Chat Over Coffee? Diffusion of Agronomic Practices Through Information Networks in Rwanda

How to cite

Zipfel, Celine; Duflo, Esther; Suri, Tavneet; Keniston, Daniel. 2022. "Chat Over Coffee? Diffusion of Agronomic Practices Through Information Networks in Rwanda." AEA RCT Registry. AEARCTR-0010224. https://doi.org/10.1257/rct.10224-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Celine Zipfel* , Esther Duflo , Tavneet Suri , Daniel Keniston

* Primary investigator

Registry ID
AEARCTR-0010224
DOI
https://doi.org/10.1257/rct.10224-1.0
First registered
Last updated
Status
Completed
Trial dates
2009-12-01 – 2012-11-30
Countries
Rwanda (Africa)
JEL codes
O12, O13, O33
Registry record
http://www.socialscienceregistry.org/trials/10224

Abstract

This paper studies the role of social learning in farmers’ adoption of improved coffee practices, in the context of training field experiment in Rwanda. While the program improved knowledge of all trained best practices and strengthened social networks, detailed tree audits reveal limited impacts on adoption. We find no evidence of diffusion through farmers’ networks; instead, control households experienced negative spillovers in high treatment concentration areas. This explains much of the 7% higher yields of treatment farmers compared to the control post training. Our results highlight the challenges of information diffusion in agricultural extension programs.

Intervention

We evaluate TechnoServe's agronomy training program, which covered several best practices in coffee growing: tree rejuvenation and pruning; fertilizer use; pest, disease and weed management; mulching; soil and water conservation; optimal shade; and record keeping. The training sessions took place once a month for eleven months in the first year of the program (starting in February 2010). TechnoServe conducted an additional five review training sessions after these sessions were over, with the last review session being held in October 2011.

Experimental design

TNS identified 1,594 farmers (from 27 villages) to be part of the sample (all volunteered for the training program). Of the 1594, 855 farmers were randomly assigned to the treatment group to receive the agronomy training. The remaining 739 farmers were assigned to the control group and received no training at any point. In addition to this basic farmer level randomization, we randomized the 27 villages that had farmers that registered to be part of the training into three different levels of treatment concentration: in a third of villages, 25% of the registered farmers were treated, in a second third of villages, 50% of the registered farmers were treated, and in the final third of villages, 75% of the registered farmers were treated.

Design details

Randomization method
Oversubscription design
Randomization unit
Household
Clusters
27 villages
Observations
1594 households across 27 sample villages
Arms
739 Control households 855 Treatment houeholds
Primary outcomes
1. Knowledge of the agronomy best practices farmers were taught in the program 2. Self-reported adoption of these practices 3. Adoption as measured by our own tree audits 4. Input use (chemical fertilizer and labor) 5. Yields (KG/tree)

Cited by

No citation data available for this registration yet.

Keywords