Many smallholder farmers in low- and middle-income countries do not adopt modern agricultural inputs sold by agro-dealers due to uncertainty about their performance. We introduce a system that aggregates farmers’ evaluations of seed into ratings and disseminates them to farmers and agro-dealers. Such a system may affect outcomes through three channels: correcting farmers’ beliefs and reducing their uncertainty about seed performance (belief updating), helping farmers identify better agro-dealers (agro-dealer sorting), or inducing agro-dealers to respond to reputational pressure (supply-side accountability). We test this approach in a randomized controlled trial involving 3,500 farmers and 350 agro-dealers in the market for high-yielding maize seed in Uganda. We find that the intervention increases farmers’ perceptions of seed performance, leads to higher adoption, and raises yields by 56 kg/acre, equivalent to 13% of the baseline mean. The pattern of results points most strongly to belief updating. Switching across agro-dealers increases modestly at midline but is not persistent, providing weak evidence for agro-dealer sorting as the primary channel. On the supply side, agro-dealers respond along service and signaling margins, but we do not detect improvements in seed handling, storage, or seed performance. The results show that peer-information systems can raise adoption and productivity in input markets even when they do not primarily operate through buyer reallocation or improvements in seller quality.
To access my job market paper, click here.