Data: gold/hrw_stocks_prediction.parquet,
gold/hrs_stocks_prediction.parquet,
gold/srw_stocks_prediction.parquet,
gold/white_stocks_prediction.parquet,
gold/wheat_us_total_stocks_prediction.parquet,
gold/durum_stocks_prediction.parquet,
gold/corn_stocks_prediction.parquet,
gold/corn_us_total_stocks_prediction.parquet,
gold/bean_stocks_prediction.parquet,
gold/hrw_stocks_residual_predictor_analysis.parquet,
gold/hrs_stocks_residual_predictor_analysis.parquet,
gold/srw_stocks_residual_predictor_analysis.parquet,
gold/corn_stocks_residual_predictor_analysis.parquet,
predictions/ers_wheat_stocks_by_class.parquet.
Current-year production for prediction models uses yield prediction models by default
(winter wheat yield predictions for HRW/SRW, both winter and spring yield predictions for White, spring wheat yield predictions for HRS, corn and soybean yield prediction history for corn/soybeans).
Simple OLS: Y(stocks) = f(X), where X = prev quarter stocks − exports (non-production quarters) or X = prev stocks + production − exports (production quarter: Sep for wheat, Dec for corn/soybeans). Soybean models use national stocks, all soybean exports, and national NASS soybean crush. Highlighted = forward projection.
Simple model: X = prev quarter stocks only (no exports subtracted). Implicitly assumes average use and average production.
Sum of all wheat classes (HRW + HRS + SRW + White + Durum) from ERS by-class data. Highlighted = forward projection (sum of per-class predictions).
Top 5 one-variable candidates for explaining the September stocks model residual. Adjusted prediction = current simple-model prediction + predicted residual.
Top 5 one-variable candidates for explaining the September stocks model residual. Adjusted prediction = current simple-model prediction + predicted residual.
Top 5 one-variable candidates for explaining the June stocks model residual. Adjusted prediction = current simple-model prediction + predicted residual.
Independently fitted national class models for HRW, HRS, SRW, and White. Durum is excluded because its existing model does not use comparable export or production inputs.
Top 5 one-variable candidates for explaining the June stocks model residual. Adjusted prediction = current simple-model prediction + predicted residual.
Independently fitted national corn model using U.S. total stocks, national production, and all U.S. corn exports.
National model: X = prior-quarter national stocks minus all soybean exports minus national NASS soybean crush, with December adding current-year production. Forward exports use the soybean export matrix.