(Jan. 24, 2012) A new research report, funded by the Office of the Chief Economist of the U.S. Department of Agriculture (USDA), examined the accuracy of quarterly estimates of corn inventories. It was conducted by University of Illinois at Urbana-Champaign Professors Scott Irwin and Darrel Good, and Southern Illinois University-Carbondale Professor Dwight Sanders.

They found that - in spite of on-going efforts to maintain the quality of information provided and the transparency of the methodology used - misunderstandings, concerns, or complaints about corn and soy reports still exist.

For additional information, see the full research report, or see a summary below.

Findings

The researchers' primary finding was that quarterly corn stocks have gotten much harder to predict since 2006. They examined numerous reasons for this, such as:

  • biased survey respondents;
  • mixing of old and new crop stocks;
  • changing geography of corn production;
  • changing number and size of on-farm grain facilities;
  • changing make-up of the survey sample during the marketing year;
  • reduced survey response rates,
  • more ethanol plants;
  • more corn in transit; and
  • inaccurate estimates of corn used for ethanol.

However, they believe the explanation with the most merit is that unresolved errors in production estimates for corn led to the large surprises.

To most market analysts, the January corn production estimate generally is viewed as the final production estimate. But USDA's National Agricultural Statistics Service (NASS) always is careful to include a detailed discussion of potential sampling errors, and those errors amount to hundreds of millions of bushels of over/under-accounted corn. The production estimate errors are believed to have been amplified in recent years by unusual growing season weather conditions.

Researchers also found that there is a general tendency for NASS from August through November to underestimate the final soybean yield forecast. While the underestimate is not significant, it does impact estimates and thus reports.

Recommendations

To improve the accuracy of USDA yield forecasts and estimates, the researchers made several recommendations:

  1. USDA's World Agricultural Outlook Board (WAOB) should outline the exact process used to determine corn and soybean yield forecasts.
  2. NASS should review its soybean yield forecasting procedures for sources of bias.
  3. NASS should provide its yield forecasts derived from both the agricultural yield survey and the object yield survey.

For corn and soybean stock estimates, the researchers recommend that NASS improve the accuracy of on-farm stock estimates by telling farmers to estimate stocks using 56 pound bushels for corn and 60 pound bushels for soybeans. Also, the researchers recommend NASS conduct an internal review of corn stock estimation procedures, and that NASS add grain stocks questions to the Agricultural Census.

For improving domestic usage forecasts and estimates, the researchers recommend NASS add surveys of corn for feed use and corn for ethanol use. Lastly, the researchers recommend that USDA create a replacement report for the discontinued (in July 2011) Census Bureau oilseed crushings report, which was used to gauge the rate of consumption of soybeans and soybean products during the marketing year and to judge the strength of demand for those commodities.

By Max Fisher, Director of Economics and Government Affairs

distributed by