Adjustments and Compromises of Household Economy Approach in Burkina Faso

The household economy approach is used as an early warning tool to identify areas of intervention and priority groups to benefit from food security support. Adopted in 2012 in Burkina Faso, the application of this tool is still in the experimental phase. After having sketched the national context, from a qualitative survey and using data from targeting surveys, paper exposes questions posed by the development practitioners and realities of application of this tool on a project in Burkina Faso in the region of Boucle du Mouhoun. Implementation of HEA in the field shows a dysfunction. The analyzes reveal a part of methodological subjectivity, the weight of opinion leaders, acting as development brokers and the absence of a complaints committee, an extremely cumbersome mechanism that is unrealistic to initiate. While HEA is seen by communities as a breakthrough, the current trend for practitioners to fill in the gaps is towards increasing the complexity of targeting methods. The initiatives developed require practitioners to maintain a close presence on targeting committees, an unrealistic presence to be deployed in a change of scale in the method and also accompanied by greater bureaucratization of aid.


Introduction
In response to repeated food crises, especially in 2012 when a national emergency was declared for 3.5 million people affected, the government of Burkina Faso published in 2013 a policy document for a National Policy of Food and Nutritional Security aimed at enabling to "all populations have fair access to a balanced, sufficient and healthy diet" (Ministry of Agriculture and Food Security (MASA), 2013). This policy targets in particular the most disadvantaged populations, whose households are mostly in moderate or severe food insecurity.
In perspective of an appropriate targeting of vulnerable populations, development and state actors have adopted household economy approach (HEA), a reference tool applied today at the scale of the region of Sahel (HEA, 2017). HEA criteria are used by humanitarian and development organizations in partnership with government services to target individuals and households to benefit from assistance operations. Originally, in Burkina Faso, the methodology was used in the provinces of the region of Sahel (Oudalan, Soum and Séno) by nongovernmental organizations (NGO), such as Save the children, Action Contre la Faim, Christian Aid, etc. More of 5 years after its application in Burkina Faso and for a better operationalization of the method on the whole of the territory, it is necessary to carry out an assessment, to question the actors, practices, weaknesses and especially to guide stakeholders.
Paper describes HEA with a focus on the experimentation process in Burkina Faso, developing country localised in West Africa and having the lowest human development index in the world (rank 2018: 182/189). Based on a qualitative survey of development practitioners who have implemented HEA, paper provides an overview of the strengths and weaknesses of the tool, along with recommendations. What follows concerns a situational analysis of the realities of the application of HEA in a Gret project in Burkina Faso in a region (Boucle du Mouhoun) strongly affected by food and nutritional insecurity. Data from surveys targeting vulnerable households are used to assess the operationalization of the method, interactions between stakeholders, initiatives taken to improve it and the issues raised.

HEA: quantified categorizations of vulnerabilities
Created in the years 1995, HEA, initiated by Save The Children, is described as a methodology for producing a quantified image of livelihoods by socio-economic group and by zone in order to know strategies of households to access their food and income. HEA entered in Burkina Faso on the period 2010-2011 as part of a European Union project with NGO such as Action Contre la Faim and Christian Aid. Training on the tool was conducted by Save the Children at the African regional level with national focal points, including 6 from Burkina Faso, who then trained other stakeholders from their structures and other interested structures. Institutional anchoring of HEA in Burkina Faso began in 2012 with food crisis. National Food Security Committee was mandated to develop and test HEA in the provinces of Sahel.
HEA in Burkina Faso can be broken down into three main stages: • Division of the country into food economy and livelihood zones: in 2009, Fews Net (Farming early warning systems Network) published, on behalf of Burkina Faso, a map of 9 livelihood zones (ZOME). A ZOME reflects the fact that rural households in the localities concerned obtain their basic survival needs, in particular food and income, in a relatively similar way, with the same market opportunities. • Decomposition of population into socio-economic groups: for each ZOME, profiles of households by socio-economic group were drawn up in 2012. HEA allows them to be grouped into 4 categories of relative wealth. These are "Very poor: VP", "Poor: P", "Middle: M" and "Wealthy: W". • Analysis of access to food and income; and analysis of dangers / shocks and flexibility / response: this involves, for example, showing how a shock impacts families considered to be very poor and way in which they deploy a "coping mechanism": daily work or self-employment allowing them to generate an income to buy survival food, the latter covering almost all of their expenses.

Participatory methodology of HEA
Targeting methodology is a "community" approach adapted from the classic targeting method of Proxy -Means-Testing. But it assesses the socio-economic profiles of households based on HEA criteria, while giving households concerned the opportunity to adjust and validate the analysis criteria. First, it is up to a group of local representatives to assign each household in the community to a defined socio-economic group. Groups are defined in terms of access to land, type and number of livestock, capital, education, labor, etc. Second step consists of a socio-economic decomposition of wealth by group. When NGO and state institutions launched the process in the pilot areas of Burkina Faso, they successively organized: • general assemblies to inform populations and validate the targeting criteria; • establishment of identification support committees and complaints committees. These committees are made up of village heads, councilors, customary leaders, association leaders, neighborhood representatives, etc.; • establishment of draft list of beneficiaries. This is a census of potential households fulfilling vulnerability criteria established by the committees and ascertainment of assets within households; • analysis of census results. Precisely, this involves processing at the data office, selection of households and adjustment of the number of beneficiary households according to available resources.

Weaknesses and amendments of the HEA: views of development practitioners
Discussions with practitioners who have experienced HEA reveal that while the establishment of ZOMEs has made it possible to better target the intervention areas, impartial targeting is difficult to carry out due to a partial census of populations and criteria. selection process chosen on an ad hoc basis by village representatives. This often results in households excluded from targeting, sometimes meeting theoretical admission criteria, when other households are included without meeting these criteria. Statistical work on HEA conducted in the province of Gnagna revealed ineffective targeting to identify the poor with a misclassification rate of almost 50%. So, it is often proposed to modify the targeting criteria and assess their impact (Bedo, 2014).
Several readjustments of the tool have been made in this direction. Catholic Relief Service carried out an exhaustive census of all households in its area of intervention and prioritized the criteria for choosing beneficiaries in a participatory manner. This is confirmed by other actors such as Action Contre la Faim, Christian Aid, National Committee for Emergency Relief and Rehabilitation.
Stakeholders agree that effective HEA targeting relies on an exhaustive census and the choice of unique criteria for the targeted ZOME. On the other hand, additional criteria must complement the basic criteria according to the mandate of each intervention. For Action Contre la Faim, criteria have been introduced in order to prioritize the targets to benefit from nutritional support, that is to say children under five, pregnant and breastfeeding women. Finally, two different committees rather than one were formed in order to refine the targeting and ensure the objectivity of the choices. The field data are then cross-checked with the ZOME criteria. Finally, the holding of the final village general assembly makes it possible to share the results of the targeting carried out by two committees and to validate a final list of households classified by socio-economic group.
However, it turns out that the needs of the populations are often greater than the means available, the field workers sometimes revise the selection criteria of household. Example, if a VP household has a coverage rate of its food needs of 9 out of 12 months and another VP household has a coverage rate of 5, it is the latter that could be used. In other cases, practitioners highlight the possible incompatibilities between the criteria set out by the village targeting committees and the criteria defined for the ZOME to which they belong. Thus, a list of VP established in the region of eastern of the country by a targeting committee contains households with cattle, which should not be the case, in view of the vulnerability criteria retained for said ZOME.
In end, agents share the same circumspect observation with regard to the possible influence of community leaders or ad hoc pre-targeting committees that could direct and bias the results for personal interests or with regard to truncated information in the declarations of certain beneficiaries. In addition, certain criteria are not objectively verifiable during the passage within households (evaluation to "say actor"). Obviously, it is difficult to circumvent the targeting biases linked to the social and political interests of the members of the targeting committees and the strong male representation. This leads practitioners to seek more rigor in the process, by carrying out exhaustive censuses or triangulation of targeting.
These findings led, during the interviews, NGO and state institutions to formulate recommendations for improving the HEA targeting process: • validate the lists by triangulation (visits and interviews); • carry out a comparative study and provide guidelines for a reference methodology; • capitalize internally and then share targeting databases with state services; • take into account the targeting period to avoid the risk of exclusion.
These recommendations go in the direction of a more complex and significant increase in the method (increase in criteria, multiplication of committees, completeness of the census). In practice, how do practitioners adjust these requirements, in the application and adjustment of theoretical HEA methodologies, when they face community pressures as well as time constraints in investigations? What are their room for maneuver, knowing that they are simultaneously experimenting with a new tool and projects? The following point, starting from the realities experienced by the authors of this paper, provides some possible answers.

Experience of HEA operationalization
As part of a project of food and nutritional security implemented between 2014 and 2017 by Gret in the region of Boucle du Mouhoun in Burkina Faso, specifically for support in agriculture and livestock, the use of HEA is part of a participatory approach involving several structures. Three targets were carried out in the intervention area. This is a pilot targeting and two extensions of the first targeting. The project intervention area concerned 2 ZOME (3&4) of the country. Parameters have led to a certain degree of subjectivity in targeting, both on the part of Gret practitioners (project team) and members of the committees, whose integrity and rigor have influenced the results.

Methodology still imprecise
HEA criteria defined for the ZOME are often imprecise and the variables have not been prioritized. Of course, this vagueness allows each institution to adjust HEA according to its specificity. However, according to the ZOME, the method does not provide elements allowing to prioritize some variables. Indeed, each socioeconomic group is determined by ranges of variables. But, from one category to another, data corresponding to a variable can show significant differences. criteria for ZOME 3 & 4 in the appendix illustrate these facts.
For example, in ZOME 3, a VP household has 15 poultry and a P household must have 7, HEA does not indicate how to classify the household with 11 poultry. Also, the border between the 4 categories of households is often difficult to establish, in this case between households P, VP and M. In ZOME 4, a household with 12 members cultivating a total area of 4 hectares, owning 20 poultry, 0 sheep, 4 goats, 1 pig and 1 plow, can be considered as being a VP, P or even M household. These method inaccuracies lead the project teams to classify the households in "subjective" and random categories.

Pilot targeting of villages: weight of producer unions
Because it is still abstract and without issue, first phase of appropriation of HEA by Gret practitioners and its partners went smoothly. It consisted of training stakeholders. Due to the specificity of the project, participants revised the collection tool to be more suited to targeting households. On the other hand, putting it into practice with the selection of first villages (10) for the pilot phase was difficult. It revealed targeting directed by farmers organizations (FO). Public technical services were asked to provide a list of vulnerable villages in the intervention area. The proposal was based on vulnerability criteria: poverty, number of boreholes and number of development partners. But first list, made up of 90% of the villages that are members of the FO, was refused by the latter. In the end, only one village appeared on the list originally proposed by the public technical services and 9 others were proposed by the FO. These officials justified their choice by the "dynamism" of their groups in these villages, whose anchoring would facilitate better involvement of community leaders and strong mobilization of parties, and therefore better support in the implementation of the project. However, among the 10 villages proposed, at least half of these beneficiary villages are the places of origin of the members of the FO executive board. By considering all the beneficiary villages of the three targets and by comparing members of the executive board to the villages of origin, it emerges that all the villages of origin of the members of the executive board were beneficiaries.
Gret's team has opted to follow recommendations of its local partners for several reasons. First of all, HEA remained focused on households, with no upstream selection advice for vulnerable villages. Due to this lack of theoretical framework, Gret, intervening for first time in the area, did not have in-depth knowledge of the specific organizational features and had to make allies. All the more so since in the project, FO were listed as the political "gateway" to the villages and public technical services only played a technical assistance role. Nevertheless, FO were committed to ensuring that households would be chosen on the basis of HEA criteria, without necessarily being members or members of the basic structure.

Household targeting: weight of opinion leaders
Next step was to meet with representatives of villages to organize general assemblies to form the committees. Before, the project team made sure that all social components were represented. She stressed that vulnerable households will be proposed by the "community" according to their own criteria of territorial vulnerability. The participants elected the volunteer representatives of the 2 committees (selection and validation). Each committee was made up of 5-7 people and one person could not be on both committees. Under the supervision of the project team, members of the selection committee proposed a list of beneficiaries. Subsequently, the project team summoned members of the validation committee to amend the proposed list.
Through their charisma, influential personalities, as committee members, when they are motivated by a desire to respect the rules of the game of fair targeting, can significantly affect the quality of results. Also, these personalities have the capacity to debate certain choices or to point out the omissions of vulnerable households. Other committee members can then avoid proposing non-vulnerable households. Thus, in some villages, committees succeeded in classifying the proposed households by level of vulnerability. Conversely, in certain villages where the influential leaders or members of the committees were not animated by objectivity, the quality of the proposed lists was often poor. Households were offered even if they were not vulnerable. Personal interests and the view of maximizing the village's chance of having enough beneficiaries were the main reasons for this type of action. Members of the committees proposed their own household and those of their relatives who were not necessarily vulnerable. Other committees simply listed all the households in their village in the VP and P categories, because members did not agree on the vulnerability criteria. Thus, this way of proceeding should maximize the chance of their community to have a large number of beneficiaries and above all to avoid the exclusion a priori of households. This configuration made the work of the Gret team difficult, where the household surveys took longer.
Another parameter would have impacted the quality of targeting. This is the lack of remuneration for committee members. During the meetings, negotiations were undertaken with the team which had to justify that no budget line existed to remunerate or support them in kind and that their participation was a "community contribution" in the implementation of the project. These types of complaints suggest that while most committee members frequently displayed selfless commitment, the lack of retribution affected their individual and collective investment.
Acting as "development brokers" (Bierschenk et al., 2000), leaders and committee members ensure, as active mediators, interface between the recipients of projects and development institutions to drain the development rent towards the local arena where they register, while deriving personal benefit by strengthening economic, social, symbolic or political capital. The procedures established by HEA (assemblies, complaint committees) to reduce or locally counter the weight of these social figures may have no effect on the quality of the lists established. Indeed, general assemblies are rarely places of contestation. If the participants come to prove that a household in their village was not to be a beneficiary, it can be immediately replaced by another from the waiting list; he would not necessarily be from the same village. On the other hand, for fear of being indexed as those who publicly or in committee rejected the households selected by the project, participants preferred to keep silent. On the other hand, behind the scenes, participants were sometimes surprised that certain households had not been chosen in relation to those who were retained. Finally, the complaints committees could have made it possible to resolve these shortcomings. However, setting up these committees is not easy. Their operationalization required a time that could be spread over 2 to 8 weeks, time necessary to collect complaints, verify them and proceed to decision-making. Also, the project team did not have enough time to make the complaints committees truly effective. In addition, denunciations could alter the social climate in the villages concerned, which would not be conducive for the execution of activities.
Impossibility of regulating these social games and their bias on targeting then led the project team to interfere with the selection and validation committees, by deploying methods of cross-checking targeting results and adjusting the selection criteria.

Targeting cross-checking of committees by the project team
From the lists amended by the validation committees, household surveys were carried out by the project team.
The tool was based on the criteria used by the public technical services to categorize rural households according to HEA method: household size, cultivated area, food and cash crops, number of animal heads, production goods. New criteria have been added by the team: number of workers, children under 2, farming practices, disabled person, etc.
Results of the surveys were submitted to the targeting committees for validation and / or correction, in particular when households had poorly estimated the number of their assets and their cultivated area, or when households were forgotten, either by absence or because they were did not see themselves as "vulnerable", finding it stigmatizing to be labeled as "Poor" or "Very Poor".
In the end, according to HEA criteria, only nearly half of the households targeted by the targeting committees fell into the categories of "P" and "VP" households. These differences varied from municipality to municipality or from village to village. Overall, the majority of proposals did not meet HEA criteria. Why such disparities in the quality of the proposals of the committees from one municipality and from one village to another, when the project intervention area was located in ZOME that were a priori homogeneous?
It is very likely that some committees proposed almost all the households in the village in order, on the one hand, to optimize their chances of obtaining as many beneficiaries as possible. This strategy allows the committees not to be the target of criticism from the populations. In some cases, when almost all or a large proportion of households have been selected, the quality of the selection is highly variable, ranging from more than half to almost a quarter of the targeted households falling into the vulnerable categories of HEA. However, less systematic targeting can lead to correct results. Result is a generalized suspicion of the choices of less vulnerable households to the detriment of needy households. In 2014, a survey of project beneficiaries revealed that nearly a third of them believed that targeting included non-vulnerable households. Some households had provided erroneous information about their property, without this being corrected by the committees.
These shortcomings led the project team to improve the collection tool in 2015 by adding three qualitative variables: type of housing, sign of the presence of undeclared goods, opinion of the investigator on household declarations. These variables enabled the team to interact with the committees to correct or verify some data. These changes have impacted the beneficiaries' perception of the quality of targeting. However, a second survey found that less than a tenth still felt this targeting was inappropriate. However, this proportion is much lower than that of 2014.
From these 2 surveys, it emerges that these shortcomings noted have not altered the good opinions of beneficiaries and even non-beneficiaries on the HEA method, in particular because it involves advisory committees and household surveys, which targets households in need, and mobilizes village leaders and population. For those considering that HEA had shortcomings, they justified it by the fact that non-vulnerable households were selected in certain villages, the number of beneficiaries was low compared to the size of the villages and the presence of vulnerable households not supported.

Conclusion
The HEA targeting of vulnerable populations has been adopted throughout the region of Sahel, and its application is currently being tested in Burkina Faso. Resulting in quantified categories of socio-economic classes and a more or less fine targeting of vulnerable households, its implementation in the field shows a dysfunction, which development practitioners identify by cross-checking the lists established by the committees with HEA criteria. Several factors explain this discrepancy. First of all, a part of methodological subjectivity (ambiguity and / or imprecision of the selection criteria) leaving room for interpretation in targeting households. Second, the weight of opinion leaders, acting as development brokers, can influence the quality of the results of targeting selected households and / or villages. Finally, the low participation of the populations during general assemblies, which are supposed to counter or validate the established lists, and the absence of a complaints committee, an extremely cumbersome mechanism, and unrealistic to initiate, especially as the members of the committees remain volunteers.
If HEA is perceived by the communities as a breakthrough, the current tendency of practitioners to fill in the imperfections goes in the direction of redoubling and more complex targeting methods on the part of the field teams: a multiplication of adjustments to criteria, heavier operating procedures and cross-checks to verify their quality. In short, these measures require development practitioners to maintain a close presence on targeting committees, an unrealistic presence to be deployed in a change of scale in the method and also accompanied by greater bureaucratization of aid.