Replication Under Spatial Uncertainty: Bias in Weather-Yield Estimation from EU FADN Data

Abstract This study examines the replication of weather-wheat yield sensitivity using farm data from the European Union’s Farm Accountancy Data Network (EU FADN) for Germany (2004–2020). Standard FADN data do not include farm locations beyond NUTS-2 or agro-climatic information. Using the example of Schmitt et al. (2022, Food Policy, DOI: https://doi.org/10.1016/j.foodpol.2022.102359), we find that farm locations at NUTS-2 introduce bias in an unknown direction when estimating weather-wheat yield relationships. We propose two mitigation approaches: probabilistic farm locations and probabilistic crop maps, to improve the selection of relevant regions before assigning weather information. We validate that both approaches replicate the core results in terms of direction and magnitude. We conclude that researchers should have access to EU FADN data at adequate spatial resolution or data must be supplemented with detailed agro-climatic information to improve agricultural policies targeting climate change.