Abstract
Using a large representative dataset of 4,428 maize farmers from Burkina Faso with information on over 7,800 plots, we study the role of risk and farmers' risk preferences in their use of nitrogen fertilizers. After characterizing the role of nitrogen on the moments of the maize yield distribution, we plug the plot-specific distributions into a structural model that allows for both risk preferences and probability distortion to elicit farmers' underlying behavioural model. We found farmers to be only moderately risk averse and to distort probabilities; i.e., farmers overweight the small probabilities of getting high yields. Finally, running simulations, we find that prices are a more important driver of the quantity of nitrogen used on maize plots than farmers' risk preferences. Our results suggest that input subsidy programs in this context, if well implemented, may have the potential to increase fertilizer use.