The Potential Implications of “Big Ag Data” for USDA Forecasts

Abstract The USDA produces yield and supply estimates for many crops that influence commodity markets and are used for implementing the Title I program, Agriculture Risk Coverage. Precision agriculture advances have increased the potential for the private sector to capture near-real time yield data, however, it is unclear whether they provide advantages in setting market positions since the samples are typically non-random. Here, we use yield histories from a large population of corn farms to quantify biases associated with different non-random sampling schemes for estimating aggregate yield, and demonstrate the effectiveness of benchmarking procedures for removing systematic prediction error.