摘要：This study evaluates the accuracy of USDA interval forecasts for corn, soybean, and wheat prices using Christoffersen’s (1998) tests for unconditional coverage, independence and conditional coverage adjusted for asymmetries in tail probabilities. The findings of this study demonstrate that due to uneven distribution of forecast misses around the interval, calibration of soybean price forecasts in several cases was rejected by basic coverage tests (suitable for symmetric intervals) but not rejected by the tests adjusted for asymmetry. Thus these forecasts were asymmetric but accurate. Symmetry was not a limiting assumption for corn and wheat interval forecast accuracy.
