Spatial support, release timing, and operational monitoring of U.S. and Brazilian soybean exports

Monthly trade statistics are released after conditions in an export system may have changed. This study uses separate weekly and monthly evaluations to examine how information timing and spatial support shape soybean-export monitoring before finalized monthly totals are available. Weekly national and port-node series are evaluated through expanding one-week-ahead forecasts; the U.S. national weekly benchmark and the common-date Brazilian national-versus-port comparison are reported separately. Monthly U.S., Brazil-total, and Brazil-to-China exports are evaluated with a common continuity-and-seasonality Elastic Net specification. The U.S. national weekly benchmark shows strong persistence and high correlation, but its continuity-plus-seasonality Ridge specification does not improve aggregate WAPE over a lag-1 naive forecast. On common Brazilian dates, the national series yielded lower WAPE than the individual port-node series. Across 144 monthly origins from 2013 to 2024, the continuity-and-seasonality specification outperformed seasonal naive forecasts for all three lines. Fixed-grid screening summarizes the historical trade-offs among event definitions and ratio cutoffs. The findings show that monitoring design depends on the timing and spatial support of the information available at each forecast origin.