The predictive power of risk preference measures for farming decisions

This paper uses a lottery-choice mechanism to measure farmer preferences over money-denominated risks. We look at the ability of these choice data to predict farming decisions for an in-person sample of 68 farmers. A coarse version of our risk preference measure has substantial explanatory power but in an unexpected direction: a farmer who was more risk averse under our measure was less likely to have diversified his operation and less likely to have a crop insurance contract. A fine version of our risk preference measure has essentially no explanatory power. We conclude that despite their widespread use in the lab, lottery-choice measures of risk preferences are unproven for predicting real world farming behaviour.