Food price forecast

A food price forecast here is a vintage: each month USDA's Economic Research Service issues forecasts of annual category-level price changes, roughly 22 CPI item categories per vintage, alongside the underlying historical CPI and PPI series. Meats get named detail - beef and veal, pork and poultry - plus dairy, fats and oils, and processed food categories. The published tables run annual percent-change history for 1974-2025 and recent change tables spanning 2024-2027; the historical CPI forecast series reaches back to the early 2000s, while archived CPI forecasts extend to 2003 and PPI archives to 2014. Files are XLSX and CSV. Because the underlying indexes come from BLS CPI and PPI programs, ERS's contribution is the forecasting overlay on top of those indexes.

Forecast vintages are revision-prone by construction, so casual handling corrupts any inflation story.

• Methodology break at 2023 is explicit. ERS warns pre-2023 archived forecasts were produced under a different methodology and should not be mixed with the recomputed historical series - averaging across that seam invents accuracy.
• Vintage timing defines the backtest. A model scored against today's revised tables grades itself against information the forecaster never had; score against as-issued vintages instead.
• Category coverage sets the ceiling. With about 22 CPI categories per vintage, sub-category claims beyond meats detail exceed the product.

1. Record the vintage month with every pull. Forecasts arrive as monthly vintages of annual category changes; undated extracts are unusable for evaluation.

1. Quarantine the pre-2023 archived series. ERS states the older methodology should not be mixed with the recomputed history - keep separate tables and labels.

1. Check category breadth against your use case. About 22 CPI item categories per vintage, with named detail for beef and veal, pork and poultry; dairy, fats and oils, and processed foods also appear.

1. Credit the underlying indexes. CPI and PPI indexes originate with BLS programs; ERS adds the forecasting overlay - cite both when publishing.

Entries that sit next to Food price forecast in this glossary:

• us-government-public-domain