Annemarie Kuhns, Richard Volpe, Ephraim Leibtag, and Ed Roeger
How USDA Forecasts Retail Food
Price Inflation
What Is the Issue?
Each month, USDA's Economic Research Service (ERS) publishes wholesale and retail price
forecasts for various food categories and subcategories, and policymakers, food suppliers, and
researchers rely on these numbers. In recent years, as commodity and food prices have become
more volatile and less easily predicted, users of the forecasts have expressed a greater need for
more accurate forecasts.
ERS continually explores ways to improve its forecasts as new data and methods become avail-
able. In 2011, ERS revised its food price forecast methodology to use more rigorous statistical
techniques and capture the impacts of the multistage U.S. food supply system on wholesale and
retail food price formation. This updated approach incorporates far richer data available for
farm, wholesale, and input prices, which could lead to more accurate forecasts.
What Did the Study Find?
· The precision of ERS food price forecasts has observably improved with the revised meth-
odology. As a result, ERS food price forecasts are, on average, closer to the realized inflation
figures.
· ERS forecasts using the new methodology required fewer and smaller revisions. For a given
year, forecasts are subject to revision during a 17-month period. An average of 3.2 changes
were made per food category using the new, current forecasting methods compared with 3.7
revisions using the previous method, and the average size of the adjustments dropped from
2.6 to 2.1 percentage points.
· Another measure of forecast accuracy was the extent to which initial forecasts differed from
the actual Consumer Price Index (CPI) values. Using revised forecast methodology, the
average difference from CPI values was 2 percentage points, compared with 2.6 percentage
points for the previous methods.
· Although forecast accuracy and precision have improved relative to less rigorous
approaches used by ERS before 2011, more years of data are needed to fully assess forecast
performance.
