Evaluation of the pilot project to improve soybean productivity in the Agricultural Development Centre 4 in Benin: An experimental study of gender-differentiated impact

The package we evaluate has two components: training and inputs delivery. The training component involves group training modules and monthly follow-up visits to individual farmers. Inputs are improved seeds, herbicide, inoculum, and fertilisers provided on credit. For the purpose of the evaluation, we provide inputs for 0.25 hectare of soybeans for each treated farmer. Effective participation in training is a precondition for receiving the inputs package.

2020-01-20

2020-12-31

Productivity (Yield)

Quantity of soybeans harvested per ha

Our experiment involves two levels of random assignment and three groups. First, we randomly assign eligible villages to either the treatment group or the control group. Second, in each treatment village, we randomly assign eligible households to two groups. In one group, households decide by themselves who between the husband and the wife will be the direct beneficiary of the intervention. In the other group, the wife is imposed as the beneficiary of the intervention.

Our experimental design involves the following groups: Group_1) A treatment group in which households decide by themselves who between the husband and the wife will be the beneficiary of the intervention; Group_2) A treatment group in which the wife is imposed as the beneficiary of the intervention; Group_3) Control group in which households will not receive any intervention.

We use a 2-step randomization process. First (Step 1), at the village level, we randomly assign villages (clusters) to the treatment (Groups_1/2) or the control (Group_3). Second (Step 2), at the farmer-level in treatment villages, we random assign eligible households to Group_1 (where the choice of the beneficiary is made by households) or Group_2 (where the woman must be beneficiary).

Village eligibility criteria are: production of soybeans in the village; no history of capacity building activities led by UNPS or ABS in relation with soybean production; be on the UNPS expansion list; have at least 16 eligible households - in which both the man and woman who are in a relationship (living together) have produced soybeans during the past crop year (2019-2020): and be at least 10 km from the nearest eligible village. Before the village-level randomization, we pairwise match eligible villages based on climatic and soil conditions as well as distance to the main road. Then, we select randomly the required number of village pairs for randomization.

Households eligibility criteria are: household in which both the man and woman who are in a relationship (living together) have produced soy during the past crop year (2019-2020); have never benefited from the capacity building of UNPS, the ABS or other partners; have plans to produce soybean production during the 2020-2021 crop year on at least 0.25 ha; be willing and available to participate in project activities either as a treated or untreated household. First, we do a listing of eligible household to construct the sample framework. Second, we randomly select the required number of household for randomization.

Public lottery

Units of randomization are villages (for the first level or randomization) and households (for the first level or randomization).

Yes

38 Villages

608 households. I each household, we will survey both the husband and the wife.

Groups_1/2 (treatment): 19 villages.
Group_1 (households decide beneficiary): 152 Households
Group_2 (woman imposed beneficiary): 152 Households
Group_3 (control): 19 villages, 304 Households

MDE = 0.63 SD (for village level assignment) MDE = 0.36 SD (for household level assignment)