Expected Utility or Prospect Theory Maximizers? Results from a Structural Model based on Field-experiment Data

We elicit risk preferences of French farmers in a field experimental setting under expected utility theory and cumulative prospect theory. We use two different estimation methods, namely the interval approach and the estimation of a random preference model. On average, farmers are risk averse and loss averse. They also exhibit an inverse S-shaped probability weighting function, meaning that they tend to overweight small probabilities and underweight high probabilities. We infer from our results that CPT explains farmers’ behaviour better than EUT in the context of our experiment. We also investigate how preferences correlate with individual socio-demographic characteristics. We find that education and agricultural innovation are negatively linked with risk aversion. Our results also show that age, education, household size and the level of secured income tend to lower farmers’ loss aversion. Finally, older farmers and farmers with large farms distort probabilities less than the others. These findings contribute to the literature which compares expected utility with competing decision theories. They also give important insights into farmers’ behaviour towards risk, which is critical for relevant public policy design.