Home / Crops / Pop the hood on USDA yield estimates

Pop the hood on USDA yield estimates

STEVE JOHNSON Updated: 08/15/2013 @ 9:20am Farm Management Specialist with ISU Extension housed in Polk County, Iowa. Areas of expertise include crop marketing, grain contracts, government farm programs, crop insurance, farmland leasing and other crop risk management strategies. Reach Steve by e-mail at sdjohns@iastate.edu.

The USDA National Ag Statistics Service (NASS) on Monday, August 12, released the August Crop Production Report. Both objective yield and farm operator surveys were conducted between July 24 and August 6 to gather information on expected yields as of August 1, 2013.

Corn production is forecast to be a record-high 13.8 billion bushels with an average yield at 154.4 bushels per acre. If realized, that would be the third-highest yield on record. Total corn production is up 28% from 2012, an increase of 31 bushels per acre. The forecast for 89.1 million acres of corn harvested remains unchanged from the July 11 report.

Soybean production is forecast at 3.26 billion bushels, up 8% from last year. NASS forecasts 76.4 million acres of soybeans to be harvested, a drop of 500,000 acres from July 11. If realized, this will be the second largest harvested acreage on record. Based on August 1 conditions, soybean yields are expected to average 42.6 bushels per acre, an increase of 3 bushels per acre over 2012.

Statistical Methodology

The objective yield surveys were conducted by a random sample of operators in the 10 and 11 major corn and soybeans producing states, respectively. These states usually account for about 85% of the total U.S. production. Farm operators were interviewed for the objective yield survey and to update previously reported acreage data and also to seek permission to randomly locate two sample plots in 1,920 selected corn fields nationally, of which 290 are in Iowa. Operators were also contacted to locate two sample plots in 1,835 selected soybean fields nationally, 210 of which are in Iowa.

The counts made within each sample plot depend on the crop conditions and the maturity of that crop. In all cases, the number of plants is recorded along with other measurements that provide information to forecast the number of ears or pods and their weight. The counts are used with similar data from previous years to develop a projected biological yield. The average harvesting loss is subtracted this fall to obtain a net yield.

The plots are revisited each month until crop maturity. Each visit the fruit are harvested and weighed. After the farm operator has harvested the sample field, another plot is sampled to obtain current year harvesting loss.

The farm operator survey was conducted primarily by telephone with some use of mail, Internet, and personal interviews. Over 24,000 operators were interviewed nationwide; nearly 800 of which are in Iowa. During the survey period grower respondents are asked questions about probable yield. These growers will continue to be surveyed throughout the growing season to provide indications of average yields.

To assist in evaluating the reliability of the August 1 production forecast, the root-mean-square error, a statistical measure based on past performance, is computed. The deviation between the August 1 production forecast and the final estimate to be released in January is expressed as a percentage of the final estimate. The average of the squared percentage deviations for the latest 20-year period is computed.

The square root of the average becomes statistically the root-mean-square error. Probability statements can be made concerning expected differences in the current forecast relative to the final end-of-season estimate, assuming that factors affecting this year's forecast are not different from those influencing recent years.

For example, the root-mean-square error of the August 1 corn for grain production forecast is 6.0% and for soybeans 6.8%, respectively. This means that chances are 2 out of 3 that the current production forecast will not be above or below the final estimate by more than 6.0% or 6.8%. Using the 90% confidence level, chances are 9 out of 10 that the difference will not exceed 10.4% for corn or 11.8% for soybeans.

Also, shown in the table is a 20-year record for selected crops comparing the differences between the August 1 forecast and the final estimate to be released in January. Using corn as an example, changes between the August 1 forecast and the final estimate during the past 20 years have averaged 422 million bushels, ranging from 2 million bushels to 1.09 billion bushels. The August 1 forecast has been below the final estimate 10 times and above 10 times.

For soybeans, changes between the August 1 forecast and the final estimate during the last 20 years have averaged 152 million bushels. The range has been from 6 million bushels to 408 million bushels. The August 1 forecast for corn has been below the final estimate 11 times and above nine times. This does not imply that the August 1 corn and soybean forecast this year is likely to understate or overstate final production.  


USDA NASS goes to great lengths to compile information and provide results in the August production report. As a result, the monthly estimates and subsequent reports are deemed statistically accurate. The numbers are still estimates, not the final yields that will be released in January. Expect minor changes in both planted and harvested acreage numbers in subsequent monthly reports. NASS will reconcile acreage data from both the Farm Service Agency (FSA) and Risk Management Agency (RMA) and include in the October 11 report.    

CancelPost Comment

Table 08/14/2013 @ 2:18pm Which table are you referring to in statement "shown in the table is a 20-year record for selected crops comparing the differences" ?

Report Abuse Reply

2013 ACRE Revenue Shortfall Triggers Corn… By: 10/13/2014 @ 6:51am The Average Crop Revenue Election (ACRE), a countercyclical program created in the 2008 farm bill…

Using a basis contract to market corn By: 12/16/2013 @ 7:53am Despite a late start, the 2013 harvest progressed rapidly, and yield prospects increased across…

Guessing crop insurance payments By: 10/14/2013 @ 1:19pm Harvest prices determined in October will have a large impact on the size of the potential crop…

This container should display a .swf file. If not, you may need to upgrade your Flash player.
Successful Marketing Newsletter