How Accurate Is USDA's August Corn Yield Estimate?

Iowa Farm Scene

USDA released its first in-the-field assessment of the impact of 2012 drought on yields with its August Crop Report. What will September show?

Published on: August 28, 2012

On August 10, USDA's National Ag Statistics Service released its first official assessment of the 2012 drought on corn yields in Iowa and the United States. This first 2012 corn yield forecast is based on actual in the field measurements by trained enumerators. Before this August Crop Report was released, the previous estimates issued by USDA each month were projections based on computer models and data gathered from USDA's weekly weather and crop condition surveys.

The current Iowa corn yield forecast, according to USDA's August report, is for 141 bushels per acre, that's down from the 172 bushel per acre average yield Iowa farmers harvested in 2011. It's also below the 30-year trend line value of 180 bushels per acre (see accompanying chart, titled "30 Year Corn Yield Trends"). The August 2012 forecast for Iowa's average yield is 22% below the trend line. The forecast for the U.S., at 123 bushels per acre, is 23% below the trend line.

How Accurate Is USDA's August Corn Yield Estimate?

August yield estimate is first one based on in-the-field counts

"These forecasts arise from meticulous late-July counts and measurements by the team at USDA/NASS," explains Roger Elmore, Iowa State University Extension corn agronomist. In late July, USDA/NASS has enumerators go into fields and record stand counts and ear counts to obtain an estimate of ears per acre. Statisticians generate yield estimates based on how these counts and compare them to those of the last five years. 

In late July final kernel counts per ear and final kernel weights are still uncertain, notes Elmore. The assumption in the forecast is that weather during the remainder of the grain filling season is normal when compared to the previous five years. Thus, kernel numbers and kernel weights are 'normal' too. Of course no one knows if that is true or not until after the fact. There's still a lot of time ahead before the crop is actually finished developing and is harvested.

Stay-tuned for monthly USDA yield forecasts and fresh estimates

Each successive monthly forecast through November is a "new" forecast -- not simply an update. The USDA/NASS group uses new and expanded observations from a large number of sample locations to formulate expected yields based on field observations for every subsequent yield forecast. By the time the September yield forecast is published (based on late-August samples), kernel numbers are likely set and thus yield forecasts are more comparable to the final yields for 2012. 

Of course, it is possible the weather could be better than average for crop yields and actually increase the yield forecasts in subsequent months. This is usually associated with temperatures that extend the corn grain filling period beyond the normal timeframe.

Remember though that corn that is not harvested because of poor yield potential will not be included in yield forecast reports, Elmore points out. Statewide "yield" will actually increase simply because "yield" is defined as "yield per harvested acre." Thus acres dropped because of drought will not be considered in the harvested area estimates for 2012.

What's the track record? How does August forecast compare to final yield?

A number of organizations and groups other than USDA/NASS also attempt to get an early assessment of yield during the growing season each year. Their procedures undoubtedly differ from those used by USDA/NASS.

But in any case, says Elmore, the biggest issue with yield forecasts that are made before harvest, no matter who makes them, is obtaining an accurate estimate of kernel numbers per acre (plants per acre x ears per plant x kernels per ear), and final kernel weights. It becomes a numbers game: the more samples taken, the more accurate the estimate; and the closer to harvest the estimates are made, the closer the estimate to final yield.

With a few exceptions USDA's August forecast is within 10% of final yield

"It shouldn't surprise you that the August USDA/NASS yield forecast doesn't always reflect the final annual yield," he says. The accompanying chart, which Elmore put together, titled "Historical Track Record For August Forecast Corn Yields," shows the track record for USDA's August forecasts. With a few exceptions, the August yield forecasts are within 10% of the final yields for each year.

How Accurate Is USDA's August Corn Yield Estimate?

Most farmers will remember plausible reasons for the exceptions which are shown in the 'Track Record' chart. Those include the 1983 drought, 1993 floods, 1992 and 2004 long seed-fill periods, etc. As we go into the final stages of kernel fill, the monthly yield forecasts will edge closer to the final annual yield estimates for the state.

Meanwhile, with high-nighttime temperatures and dry conditions during the early seed set and seed-fill period, if that scenario occurs, "we can expect fewer kernels per ear and lighter kernel weights than we've seen for several years," Elmore says. "On the other hand, cooler nighttime temperatures and periodic gentle rains improve yield potential through increasing seed set and lengthening of the seed-fill period."