].

Although complete agency or perfect sovereignty may not entirely eradicate every food security issue, it has the capacity to minimize them significantly, and as such, it should be a top priority for those seeking to increase food security around the world [15]. The focus on agency complements the already existing research on agency regarding food production. Agency over production in agriculture concerns itself with an individual’s right to choose what to grow and how to grow it [13]. With the continued food crises in Africa resulting in part from growing crops not for local dietary consumption, attention to this aspect of agency continues to gain priority [12,16]. Agency over production is a concern related to food and nutritional security that has been explored [17]. With production often overshadowing it, agency over consumption has received less attention [18].

Separating consumption from production is an important step in understanding how agency can be integrated into food security metrics. Specifically in Africa, consumption is at times overlooked [19]. The relevance of consumption rests in the extremely wide range of needs and desires, and on the necessity to adapt the goods available to local preferences. With few exceptions—cloth and beads, for example—the localization of commodities in Africa grew through the consumers’ own agency [19]. This agency has been consistently disregarded such that finding ways to measure it remains pressing.

This broader pattern of large-scale production without consideration for local consumption holds true for the agricultural sector as well, and it ultimately created the food and nutrition crisis hitherto described. This crisis has resulted in a lack of food more generally and a lack of nutritious food more specifically [20]. First, these agribusiness models emphasize the production of non-consumable products, such as cotton in Mali, or foods meant for export, such as rice and peanuts in Western Africa [13,21]. This has resulted in both a lack of food and decreased access to the food that remains [22]. Additionally, undernourishment remains an immense issue [23]. More than 250 million undernourished individuals live in Africa, where the number of undernourished is growing faster than anywhere else in the world [24]. With the hunger and nutrition crises continuing, considering agency within food insecurity is essential, as it will allow us to understand what individuals hope to change in the system beyond securing adequate food. Although completing caloric requirements is undeniably a higher priority, allowing individuals to consume culturally appropriate foods and the foods they prefer is key to enacting individual agency over food consumption.

At some point, agency cannot account for a complete lack of resources; however, when unable to listen to hunger cues or forced to give up food for other family members’ preferences, individuals run the risk of experiencing undernutrition [25]. Furthermore, ensuring culturally appropriate food will be integral to increasing access to safe food meant to combat food insecurity [25,26]. Although the HFIAS questions did not explore this question of food preference, it is an important component of it, which is essential to mention [25]. Because increasing agency has the capacity to aid in food insecurity reduction, it is key to consider it as a component of food security; however, in doing this, the question of how to establish a metric arises.

Although agency is increasingly viewed as a fundamental dimension of food security, only limited conceptual and methodological work has been done to measure it [6,12]. Our focus in this paper was to propose and test a measurement of agency over food consumption. This metric, which we call the Food Consumption Agency Metric (FCAM), was built on the responses to three of the survey questions from the widely used Household Food Insecurity Access Scale (HFIAS). We sought to demonstrate that agency is not just measurable, but that it is relatively easy to do so with existing data. More specifically, we sought to answer the following questions in this paper:

(1)

How can including agency as a dimension of food security inform our understanding of food security?

(2)

How can we tangibly measure agency?

(3)

How does agency influence food security and dietary diversity?

These questions are driven by the academic and policy literature highlighting the importance of the Right to Food, agency, and food sovereignty when addressing food security [6,12,27]. To answer these questions, we first considered the relevant debates related to food security, agency, and metrics. Based on those conversations, we created and tested the FCAM. Although this is the first time we have tested the FCAM, because the measurements are based off of HFIAS data, we believe the scores themselves to products of reliable data. Until more studies use the FCAM and produce findings for us to compare, we will not be able to speak to its reliability in measuring agency; however, we believe that it is reliable, as it measured agency over food consumption and procured worry and preference subscores as intended. We hypothesized that an increase in agency score would correlate with increasing income and dietary diversity scores. We then explored the actual findings related to FCAM’s application and the academic implications and policy recommendations associated with them.

2. Study Site and Methods

This section provides contextual information on the study area, Burkina Faso, an explanation for why we chose this research area, and a discussion of research methods. Burkina Faso has a largely tropical climate with two very distinct seasons: rainy and dry. In the rainy season, the country receives between 590 to 885 mm of rainfall per annum [28]. The rainy season lasts approximately four months from May or June to September. There are three climatic zones in Burkina Faso: the Sahel, the Sudan-Sahel, and the Sudan-Guinea. The villages from where data was collected in this paper are situated in the Sudan-Guinea region, not too far from Bobo-Dioulasso, Burkina Faso’s second largest city. Figure 1 below shows the location of the five study villages relative to each other and the research base.

About 80% of Burkina Faso’s population is engaged in subsistence and cash crop farming [29]. Not only is most of the population of Burkina Faso dependent on agriculture as a source of income, but they also rely on the agricultural sector for food to directly feed the household [29]. Due to the vulnerability of agriculture to rainfall variability and climate change, more and more families are having to look for other sources of non-farm income, and they often have to travel outside of their regional zone to find work [30,31,32]. It is important to note here that livelihood systems have changed to fit export systems [33]. Many of the original agricultural methods which helped manage climatic risk for these populations are now moot, as they produce cotton for export and use farming methods insensitive to the ecosystems they find themselves in [32,33]. All of this is exasperated by the worsening effects of climate change, which continues to impact Sahelian regions in the face of land degradation and water shortages, among other climatological challenges [33].

Concerning the choice of the study region itself, the second author had ongoing research in southwestern Burkina Faso regarding the Green Revolution for Africa (GR4A) approach and projects such as the BRICOP rice farming project [34]. Figure 1 further distinguishes BRICOP and non-BRICOP villages [35]. Using food security and dietary diversity surveys, which are discussed below, he collected household level data between 2016 and 2020 to assess the success of the BRICOP project and how the market and gender intersect in human–environment interactions [34]. Due to our limited ability to conduct new fieldwork during the COVID-19 pandemic, we developed this paper’s metrics and analysis with that data. Aside from the standardized Household Food Insecurity Access Scale (HFIAS) and Household Dietary Diversity Score (HDDS) surveys, which increase the replicability of the study, the research sites in Burkina Faso are somewhat typical of the other rural, agrarian communities in the Sudano-Sahelian zone of West Africa. Exploring women’s agency over their food consumption in this context is important given that these are some of the poorest and most food-insecure regions and that women have historically played large roles in these rural food systems.

The individuals surveyed for this study were randomly selected from five different villages in southwestern Burkina Faso, as shown in Figure 1 [35]. In consultation with our local research collaborators, the second author selected these five villages from a larger sample based on certain criteria such as safety, travel time, and whether they participated in a GR4A rice project. Once identified, the women farmers to be interviewed were randomly selected (using a random number generator). They were orally surveyed face-to-face in Mossi, Dioula, or Guani by the second author, his students, and/or Burkinbe research assistants between 2016 and 2020. This included baseline surveys in 2016 and four waves of food security and dietary diversity surveys administered between 2017 and 2020 to the same women. Throughout the paper, the women interviewed are compared to themselves over time as well as across wealth groups.

The interview results from 130 participants were used in the analyses of this paper. To estimate household and individual dietary diversity, the field research team (which included the second author) used the United Nations Food and Agriculture Organization’s (FAO) household dietary diversity score (HDDS) survey, a useful indicator of nutrition among similar populations in Burkina Faso and other countries in the past [22,36]. This survey asks participants to recall all of the foods they consumed over the course of the previous day, including the ingredients used in each meal. The presence and absence of foods within certain food groups—also designated by the FAO—were used to create a dietary diversity score. It is important to note that this measure counts the number of food groups from which ingredients were eaten, not the quantity that was eaten.

Interviewees were invited to participate in the study through communication with local leaders who held different roles in each particular village. The field surveys had IRB approval, participation was completely voluntary, and the participants’ anonymity was protected. Participants were asked about their age, wife status, and the number of children in their household, as shown in

Table 1. Whole Study Summary Statistics [39].

		Medina Koura	Saki	Seguere	Siniena	Yeguere	Total Sample
Age	Pre-Menopause	84%	67%	86%	46%	69%	70%
	Post-Menopause	16%	33%	14%	54%	31%	30%
Wife Status	First	41%	33%	57%	25%	21%	32%
	Second	34%	39%	43%	42%	17%	32%
	Third	9%	11%	n/a	8%	5%	7%
	Fourth *	6%	n/a	n/a	n/a	n/a	2%
	Monogamous	9%	17%	n/a	25%	57%	28%
Children in Household	Zero	3%	5%	n/a	21%	2%	6%
	One-two	6%	17%	21%	21%	21%	17%
	Three-five	47%	50%	50%	50%	50%	49%
	Six plus	44%	28%	29%	8%	26%	28%

* Many families did not have a fourth wife, so some fields are not available.

, to guage their social roles and familial obligations. Participants were also asked about their agricultural activity and crop sales. Lastly, participants were asked about the large goods that they owned both individually and within the household, including forms of livestock and agricultural equipment. This information was used to create approximate household and individual women’s wealth statistics in West African CFA Francs based on the market prices for these goods [37]. This resulted in three strata of wealth groups: high, middle, and low. Participants were then asked a series of questions in the HFIAS about their perceptions of their level of food security over the previous four weeks. This, again, was done by using a standardized set of questions developed by USAID [38]. Questions were aimed at gathering information on uncertainty, anxiety, and worry about food supply, food quality, insufficient consumption, and the associated physical discomfort.

In terms of the metric construction, the questions we used for FCAM (listed below) were asked as part of a HFIAS survey administered four times to the same women over the 2017–2020 period. These survey questions received a response score of 1–4. Women were surveyed once in the rainy and dry seasons in 2017 and then again during these two seasons in 2019–2020. We took the answers from the rainy and dry seasons in 2019 and 2020 in order to test the metric by comparing the patterns or lack thereof between 2017 and the 2019–2020 seasons. Additionally, later in the paper, we consider how the agency scores align with the dietary diversity scores of each individual to see if there is any correlation between agency and dietary diversity scores from the HDDS surveys. Much like the food security surveys, the dietary diversity scores are also the results of the surveys conducted in the same manner as described previously. The individual case profiles are also based on survey questions related to the HFIAS and HDDS.

We designed the FCAM to standardize how to measure individual agency and power over food consumption. The metric measures agency by summing the frequency of events indicative of a lack of agency and appropriately weighting them to produce a final score. The FCAM is composed of two components: one considers preference, and the other concerns worry. The aim of the first component (derived from two questions) is to determine if the individual can eat what they desire and if they are being forced to eat what they do not like [40]. The other component concerns worry. This second aspect of agency over consumption relates to whether an individual is confident that they can provide enough food for themselves [12]. It accounts for a third of the final score.

We chose to consider preference and worry as the main components of agency based upon how agency is typically conceived in the food security literature [40]. For example, access to culturally appropriate foods, which people generally prefer, is largely considered essential to full agency over consumption [12,40]. In fact, forcing individuals to consume inappropriate foods is considered a human rights violation [41].

Working from Sen and Nussbaum’s capability approaches, we used the HFIAS questions available that most appropriately indicated control over the relevant capabilities, such as control over one’s environment, health, food, and fear of security in the future [42,43]. In doing so, we treated empowerment as “the ability to freely choose one’s own path in life in accordance with one’s distinctive talents and abilities” [44]. We applied this to food security specifically by highlighting the ability to make choices which meet one’s preferences and minimize worry. Is one able to eat the food they prefer? Are they forced to eat undesirable foods out of desperation? Does one have so little agency over their diet that they are constantly worrying about where their next meal will come from? These all feel like important dimensions of food consumption agency that are captured in the HFIAS questions we used to develop the FCAM index based upon past literature on food security and agency more broadly.

For these reasons, we used the following three specific HFIAS questions and their corresponding answers in the FCAM index.

• During the past four weeks, were you or any household member not able to eat the kinds of foods you preferred because of a lack of resources?
• In the past four weeks, did you or any household member have to eat some foods that you really did not want to eat because of a lack of resources to obtain other types of food?
• In the past four weeks did you worry that your household would not have enough food?

The first two questions specifically look at food preference, and the last addresses worry. If one has control or agency over their diet, then they are able to consume the types of food they prefer and to avoid eating those food items they find distasteful [17]. In contrast, being forced to eat something one does not want is an indication of a lack of agency [6,12,40,43]. This aspect of agency clearly declines during periods of acute food insecurity. For example, some famines in the Sahel have been accompanied by the distribution of grains that local people found distasteful, such as red sorghum from the United States, or the consumption of famine foods that are only consumed in periods of desperation [45]. As for the third question about worry, anxiety about where your next meal will come from relates to how much power one feels over future food consumption. In rural Burkina Faso, this worry may be somewhat seasonal, occurring more frequently in the hungry season before the next harvest, and less often during the month after the crops have been harvested [31]. In more urban contexts, people may worry more about the next meal at the end of the pay period when money is running low [32]. In some, our interviews revealed, and the responses to these three questions helped us measure, the fact that many farming households in rural Burkina Faso would feel greater power or agency over their diets if they knew they would have enough food to eat throughout the year, that they had the ability to vary their diets, and that they were not forced to eat certain foods out of desperation.

The numbers we used to calculate the score are numerical responses corresponding to questions selected from the Household Food Insecurity Access Scale (HFIAS). Individuals were asked the questions, and their responses were recorded on a scale of one to four, where one was “no” or “zero times,” two was “rarely” or “one or two times,” three was “sometimes” or “three to 10 times,” and four was “often” or “more than 10 times.”

Because HFIAS measures insecurity, a negative feature, the lower the score, the better. Here, HFIAS measures a positive feature, agency; thus, to make the scores more intuitive, we manually rescaled every HFIAS 4 to an FCAM 0, HFIAS 3 to an FCAM 1, HFIAS 2 to FCAM 2, and HFIAS 1 to FCAM 3. This conversion did not change the relative value of each HFIAS score. It simply allowed us to have more intuitive FCAM scores. Once we converted the HFIS scores, we averaged scores from the three survey questions for each participant to produce their final FCAM scores. Because two of the questions concern preference and one concerns worry, the preference subscore accounts for 66% of the score, and the worry accounts for 33%.

These numbers between 0 and 3 represent an individual’s agency score over consumption by measuring how often they did not eat their preferred foods, had to eat food they did not like, and worried that they would not be able to eat when they wanted to. Next, we took these composite scores and interpreted them using the FCAM scale: a scale from 0 to 3, where zero is low agency and three is high agency. More specifically, the ranges are as follows: 0–1.5 is low agency, 1.5–2.5 is medium agency, and 2.5–3 is high agency.

We chose to break the range down in this way—not along even breaks or equal intervals—because the effects of a lack of agency tend to compound [17]. The difference in the amount of agency between a 3 and a 2.5 are as extreme as a 0 and 1.5, as illustrated in the case studies. Furthermore, the causes of agency are often highly connected, and failing to consider how they compound by using natural breaks would not do justice to the experiences of each individual. In fact, 66% of the score is determined by a lack of preference, and the answers to these questions tended to grow and shrink together, suggesting that the factors at play are connected.

We recognize that FCAM is a relatively simple metric based on three questions that do not cover all aspects of food system agency. However, we would argue that this simplicity is one of its strengths. Given that food security studies using the HFIAS are widely used across the globe today, and that much of this data is publicly available, the FCAM would be relatively easy to replicate using these existing data sets as well data that is collected on a recurring basis [25]. Although one could certainly imagine more sophisticated models of food system agency, using them on a widespread basis may be cost-prohibitive.

3. Results

3.1. Agency and Wealth Groups

Although the wealth group categorization had the weakest correlation and variances across the three variables, it was the most consistent with our hypothesis as seen in Figure 2. The correlation between economic groups and the agency scores were higher during the rainy season, positively correlated during each season, and were statistically significant. As income went down, so did the agency score.

As for the strengths of the correlations and variances themselves, they were, again, stronger during the rainy or hungry season, as shown in

Table 2. Wealth Group Predicting FCAM Regression Summary [39].

Wealth Group → FCAM Regression Summary	Dry Season	Rainy Season
Correlation	20%	27%
Variance	4%	7%

. The movement across seasons was consistent with what we expected.

During the rainy season, when grain was depleting, we suspected that an individual would feel their lack of wealth more acutely, causing their worry to increase and their “preferences satisfied” to decrease. We compare this to urban families living paycheck-to-paycheck who feel more financial pressure at the end of a pay period. This means agency scores would decrease overall for the rainy season, especially for poorer women, which the statistics here reflect. This was also true for the results from the dry season. Although an individual’s wealth group did not shift up during harvest season, their access to both income and food was higher during this time. As we expected, the correlation between agency score and wealth group was positive during this season, and the intensity of the correlation, or lack thereof, indicates the apparent preference for food available during the dry season and decrease in worry relating to seasonal income rather than the wealth categories individuals fell into. Although there were a few exceptions, it is important to note that a higher wealth group did not always bluntly correlate with higher agency, which contradicts our hypothesis.

3.2. Agency and Dietary Diversity

Dietary diversity and agency scores had moderate positive correlations across both seasons. This means that as dietary diversity went up, so did agency. The subscores, or parts of the agency metric, reveal even more interesting details about these results. When we examine the agency score more closely, it is interesting to note that both the preference and worry subscores went up during the rainy season. Despite the way the subscores seemed to work “against” each other during the two seasons, they can be explained by the food acquisition practices we see during the rainy season in particular. Agency is at its lowest during the wet season, when dietary diversity is at its highest. Dietary diversity is higher during this time because the rains result in the flourishing of local flora and higher biodiversity in the landscape [46]. This, in turn, results in more foraging and higher dietary diversity, but it also indicates a high degree of worry because what someone will eat and how much one can find will vary every day as we see in

Table 3. Dietary Diversity Predicting FCAM Across Seasons [39].

HDDS → FCAM Regression Summary	Dry Season	Rainy Season
Correlation	28%	45%
Variance	7%	20%

. Even though these foraged foods seem to meet the preferences of individuals who rely on them, this does not alleviate worry about insufficient calorie consumption. During the dry season, access to locally available biodiverse foods decreases. As a result, grains are more prevalent; thus, dietary diversity decreases, but there is less worry about having enough to eat. The increased income from grain sales also increases access to preferred foods from the markets, although apparently not to the same extent as foraged foods. Preferences related to foods at the market versus food in the wild are explored further in the case study section.

Preferences being met more during the rainy season calls into question assumptions about the nutrition transition. Part of the nutrition transition describes the shift in food and drink consumption patterns that accompany economic changes [47]. The model is often used to describe the transition in developing countries from traditional diets to more Western patterns of consumption, with diets higher in sugars, fats, and processed foods [47]. In this instance, we saw a preference for the more traditional, foraged foods even when participants were in the higher income bracket. The subscore related to preference indicates that individuals would rather consume traditional foods even when they could do otherwise. What is fascinating about this is that people do not seem to be foraging just as a last resort, but instead because it is what they prefer, even when finances would allow them to purchase processed foods.

Furthermore, the results demonstrate that agency is positively correlated with dietary diversity scores and increased income, as seen in Figure 3, indicating that it can add an extra dimension to our understanding of these food systems. Given that it was negatively correlated with the food insecurity score, meaning increased agency resulted in increased food security, agency does apparently have a distinct relationship to food security and food systems. Because agency does not exist in a vacuum, assessing it alongside the other dimensions of food security would further elucidate this complex topic.

What we found most interesting about these results was the role of worry, a component or subscore of the FCAM score. Does an increase in wealth alleviate worry? The case studies below seem to suggest so. Because wealth in an increasingly monetized economy can provide greater certainty over when one’s next meal is coming, food security in this type of economy increasingly relies on money [48]. For the three women in the high-income bracket in our data set, they all indicated low worry across their agency scores, and if the scores decreased, it was linked to their preferences not being met. Again, there is a gap in the data here demonstrating their specific tastes, but the larger trends demonstrate that they prefer foods acquired in the rainy season. This raises interesting questions about how preference does or does not change as wealth increases.

Another interesting piece of this puzzle is related to the role of preference when the preferred foods decrease dietary diversity. For example, Maggi cubes, a seasoning substitute packed with MSG, are becoming a commonplace ingredient, but they lack many of the micronutrients found in traditional seasonings. That said, they taste better for some of these women, and they prefer to use them because they save time and/or may be cheaper. Understanding how always meeting preference could be detrimental to nutritional health complicates the impact of agency in the food security equation.

3.3. Case Studies

This section of the paper explores three case studies of individual women that help illustrate some of the issues at play in the statistical findings for the larger data set. It reconciles and explains some of the patterns seen across the larger study population by engaging with the specific foods eaten. It also considers how economic circumstances across each season affect the corresponding scores for individual women. We also look closer at the details concerning food preparation, the repetition of certain meals, and food acquisition techniques.

To understand the results of the case studies, some background related to traditional food consumption patterns is necessary. For the women in this study, a typical noon or evening meal would consist of some kind of grain or starch (typically maize, sorghum, or rice), which contains the macronutrients necessary for the day, and a sauce packed with more substantial foods such as vegetables and nuts, all rich in micronutrients (e.g., baobab leaves, peanut butter, hibiscus leaves, and fish meal). Within the family, the men are largely expected to provide the staple grain, which is often maize, millet, or sorghum [46]. Although these case studies focused on women who farm rice, this grain is not regularly eaten by most rural people in the study area. Sometimes, it is eaten on market days or by the better-off or for a special celebration such as a wedding (this differs from urban areas, where rice is eaten more regularly). Besides grains, men are also often responsible for providing meat for the household. When meat is served, which is rare, this is typically in the form of chicken or fish, which can be fresh or dried and crushed into a powder for flavor. The women are responsible for the sauce. Whether they forage, grow, or purchase those ingredients varies [46]. Common sauce ingredients include peanuts, cowpeas, baobab leaves, and cultivated or foraged vegetables.

Tea and coffee may be consumed throughout the day (although mostly in the morning) or alongside meals, but for the most part, water is the dominant beverage consumed. Western foods and beverages are also beginning to change the traditional meals eaten. Although many of these foods reduce cost and may taste better, they often come at the expense of nutritional value. As noted earlier, Maggi bouillon cubes are being added to many sauces. These kinds of foods are often bought at the market to supplement the flavor of foraged and grown foods. Additionally, some of these ingredients are easier to prepare or to add to recipes. In the case of Maggi, it is a seasoning cube that is simply quicker to include in a sauce, especially compared to something like soumbala, which requires hours of foraging, fermenting, and processing to develop the desired flavor.

Understanding how staple and newly introduced foods are produced and consumed yields more insight into how food acquisition and consumption factor into levels of individual and community food security [47]. As many individuals in this case study rely on a mixture of purchased, foraged, and grown foods to prepare their meals, the variety of acquisition techniques as well as seasonal variability can greatly affect the food security of an individual.

The aim of this section is to demonstrate how agency can increase our understanding of food security and how it manifests itself in terms of food security and dietary diversity. To do this, we explore three case studies of individuals that are meant to consider the different facets of our test groups and how each of the variables interact with each other. We explore three individuals, one each with a high, medium, and low income. We used a pseudonym for each woman.

3.3.1. High Income: Miriam

Miriam is an upper-income individual with no co-wives. She does all of the food preparation for her household by herself daily. Miriam had the highest possible agency score across both seasons (scored a 3) as well as relatively high dietary diversity scores across both seasons (6 in the dry season and 10 in the wet). Her food insecurity score was low in the dry season; however, she was less secure during the wet season (1 and 3, respectively). She fits into the observed pattern of security decreasing during the rainy season, whereas diversity grows conversely. The stability of her agency reflects that access to income helps to minimize worry. It also appears that her preferences are being met across both seasons, leading to these consistently high agency scores. She is the only individual in the high-income group to score a 3 for FCAM, the highest possible score, across both seasons, as seen in

Table 4. High Income Case Study Summary Scores [39].

Miriam
	Dry Season	Rainy Season
FCAM	3	3
HFIAS	1	3
HDDS	6	10

. We were interested in this occurrence; thus, we picked her for analysis.

During the dry season, she had coffee with sugar and bread for breakfast. For lunch, she had corn toh (like corn polenta) with hibiscus sauce with onion, tomatoes, and soumbala, all cooked with cotton oil and seasoned with salt. For dinner, she had spaghetti with tomato sauce, again cooked in oil and seasoned with salt and Maggi. During the wet season, she reported eating rice with a peanut sauce containing onion and smoked fish, seasoned with salt and Maggi, and drinking coffee with sugar for breakfast. For lunch, she had corn toh with baobab sauce, containing onion, soumbala, Maggi, salt, and fish powder. She ate shea fruit as a snack, which is often foraged. Lastly, for dinner, she had oily rice (riz au gras), which had onion, fresh fish, and cabbage in it, and it was seasoned with Maggi and soumbala. The presence of pasta and rice in her diet is more common among women with higher incomes. Note also the presence of Maggi in her sauces.

Considering the differences between what she ate over these two days, it is clear that she had higher dietary diversity in the wet season, as we expected. The food that increased her diversity was likely foraged [47,49]. Additionally, based upon her agency scores, she reported eating what she preferred when she wanted frequently. She did not report having to eat what she did not want. Even during the hungry session, her worry levels remained low; thus, she demonstrated very high agency across both seasons. As her scores are consistent across both subscores, we believe that she is able to meet her preferences with relatively low stress consistently. When considering these results in light of her food security scores, it is interesting to note that she was less food-secure during the hungry season; however, this did not impact her agency levels. Based upon the other interview and survey data, she attributes this insecurity to the inaccessibility of certain foods.

3.3.2. Medium Income: Aminata

Aminata had medium levels of agency and medium income. She did the food preparation by herself five days a week on rotation (five days on, five days off) with help from her co-wife during both seasons. She described the kitchen as shared, and women may cook outside of their rotation should they so desire. Aminata’s agency was higher in the dry season than the wet, scoring a 2.33 and a 2.67, respectively, following the pattern of the whole group. Her food insecurity stayed relatively high, with a 3 across both seasons, as did her dietary diversity, which was an 8 in both the dry and rainy season, all summarized in

Table 5. Medium Income Case Study Summary Scores [39].

Aminata
	Dry Season	Rainy Season
FCAM	2.33	2.67
HFIAS	3	3
HDDS	8	8

. Although these did not fluctuate as much as expected, they were at the degree we expected for her wealth group and agency score classifications.

Her food categories align with what we would expect given the foraging practiced in the region. For breakfast during the dry season, she had rice porridge with shea butter. For lunch, she ate corn toh with potash and a dried baobab sauce containing soumbala, dried fish, salt, and hot peppers. For dinner, she had leftovers from lunch with an additional sesame sauce containing dried fish and salt. She had cashew apples as snacks in the morning and afternoon. As a morning snack, she had Bambara nuts with shea butter. During the wet season, she ate corn toh with potash and kirikiri sauce containing peanut butter, Maggi, salt, and hot pepper, as well as Lipton tea with sugar, and wheat bread. For lunch, she had rice with a peanut butter sauce containing onion, Maggi, salt, and dry fish. Finally, for dinner, she ate corn toh with potash and with baobab sauce containing dry fish, Maggi, piment, onion, soumabala, and salt.

Aminata’s case is largely unsurprising. Her scores and the foods she ate align well with what we would expect an individual in her position to experience. She has a high proportion of foraged foods in her diet, and this is reflected in her dietary diversity score as well as in her worry subscore. She had relatively high dietary diversity scores. Furthermore, her worry score went up during the rainy season as we would expect to find for someone who is foraging more during the hungry season. It is interesting to note that she had preference for the food she was eating in the wet seasons, as her preference score was higher in the dry season, suggesting she was not able to eat what she wanted to as frequently during the dry season, like the majority of the study population.

3.3.3. Low Income: Yasmine

Yasmine was the only individual in the low-income classification with a high agency score. Understanding why her scores are high may shed light on useful adaptations to seasonal variation and food security concerns. It is also important to note that her score in the wet season switched to medium agency, showing just how precarious this can be. Yasmine is an apparent exception to the wealth group indicator as the only woman to have a high agency score in the low-income bracket. Her FCAM scores were 3 and 2 in the dry and wet season, respectively. In the dry season, she was more food-secure with a 3 and increasing to 4 during the wet season, as we would expect. Her dietary diversity scores also followed the expected lines, going from a 7 to a 9 from the dry to wet season, respectively, the summary of which is shown in

Table 6. Low Income Case Study Summary Scores [39].

Yasmine
	Dry Season	Rainy Season
FCAM	3	2
HFIAS	3	4
HDDS	7	9

Aside from the high agency categorization, Yasmine follows the trendline closely.

The food preparation questions indicated that there were multiple women in the household and that they each prepared food for themselves. In a typical day of eating during the dry season, Yasmine had leftover maize toh with a sauce of tomatoes, peanut butter, dry fish, salt, and sorrel for breakfast. For lunch, she had something similar: maize toh with a sauce of tomatoes, peanut butter, dry fish, and sorrel again. Her dinners consisted of rice with a cabbage sauce that had onion, smoked fish, salt, and oil in it. During the wet season, Yasime had maize toh with a baobab sauce, containing soumbala, dry fish, and salt for breakfast. She had rice with hibiscus sauce containing peanut butter, onion, soumbala, fish, and salt for lunch. Lastly, for dinner, she had corn toh with potash, with a hibiscus sauce again; however, this sauce had fish, soumbala, and cotton oil in it.

Yasmine’s scores, in tandem with her meal descriptions, highlight the inequality present among those with less income. Her food categories changed only during the rainy season when she was foraging more. Additionally, her agency score itself dropped during the rainy or hungry season. Considering her subscores, she was more worried during the hungry season, as expected, but she also was not eating what she wanted frequently and having to eat what she did not want a few times that month. Part of the decrease in her agency levels can be explained in large part by her preferences not being met during the rainy season and her worry score increasing slightly.

As we considered these case studies, especially in tandem with the results from the larger study, we realized that these case studies elucidate the impact of income on levels of worry. First, while wealth was only a moderate indicator mathematically, it was typically true that higher wealth indicated higher agency, as seen in the wealthy and even medium income individuals in the study. Only one woman in the low-income group had a high agency score, and even then, her agency score was in the medium category during the wet season. Additionally, Hema was one of two women who fell into this category, and she was one of the few women to repeat the same meals over the same day. That said, wealth does not explain everything.

4. Discussion

Agency is increasingly recognized as an essential component of food security. However, because it is difficult to measure, it is often deprioritized or missed in policy debates and programs. By beginning to measure agency, this dimension of food security both becomes more visible and helps policymakers better understand the role of agency in food security. Ultimately, our hope is that agency will inform program design to better address marginalized groups’ needs. In this context, agency is the ability to choose what to eat, when to eat it, and how that food is acquired and grown. By enabling greater agency over food consumption and production decisions, we hope to better address food insecurity on a wide range of scales. We find that improved agency and the integration of local and insider knowledge enhance food security [50,51,52]. Other aspects of our findings are discussed in this section.

Primarily, we demonstrated that agency in relation to food consumption is possible to measure. To do this, we designated preference and worry as the key components of agency over food consumption. Returning to the stool metaphor from the introduction, creating the FCAM allowed us to make that leg of food security more visible and enhanced our ability to measure progress in strengthening agency over food consumption and food security more broadly. The FCAM produced surprising results. We found that preference was met more during the rainy season, when foraged foods are more plentiful.

On one hand, providing more income or money seemed to slightly alleviate worry pertaining to food systems in this case but not as strongly as pushed for by proponents of the GR4A and its theory of change [47,48]. Further contradicting their development narrative is not only the weak correlation between income and preference, but also the fact that preference is apparently more difficult to meet even when access to income is increased. Here, we chose to weigh preference at 66% and worry only at 33% (largely as a function of the three questions we used from the HFIAS survey). Even in the rainy season, when preference was met more, the worry variable, accounting for only a third of the score, was able to trump the preference subscore and decrease agency scores overall. With that in mind, we chose to focus our recommendations on approaching worry first. Even with its lower weight, it is clear that feelings of worry trump respondents’ preferences when considering agency. This is not to minimize the surprising results of the preference scores, but to return to our goal of increasing food security within this new agency framework.

This result has significant academic implications in that it questions previous assumptions about food preferences. As shown in this study, dietary diversity, income, and seasonal variation all affected how agency manifested itself in food security. It appears that there was a preference for more traditional foods in general, even among wealthier groups, when we expected to see preferences met more during the dry season. The harvest season is when income is highest; thus, access to processed and typically more desirable foods increases. However, in our study, participants did not seem to prefer these products. The typical pattern we see in the nutrition transition literature suggests a preference for packaged and processed foods as income increases [53]. However, what we saw in this case was that even those in the higher wealth category still had their preferences met more frequently during the rainy season when foraged foods were more available [39,53]. Acknowledging that Burkina Faso is still a low-income country, these findings challenge present understandings of nutrition transition at the micro level and calls assumptions about preferences evolving with income into question [48,54]. During the dry season, the ability of a woman to acquire food sufficient for survival is higher. Furthermore, her ability to acquire foods at the market, as opposed to foraging, is also higher. For example, we see this with the consumption of Maggi cubes, which are a processed seasoning but are also likely labor-saving, especially compared to traditional seasoning methods. However, beyond this seasoning, it was rare to see women eat foods in the salty and sweet snacks categories. It seems that most of the time, they were not eating processed or packaged foods, rather relying on foraged and grown foods to meet their preferences. The way preference is met more during the rainy season also raises some interesting questions about how income relates to preference and taste that would be interesting to explore down the line [54].

This contradiction to the nutrition transition and larger development narrative impacts our policy implications, and in light of the low correlation between wealth and agency, increasing income alone is not the answer to increasing food security. In fact, if we considered agency as the sole measure for food security, we would see a strong contradiction with current, standard policy recommendations in the food security arena [48]. Now, we do not consider agency to be the holy grail of food security measurements, but introducing it to discussion decreases the emphasis on increasing production to increase income and, in turn, food security [47]. Typically, recommendations would call for more agricultural production to increase what farmers can sell to accrue more income [39,47]. The fragility of these models was made clear in the face of COVID-19 [16]. The goal is not just to find a steady supply of healthy food but to also find preferred foods. As we see here, income is not necessarily the only means of achieving that goal.

Policies that increase access to preferred foods will need to account for nutritional needs if they are truly going to address unjust food systems [55,56]. Additionally, understanding the way socially constructed gender norms play out in human–environment interactions further informs this. The need for culturally sensitive and ecologically aware policies cannot be removed from this equation [12,40]. Although the growing tendency for market-based solutions to poverty currently spearheads the process of agrarian modernization, this finding seems to suggest that it is not as impactful as once thought [54]. Increased income does not necessarily equate to increased access; thus, even if income were to remain stable or increase for these women across both seasons, it is not likely that this would produce the desired effects on the agency scores unless the availability of those preferred foods also increased. In conclusion, the development of this metric revealed surprising results. They call into question our assumptions about preference and how we can alleviate worry for populations such as those represented in our sample.

The use of FCAM invites other studies to consider whether our results are true across the globe. Because this is its first application, we cannot speak to its validity with certainty. We believe that FCAM is constructively valid, as it measures agency over food consumption, but we cannot yet assure content validity until the metric is evaluated further. On its face, we believe FCAM measures agency over food consumption, but again, until more case studies are run, we cannot state with certainty that it produces valid agency measurements. We are especially interested in how an econometric theory could further explain the surprising results we find here, and we hope to add this theory to the arsenal of analytical tools to explore the results pertaining to the worry measure in particular and how it can be further quantified and integrated into policy recommendations. If these results hold true under the application of this theory, a notable change in policies increasing food security may follow.

Although it seems that increasing access, and not purchasing power, plays an integral role in increasing agency and feelings of empowerment, we want to see FCAM applied to other contexts before we state with conviction that the policy measures at play need changed. It seems that these results contradict the traditional theories of nutrition transition, and they may leave us uneasy at first, but as we consider the long-term implications of constructing more just and equitable food systems, these results are but a drop in the bucket of positive food policy changes for the future. This research was limited due to COVID-19 restraints, and thus, we urge others to test FCAM and see if they yield similar results. If there is larger body of data analysis suggesting that agency over food consumption can be increased by atypical policy plans, this could have a substantial impact on food security programs around the globe. Although increased income decreases worry, it did not necessarily increase preference, and thus, we are curious to see if this is unique to our case studies or true across other studies. Until other researchers implement FCAM, its results will be limited. Our hope is that the results of our study will spur others to measure agency over food consumption and see if they produce similar results. If that does occur, then the legibility of agency and ensuing policy recommendations will need to change dramatically.

The Burkina Faso case study was based on female farmers in the southwestern part of the country. Using this data to see if certain factors beyond season, wealth, and dietary diversity affect agency makes us question if these variables can further positively transform our understanding of food systems. In focusing on worry, increasing access to enough food should be the priority; although, it does not seem that increasing income is the most certain way of accomplishing this. Rather, it seems that increasing available food by optimizing subsistence farming programs for food, not for market or export, would be more effective for ensuring accessible food. We need an agricultural development approach accessible to even the poorest of the poor, which involves decreasing purchased inputs. We saw that even those with higher purchasing power had a preference for local foods, not processed foods, and thus, increasing their ability to access those foods consistently will, in turn, begin to address their worry concerns. We believe that households will worry less and eat foods they prefer if they are actually growing food crops that they want to eat themselves. The benefits of development in this direction are twofold. On one hand, it decreases the likelihood of debt from purchasing expensive inputs, such as seeds and fertilizers. It also increases their personal feelings of empowerment as they produce this food.