Wide divergence between U.S.D.A., private crop estimates.

It is not often corn and soybean crop production expectations from private analysts miss the mark of estimates from the U.S. Department of Agriculture by a wide margin. But that was the case on Aug. 12 when U.S.D.A. numbers came in much higher than the trade expected, sending futures prices into a tailspin.

The U.S.D.A. forecast the 2015 U.S. corn crop at 13,686 million bus, down 4% from a record 14,216 million bus in 2014 but still the third largest ever. Based on Aug. 1 conditions, average yield was forecast at 168.8 bus an acre, down 2.2 bus from a record 171 bus an acre in 2014 but still the second highest ever. The U.S.D.A. numbers were 3% above both the average pre-report trade estimate of 13,318 million bus for production and 164.4 bus per acre for average yield.

Soybean production was forecast by the U.S.D.A. at 3,916 million bus, down 1% from 3,969 million bus last year. Average soybean yield was forecast at 46.9 bus an acre, down 0.9 bus from 47.8 bus an acre in 2014. The numbers were 7% above the trade average of 3,719 million bus for production and 5% above the trade average of 44.6 bus per acre for yield.

Many major trade houses issue forecasts ahead of key U.S.D.A. reports. Those forecasts are used internally and gathered by the major news services — Dow Jones Newswires and Reuters — which also calculate an average of all the estimates a few days before the monthly reports are issued. When the reports are released, the trade then gauges the U.S.D.A. numbers in relation to trade expectations. If the U.S.D.A. numbers are relatively close to the trade average, there typically is little movement in futures prices. If the U.S.D.A. numbers vary significantly from the trade average, and especially if the numbers miss the full range of trade expectations as was the case for some of the Aug. 12 data, a much greater reaction may be expected.

The trade develops its numbers by various methods, from internal computer models that utilize U.S.D.A. weekly crop ratings, acreage data, historical trends, weather and other inputs, to outside crop indexes, anecdotal reports, farmer surveys and field inspections. The first production number of the season typically is the most difficult to peg.

Beginning in May for winter wheat, July for spring wheat and other small grains and August for corn, soybeans, sorghum, rice and other row crops, the U.S.D.A. conducts massive farmer surveys as well as actual field visits. The August forecast included about 23,000 producers in major states that account for about 75% of U.S. production. It is generally accepted that the U.S.D.A. procedures and results are the best of any other country in the world.

A number of factors appear to have contributed to the wide variance in trade and U.S.D.A. numbers in August. Some analysts tend to look at the U.S.D.A. side while others focus on the trade side.

One analyst suggested higher-than-expected U.S.D.A. yield forecasts in areas most affected by excessive June rains (the eastern Corn Belt) may have resulted from the U.S.D.A.'s use of average ear weights for corn and pod counts for soybeans. Some also suggest the U.S.D.A. may have put more weight on June and July weather models that showed the weather helping yields rather than hurting yields. But others said the U.S.D.A.'s methodology should account for such variations.

Another analyst said he thought the trade put too much emphasis on the bad areas in the eastern Corn Belt versus the good areas in the central and western parts of the Corn Belt. He also suggested varietal improvements from seed companies may be making a greater impact on yields last year and this year than many in the trade realize, offsetting some weather stress and thus boosting the yield trend faster than in the past.

Adding to the debate last week was the annual Pro Farmer Midwest Crop Tour, which visited fields in Nebraska, South Dakota, Minnesota, Iowa, Illinois, Ohio and Indiana — the heart of the Corn Belt. Analysts noted the tour's daily estimates did not support the U.S.D.A.'s high numbers in several states.

Corn and soybean futures prices firmed last week on reports from the tour.

Also added to the mix last week was "prevented planting" acres reported by the U.S.D.A.'s Farm Service Agency for crop insurance and farm program purposes. That report said 2.3 million acres intended for corn and 2.17 million acres intended for soybeans were not planted, with the soybean number the highest since the U.S.D.A. first started releasing prevented planting data in 2007. Still, analysts maintain that yield is more significant in determining final production numbers than are acres, although they note the F.S.A. data was not included in the August crop production estimates.

Most analysts still expect final estimates for both corn and soybean crops, which won't be known until January, will end up smaller than the initial U.S.D.A. August forecasts.

While the trade may not always agree with the results, the U.S.D.A. numbers are considered the "official" forecast, even if they are subject to change in later reports. As one analyst said, "The market will trade the numbers until they change."