18 pages, Recent studies cast doubt on the ability of abstract experiments to predict decision-making in the field. Thus, scholars have argued for more ‘realism’ by introducing context to field experiments. Yet, such realism may work against the induced values of monetary incentives in economic experiments. It is an open question whether contextual framing works best with or without inducing values, through methods such as the use of monetary incentives. Using a sample of 146 German farmers, we compare experimentally the predictive power of a framed lottery in an agricultural context vs. using an abstract version. For one half of the sample, lotteries are incentivised; for the other half, they are hypothetical. Although risk preferences differ between treatments, all four lottery tasks correlate poorly with farmers’ real-world use of risk management instruments such as harvest or hail insurance. Subjects who start with an agricultural framing are willing to take significantly greater risks in the lotteries. More generally, our findings cast doubt on the ability of lottery tasks to predict risk-taking in the field.
