Conference Agenda

Many smallholder farmers in developing countries rely on memory and experience alone to make planting and investment decisions. For those with limited access to advanced technology and financial services, errors due to memory inaccuracy and faulty forecasts can impact welfare. Using a recall experiment, survey data, and secondary data from various sources, I study farmers' recall and forecasts of rainfall, agricultural output, and crop prices in Burundi. While I find evidence of recall inaccuracy and decay, farmers’ recall accuracy for the main crop of each season is high and remains stable over time. However, recall is subject to memory interference from non-agricultural events. I also explore the link between recalled and forecasted values, providing evidence of long-lasting effects in how farmers form their forecasts. Finally, I evaluate the accuracy of farmers' forecasts. I show that while price forecast errors can be significant, the portion attributable to recall inaccuracy is small. Farmers' forecasting behavior is best described by a mean absolute error loss function. My study provides a detailed assessment of how smallholder farmers use past information to form their forecasts and identifies limitations in both memory and forecasting.