How Well Can Experts Predict Farmers' Choices in Risky Gambles?

Risk is ubiquitous in agriculture and a core interest of agricultural economists. While farmers' risk preferences are well studied, there is limited knowledge on the perspectives of other stakeholders on farmers' risk preferences. We address this gap by eliciting predictions for a multiple-price-list task from 561 students, farm advisors, and experts from Italy, Poland, Croatia, Spain, France, Sweden, and the Netherlands. First, we investigate whether the risk preferences of farmers from different European production systems differ in terms of predictability for the experts. Second, we compare the predictions of different groups of experts, as well as their accuracy. Third, we evaluate whether the accuracy of predictions can be improved by changing incentive mechanisms. Overall, we find substantial variation in individual predictions. Yet, average predictions are close to the averages of the observed responses of farmers. We find that an international group of researchers in experimental economics provides more accurate predictions than farm advisors and other experts or students of agriculture. Differences in predictions by production systems are small. Incentivizing predictions by either a tournament scheme (the best prediction receives a reward) or high accuracy (randomly selected participants are paid depending on the quality of their prediction) do not strongly affect the accuracy, but may slightly reduce noise in the predictions.