Evidence-based Scaling of Improved On-Farm Storage among Smallholders (Kenya)

Each farmer groups receives a standardized training session on improved on-farm storage (Huss et al., 2021). This training takes place around harvest time when farmers would typically require hermetic storage bags. The trainings are implemented by project partner icipe. The training will be provided both to the original treatment and control groups. Following the training (during the same visit), all farmers participate in a short survey, including on demographics, hermetic storage bag ownership/use, and our key outcome variables. Subsequently, farmers participate in the hermetic storage bag sales. Some farmer groups are randomly allocated to participating in a BDM auction, while other farmer groups receive a take-it-or-leave-it offer (TOLI) to purchase bags (reflecting current market prices for the bags) for comparison.

2024-08-12

2024-09-30

The primary outcome variable to be measured is hermetic storage bag purchase rates at different prices. For the take-it-or-leave-it offer, another primary outcome is the number of bags purchased.

Data is collected during implementation of the BDM mechanism.

As secondary outcome variables, we measure household food security, income and poverty, farm investment (e.g. seed type) and local market prices.

To measure self-assessed food security, we use the reduced Coping Strategies Index (rCSI), and apply standard weights and thresholds for classification in food (in)security categories (Maxwell et al., 2014; Vaitla et al., 2017). To assess poverty and income, we use the Progress out of Poverty Index (PPI) to analyse long-term effects (Desiere et al., 2015). To analyse potential heterogenous treatment effects, we further collect socio-demographic data.

We implement a cluster randomized control trial (RCT) to study the role of (i) financial constraints, (ii) liquidity constraints, and (iii) risk perception in smallholder farmer’s decisions to adopt hermetic storage. The RCT is implemented in Kenya (Kakamega county).

We draw from a sample (the “original sample”) that has previously been recruited and continues to participate (via regular SMS-based surveys) as part of the first phase of our research project.

In September 2019, farmers in half of the experimental clusters (randomly allocated) were provided with five hermetic storage bags per household and training in their use. Before and after the interventions, data was collected through regular (weekly and monthly) SMS-based mobile phone surveys.

Not available

Randomization done in office by a computer. BDM prices drawn on-site.

Original randomization unit are experimental clusters (spatial clusters of farmer groups).

Yes

The original sample consists of 72 experimental clusters (spatial clusters of farmer groups), matched in pairs.

The original sample consists of a total of 285 farmer groups with a total of 5444 smallholder households.

BDM: 36 clusters; TOLI (Pay now): 18 clusters; TOLI (Pay later): 18 clusters. Half of the clusters assigned to each of these conditions had been treated as part of the original RCT.