Abstract
Disorganized agricultural value chains often prevent the transmission of quality incentives to upstream farmers, especially when quality is unobserved at the farm gate. When the quality-revealing costs are prohibitively high, and without an effective traceability system throughout the value chain, farmers might not be rewarded for producing high-quality products even if quality incentives exist in the downstream markets. Thus, farmers have low incentives to invest in quality upgrading. In this study, I establish digital traceability systems among Kenyan dairy cooperatives and develop an innovative quality monitoring method based on Bayesian models to overcome the high test costs for individuals’ milk. The model uses the aggregated milk container quality information and traceability data to detect whether individual farmers are producing high- or low-quality milk. The model-predicted quality significantly correlates with the one-time milk test among 940 farmers. I conduct an individual-level randomized controlled trial to provide both cooperatives and farmers with either the model-predicted milk quality or one-time random quality test results. I evaluate the differential effects on cooperatives’ reactions to different types of farmers, farmers’ investment in quality upgrading, farmers’ milk quality, and sales.