Experimental Contextualization and External Validity of Farmers’ Risk and Uncertainty Preferences

Risk and uncertainty preferences are key determinants of agricultural production decisions. Yet preferences elicited through standard lottery tasks often exhibit weak external validity. While task contextualization is meant to bridge this gap, its application to risk and especially uncertainty elicitation remains underexplored in developing countries. This paper fills two gaps in the literature. First, we provide the first application of contextualized experiments to elicit uncertainty preferences among smallholders in a developing economy. Second, we conduct a comparative evaluation of the external validity of risk versus uncertainty preferences across contextualized and non-contextualized formats. We use a lab-in-the-field experiment paired with a survey of farmers in Zimbabwe. Participants completed both non-contextualized and contextualized lottery tasks, with the latter framed around contract farming decisions across various crops to manipulate salience. Our results reveal that contextualized preferences differ from non-contextualized counterparts, with standard tasks overestimating risk and uncertainty aversion. Regarding external validity, elicited risk preferences fail to correlate with real-world decisions. Conversely, uncertainty preferences exhibit selective external validity, correlating with farming choices only when elicited with a salient framing. These findings highlight that while contextualization can enhance external validity, its efficacy is highly context-dependent, necessitating careful methodological refinement before integration into economic modelling.