Integrating value chains to improve food safety and increase smallholder incomes in Kenya
How to cite
Hoffmann, Vivian; Treurniet, Mark; Kariuki, Sarah; Bulte, Erwin; Pieters, Janneke. 2016. "Integrating value chains to improve food safety and increase smallholder incomes in Kenya." AEA RCT Registry. AEARCTR-0001373. https://doi.org/10.1257/rct.1373-3.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
Adoption of improved agricultural technologies in developing countries may be limited by common informational and market inefficiencies. In this study, we examine the role of such inefficiencies in the adoption of a new food safety technology, Aflasafe. Aflasafe has been shown to reduce aflatoxin contamination in maize by approximately 90%. However, this technology faces several barriers to widespread adoption. First, since Aflasafe is applied while the crop is still growing, its use increases farmers’ exposure to yield risk. In the absence of a market for crop insurance, this may limit its adoption. Second, the lack of price incentives for food safety in markets served by smallholder maize producers in Kenya may constrain the adoption of food safety technologies by these farmers. In this study, we explore the impact on adoption of Aflasafe of 1) bundling the product with a rainfall index based money-back guarantee, and 2) access to an output market that rewards aflatoxin safety.
Intervention
The study will include 3 interventions: an Input Linkage treatment, a Money Back Guarantee treatment and an Output Market Linkage treatment. The Input Linkage treatment groups will receive information on the benefits of Aflasafe and instruction on its use. They will also be given an opportunity to buy Aflasafe, which is not currently available in the study area. These groups will further be divided into two treatment conditions concerning money back guarantee (MBG): an Optional condition, offered the option to buy Aflasafe either with or without the MBG, and a Bundled condition, in which Aflasafe is only offered in combination with the MBG. The Money Back Guarantee treatment will be crossed with the Output Market Linkage treatment, which consists of a linkage between the farmer group and a buyer that pays a premium for safe maize. Hence, in total there will be five categories, including the pure control group, and four treatment groups (Optional MBG / Output linkage , Optional MBG / No output linkage, Bundled MBG / Output linkage, and Bundled MBG / No output linkage).
Experimental design
Farmer groups are drawn from a list of 250 farmer groups obtained from the Cereal Growers Association and county Ministries of Agriculture. Based on power calculations, we allocated 160 farmer groups across the four treatment groups and the control group as follows: Optional MBG / Output linkage: 38, Optional MBG / No output linkage: 38, Bundled MBG / Output linkage: 38, and Bundled MBG / No output linkage: 38; Pure Control: 8.
In order to limit spillovers across money back guarantee treatments, we created comparable, but geographically distinct, clusters of farmer groups within each of the study counties (4 in Meru, 2 in Tharaka-Nithi and 2 in Embu) and subsequently assigned these clusters to either the bundled or the optional MBG treatment. Subject to a minimum geographical distance of 5 km between clusters, we aimed to select similar farmer groups into the clusters within each county. Similarity was defined based on the Euclidean distance in the six-dimensional space formed by the standardized values of variables listed in section 5.4.
To this end, we first dropped clusters close to the county borders to achieve a minimum distance of groups in different counties of at least 5 km. Subsequently, we excluded any groups within a 5 km bands dividing the remaining groups into similarly-sized clusters. The direction of this band was selected to minimize the Euclidean distance between the groups on either side of it. From the remaining groups, we then iteratively selected matched pairs across each cluster with the lowest Euclidean distance into the sample. To ensure that the MBG treatments were spread out geographically, we manually decided which clusters would receive the same treatment. Finally, we randomly assigned the bundled MBG to one of the two groups of clusters.
Within each of the money back guarantee clusters, the market linkage treatment was randomly assigned at the village level.
Pure control groups were selected as the 8 nearest geographical neighbors to any BY group, stratified by county approximately in proportion to the total number of groups on the initial list.
Design details
Cited by
No citation data available for this registration yet.