Is Being Bold Better? Industry Expectations of USDA Corn and Soybean Production Estimates
利用企业层面预测美国农业部玉米和大豆产量的独特数据，发现这些预测存在归因和锚定等认知偏差，且越大胆的预测准确性越低，偏离群体共识并无益处。
ABSTRACT Despite the extensive use of industry expectations in measuring forecast accuracy and price reactions to USDA reports, very little is known about their properties beyond the basic statistical characteristics of bias, rationality, efficiency, and relative accuracy. Using unique proprietary data of firm‐level expectations for upcoming USDA corn and soybean production estimates, we demonstrate that these forecasts exhibit cognitive biases such as attribution and anchoring. Prior success leads to overconfidence and bolder forecasts, and firms base their forecasts on a known reference value. We also show that the bolder the forecasts, the lesser the accuracy, indicating that substantially deviating from the herd does not pay off when it comes to crop production forecasts.