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.

中文翻译：

---

农民在做出气候适应决策时的贝叶斯如何？期望形成参数化模型的计算机实验室实验

随着气候变化对农业生产的影响在地方层面慢慢显现（有时会出现相互矛盾的信号），农民的信息处理和决策与政策分析和建模变得更加相关。主要的挑战是揭示农民形成对未来生产结果的期望的模式，并将这些发现编码到异质期望形成模型中。我们开发并测试了一项以支出为动机的现场实验，以观察气候变化下农民的决策，并检查他们如何在递归动态环境中形成他们的期望。参与者暴露在模棱两可的环境中，并通过反复随机抽签获得有关可能气候结果真实分布的增量证据。农业和环境研究中使用的模拟模型通常实现简单形式的自适应代理期望，或者通过假设完美的远见或恒定的期望来完全忽略这个问题。我们的计算机实验室对来自德国西南部的蓝领和白领农民进行了实验（n  = 97) 表明大部分农民的期望行为可以用贝叶斯类型的期望模型很好地复制。