Smallholder Farmer Risk Preferences in Northern Ghana:Evidence from a Controlled Field Experiment

Abstract

We conduct a controlled field experiment to elicit risk preferences among maize farmers in Northern Ghana. Farmers participating in the experiment were asked to choose from a menu of lotteries representing different hypothetical probability distributions over yields produced by ‘traditional’ and ‘high yield’ maize varieties. We estimate a Rank-Dependent Utility Model (RDU) with an Expo-Power utility function, allowing for systematic subjective underweighting or overweighting of outcome probabilities and non-constant relative risk aversion. Based on our estimates, we cannot reject the hypotheses that decisions made by farmers in our study can be uniformly characterised by conventional Von Neumann–Morgenstern expected utility theory (EUT), but reject the hypothesis that farmers exhibit constant relative risk aversion.