Nutrition-sensitive agriculture: Impact evaluation of interventions to relax constraints of orange fleshed sweetpotato (OFSP) adoption

There are three sets of interventions in this evaluation - agriculture extension, health-nutrition, and financial products.
The agriculture intervention involved providing inputs and agricultural extension services to women farmers assigned to treatment. This intervention entailed using community agents to train beneficiaries using demonstration plots in the community and the provision of an input package. The input packages contained planting material (vines) for OFSP and a few other nutrient-enriched crops including beans, tomatoes, and groundnuts. There were three types of community agents used to deliver the agriculture interventions: community agriculture promoter (CAP), vine producer, and marketing agent.
The health and nutrition information intervention also involved community agents that focused on information dissemination. Community health promoters (CHP) conducted quarterly health and nutrition community forums targeting female caregivers and husbands to promote consumption of nutrient-rich foods and improved feeding practices for under-2 children, adolescent girls, and pregnant women. In addition, monthly growth monitoring and promotion (GMP) sessions were held for under-2 children to promote dietary diversity and optimal infant/young childcare and feeding practices. Nutrition fairs to promote nutrient-rich foods were also conducted jointly by CHPs and CAPs annually to build community awareness and demand for nutrient-rich foods and food preparation and processing techniques.
The innovative financial products were offered to address three specific constraints in technology adoption. A credit product was customized to alleviate credit constraints for input purchase. The loan had an 8-month duration, flat interest rate, and backloaded repayment schedule that allowed farmers to repay 25% of the loan during the first four pre-harvest monthly installments and the remaining 75% in the final four post-harvest installments. An input voucher was designed to act as a commitment device for the purchase of agricultural inputs (seeds and fertilizers) to overcome time-inconsistency and present bias. Farmers were offered to purchase input packages for OFSP and other nutrient-rich crops after the harvest period when they had cash on hand. These inputs would then be delivered to farmers when they were needed during the subsequent planting season. A price insurance product was offered for the first cropping season that covers the potential downside price risk created by unpredictable demand for OFSP in the market. This product is a surplus buyback guarantee at prevailing market prices and tried to ensure that in the event of low demand, farmers would continue to produce adequate supply of the crop.

2014-10-01

2018-04-30

OFSP adoption - whether cultivated, amount of land used for cultivation, amount produced and amount sold.
Consumption - consumption of OFSP in the past week and month, household dietary diversity.

The crop production data is collected by season and cover the 2 seasons preceeding the survey.
Household dietary diversity score will follow the scoring method suggested by WFP.

Child health (anthropometric measures of 6-59 months old children),
Women respondents knowledge on nutrition,
Women empowerment in agriculture index (WEAI),
Eye health

Child health will focus on underweight, but will also look at stunting and wasting as longer term outcomes. In addition to the 6-59 months cohort, the children in this age bracket at baseline will be followed up at endline.
Eye health/visual acuity is an additional measure (using Sjorgen Eye Test that allows testing for participants with literacy) is added at endline.

This cluster randomized trial has three treatment arms and a control group. The treatment groups are
C: Control (No intervention, 30 clusters)
T1: Agriculture only (30 clusters)
T2: Health only (30 clusters)
T3: Agriculture and health (120 clusters)
Within the T3 arm, there are 4 sub-groups
T3A: Agriculture and health (30 clusters)
T3B: Agriculture and health + credit (30 clusters)
T3C: Agriculture and health + input voucher (30 clusters)
T3D: Agriculture and health + price insurance (30 clusters)

The randomization was done by using stata codes after a village mapping.

The randomization is done at cluster level. Each cluster is typically a village, but small villages neighboring each other were combined in some cases to create clusters. Households were sampled for interview by using a village level census in the target clusters as the sampling frame. The same selection criteria were applied in all treatment group and control clusters for identifying eligible households and respondents.

Yes

210 clusters (i.e. villages) from 4 districts.

8,400 households sampled from the village census (40 household per cluster).

C: No intervention (30 clusters - 1200 households)
T1: Agriculture only (30 clusters-1200 households)
T2: Health only (30 clusters - 1200 household)
T3A: Agriculture and health (30 clusters - 1200 households)
T3B: Agriculture and health + credit (30 clusters - 1200 households)
T3C: Agriculture and health + input voucher (30 clusters - 1200 households)
T3D: Agriculture and health + price insurance (30 clusters - 1200 households)

The minimum detectable effect of any of the interventions vs. control group for binary outcomes (e.g. adoption of OFSP) is 9 percentage points at 90% power (assuming control group mean of 0.05, intra-cluster correlation of 0.1, significance level at 5% and 10% attrition rate). For scale outcomes (such as dietary diversity score) the MDE is 0.3 SD.