How Bayesian Are Farmers When Making Climate Adaptation Decisions? A Computer Laboratory Experiment for Parameterising Models of Expectation Formation

As the consequences of climate change for agricultural production slowly unfold at the local level (sometimes with contradicting signals), farmers’ information processing and decision making become more relevant for policy analysis and modelling. The major challenge is to reveal patterns in the way farmers form expectations about future production outcomes and to encode these findings into models of heterogeneous expectation formation. We developed and tested a payout‐motivated field experiment to observe farmer decision‐making under climate change and to examine how they form their expectations in a recursive‐dynamic context. Participants were exposed to ambiguity and acquired incremental evidence about the true distribution of possible climate outcomes through repeated random draws. Simulation models used in agricultural and environmental research usually implement simple forms of adaptive agent expectation or completely neglect this issue by assuming perfect foresight or constant expectations. Our computer laboratory experiments with blue‐ and white‐collar farmers from Southwest Germany (n = 97) suggest that expectation behaviour of a large share of farmers can be well replicated with Bayesian types of expectation models.