A Horse Race Comparison of County-Level Crop Yield Prediction Methods
AgEcon Search (University of Minnesota, USA) · 01.01.2025
DOI: 10.22004/ag.econ.391348 ↗Kurz gefasst
Fachartikel in der Zeitschrift AgEcon Search (University of Minnesota, USA), erschienen am 01.01.2025. Verfasst von Junkan Li und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Accurate county-level crop yield prediction is essential for agricultural outlooks and risk management, yet the predictive value of complex models remains uncertain. This study conducts a horse race comparison of alternative yield prediction methods for corn, soybeans, and cotton using USDA NASS yield data and PRISM weather data. Models range from simple historical averages to specifications incorporating spatial dependence, time dynamics, and weather variables. Evaluated using out-of-sample forecasts from 2015 to 2024, results show that simple models based on recent county-level yield averages consistently outperform more complex approaches. The findings highlight the robustness and practical value of parsimonious benchmarks for operational yield forecasting.
Schlagwörter
Zitieren
APA
Li, J., & Tsiboe, F. (2025). A Horse Race Comparison of County-Level Crop Yield Prediction Methods. AgEcon Search (University of Minnesota, USA). https://doi.org/10.22004/ag.econ.391348
Vancouver
Li J, Tsiboe F. A Horse Race Comparison of County-Level Crop Yield Prediction Methods. AgEcon Search (University of Minnesota, USA). 2025. doi:10.22004/ag.econ.391348
BibTeX
@article{Li2025,
author = {Li, Junkan and Tsiboe, Francis},
title = {A Horse Race Comparison of County-Level Crop Yield Prediction Methods},
journal = {AgEcon Search (University of Minnesota, USA)},
year = {2025},
doi = {10.22004/ag.econ.391348}
}
EUVIMED