Brück, T, Ferguson, NTN, Ouédraogo, J and Ziegelhöfer, Z, 2018. Impacts of the World Food Programme’s interventions to treat malnutrition in Niger. 3ie Impact Evaluation Report 80. New Delhi: International Initiative for Impact Evaluation (3ie).
Link to Source
This study evaluates the impacts of four different aspects of WFP’s Protracted Relief and Recovery Operation (PRRO) on the prevalence of moderate acute malnutrition (MAM) in Niger.
The broad remit of the PRRO in Niger is to reduce MAM prevalence through multiple targeted and universal activities. This evaluation focuses on four of these activities: the first two activities – Targeted Food Assistance (TFA) and Blanket Supplementary Feeding (BSF) – are implemented during Niger’s lean season with the aim of preventing MAM developing due to seasonal fluctuations in food availability. The third – Targeted Supplementary Feeding (TSF) – is a treatment programme and is provided to children already assessed to be suffering from MAM. The fourth are activities designed to increase poor households’ access to food and assets, particularly through land rehabilitation, water harvesting and local purchases - Food for Assets (FFA), an agriculturally sensitive form of assistance.
This impact evaluation focuses on three primary evaluation questions:
- What is the impact of, in addition to FFA, gaining at least one of BSF, TSF or TFA in comparison to a control group who receive no assistance at endline?
- What is the impact on individuals who receive at least one of TFA, TSF and BSF at endline, compared to individuals who receive no assistance?
- What is the impact on individuals who receive FFA and at least one of TFA, TSF and BSF at endline to those who do not receive FFA but who receive at least one of TFA, TSF and BSF?
Using two waves of panel data collected in March 2014 and in September 2016, the impact of receiving at least one of TFA, TSF or BSF, compared to receiving nothing is isolated; and any synergies that are present between the receipt of FFA and the three MAM interventions are studied.
The impacts for the following three groups are compared, where all the groups received FFA during the baseline survey, but their status during the endline survey had changed:
- Group 1 received only FFA in the baseline and no assistance in the end line (-FFA);
- Group 2 received only FFA in the baseline and no FFA but at least one other form of assistance (TFA, TSF or BSF) in the end line (-FFA+); and
- Group 3 received only FFA in the baseline, continued to receive FFA in the end line, and received at least one other form of WFP assistance (TFA, TSF or BSF) in the end line as well (FFA+).
The evaluation uses joint instrumental variables and selection correction model within a difference-in-difference framework to measure and compare nutritional outcomes across the different treatment groups to understand the different PRRO impacts and their synergies.
Nutritional outcomes were measured by three indicators of specific interest for MAM (a binary indicator for whether a child suffers from MAM, the weight for age z-score, the mid-upper arm circumference) for children aged 6-59 months at baseline (March 2014) and endline (September 2016).
The evaluation finds no evidence of positive impacts of treatment and/or prevention programming only. Rather, this programming approach can have negative impacts. At first sight, this is a surprising finding. It could be that the expected benefits decline over time and come to be dominated by one or two other influences on child malnutrition.
Specifically, the evaluation posits two reasons for this finding: either that other coping strategies are more effective than the receipt of this form of food aid; or that receipt of this assistance alters intra-household decisions on nutrition. For example, a highly food insecure household that does not access a standard WFP programme may choose to engage instead in migration. Remittances received can, in turn, reduce food insecurity and may, in fact, dominate the impact of the programme itself. Alternatively, a household that receives such programmes may substitute scarce resources away from eligible children to other household members, leading to reductions in those children’s level of nutrition but not that of the household as a whole.
Broadly speaking, qualitative results support the quantitative findings and provide intuition for the underlying theories that might explain these findings. The qualitative research suggests that the behaviour of men and women is changed positively due to the WFP interventions. Nevertheless, evidence of perverse effects was also found to be quite common. For example, households sometimes sell the received malnutrition treatment products in markets rather than consuming them at home. In the case of some households, it was found that malnutrition was induced in “cured” children, in order to maintain eligibility for support (mostly by feeding them tamarind to trigger diarrhoea). WFP's agriculture-sensitive intervention has enabled the most vulnerable households in some villages to meet their food needs and even make financial savings. The interventions allowed households to mitigate food crises and malnutrition and to create productive assets.
When the indicators for which the analysis shows a negative or no impact are analysed, it is noted that there is a positive impact of bias, implying that treatment may not necessarily be reaching the correct groups. This suggests that non-FFA programming is not strongly linked to improvements in the MAM status of those children who receive such assistance. That is, the children who receive no FFA but one of BSF, TSF or TFA may have better nutrition indicators, a priori, than children in households who do not receive the assistance. Either this implies errors in the targeting strategy, or that uptake of the assistance is lowest amongst those who may stand to benefit most from it. This is in contrast to the FFA programming which appears well targeted.
Cost efficiency analysis
The cost effectiveness analysis undertaken as part of this evaluation indicates that the combined approach in WFP programming has been the most cost-effective compared to single interventions. Interventions combining BSF, TFA and FFA activities lead to improved nutrition indicators in the range of 0.28 standard deviations per US$100 spent, implying the average cost of bringing a MAM child to a non-MAM status is estimated at US$ 352.60.
Implications for implementers
The evaluation also notes important lessons learnt on monitoring and evaluation (M&E) and data collection. A large initial investment in the collection of baseline data did not lead to optimal M&E due to the poor design of the follow-ups. While a large baseline was collected at the beginning of the PRRO period, this was never designed to be followed-up, which resulted in high attrition.
Implications for policy and practice
The impact evaluation puts forward the following points for consideration:
- The World Food Programme, in Niger and more generally, should consider FFA as an effective nutrition-sensitive model for malnutrition programming and adopt and further analyse it in other complex environments. The relative and absolute extent of food for assets programming should be expanded as far as possible, within budget allowances. Where possible and feasible, FFA programming should be provided alongside treatment and/or prevention programming. There is an urgent need to understand the intra-household behavioural responses to the provision of treatment and/or prevention programmes. A second endline survey should be conducted with the original sample in Niger in 2018 to analyse long-term programme impacts