Forecasting Rainfall: Are Farmers Bayesian? Evidence from Northeast Brazil

Abstract:

While higher insurance take-up remains a challenge in the developing world, more accurate rainfall forecasts could play the same role (Rosenzweig and Udry, 2014). In practice, there are multiple sources of rainfall forecasts, with wide variation in accuracy, and it is unclear how farmers weigh in different sources to form expectations. This paper randomly assigns information about the local accuracy of alternative sources of rainfall forecasts and documents how doing so affects farmers' perceived reliability of each source, and their expectations about rainfall throughout the rainy season. We track farmers’ expectations weekly over SMS, and incentivize correct forecasts for truthful reporting. Results are forthcoming.

Last updated on 11/02/2015