Column: U.S. Corn, Soy Output is Basically Settled, According to Analysts

(Reuters) – Some market participants have been waiting for the U.S. Department of Agriculture to finally acknowledge that the 2019 U.S. corn and soybean harvests will not be as large as the agency has previously stated, primarily due to historically late planting and crop development.

But industry analysts polled ahead of USDA’s next update, set for Thursday at noon EDT (1600 GMT), do not see an extraordinary amount of further downside for the harvests. The ranges of ideas are also relatively narrow.

On average, the sixteen analysts polled by Reuters believe USDA will peg U.S. corn yield at 167.5 bushels per acre on Thursday, below the agency’s prior estimate of 168.2 bpa. The guesses range from 166 to 168.7 bpa, the lowest corn yield range in an October poll in 13 years.

Soybean yield is seen at 47.3 bpa, below the September figure of 47.9 bpa, and the range is 46.8 to 48.2. That spread of 1.4 bpa is the smallest for an October poll in four years. The previous two years featured a trade range of 3 bpa on soybean yield in October.

But why such a slim range of guesses? Perhaps analysts are hesitant to put out a riskier estimate for fear of standing out and then being wrong. Or maybe USDA’s recent estimates are actually within the ballpark.

The way Reuters conducts its polls might initially seem to offer explanation on the narrow ranges. Several years back, Reuters split the polls into “Predicting USDA” and “Analysts’ Own Estimates,” since what USDA says on report day influences the market to a much larger degree than what analysts believe the final number will be several weeks or months down the road.

Some market participants have been critical of this format, but the separation does not explain this month’s mob mentality. Of the 18 analysts that responded with their own estimates, some 83% of them see corn yield at 165 bpa or above, and 83% see soybean yield between 46 and 47.9 bpa.

The Bloomberg poll has even more data points indicative of a group think. Of the 31 respondents, some 81% saw corn yield between 166 and 169.3 bpa, and 90% pegged soybean yield between 46 and 47.9 bpa. Bloomberg has just one poll that does not specifically indicate whether the analysts are predicting USDA or offering their own guess.

Mob mentality ahead of a USDA report has already plagued the market several times this year, most notably on June 28 and Aug. 12, when analysts were certain of drastic cuts to corn area, yield, and/or stocks. Chicago corn futures traded down the daily limit on Aug. 12 and close to it on June 28.

RECENT TRENDS

For the analysts attempting to guess USDA’s numbers, history indicates they are justified in their conservative yield estimates this month.

In the last 25 years, October corn yield has landed more than 2 bpa below the September figure only four times. The 2010 deviation was the largest at 6.7 bpa, and corn yield landed 4.5 bpa lower in 1995. The other two years were 1993 and 2000.

For soybeans, there are only three instances in the last quarter century where October yield was at least 1 bpa lower than in September. The biggest deviation was 2.4 bpa in 2003, when an aphid outbreak significantly reduced yield. The other two years were 1995 and 1998.

But it is not just yield in question this month. In its October report, USDA’s National Agricultural Statistics Service (NASS) adjusts acreage based on registration data, if necessary. On average, analysts see corn harvested area falling from the current estimate by 360,000 acres, and soybean harvested area down 161,000 acres.

These would be relatively light changes if realized, especially for soybeans. Except for a no-change in 2016, soybean harvested area has moved as little as 600,000 acres and as much as 1.1 million acres in the last five Octobers.

The largest move between September and October on soybean harvested area within the last couple of decades was a 1.2 million-acre reduction in 2010.

Corn area has been more controversial this year as the acreages have continually come in higher than many had expected. However, corn harvested area tends to move less than that of soybeans in October.

The figure was basically unchanged last year, but in the previous four Octobers, harvested corn area moved as much as 700,000 acres and as little as 200,000 acres. The largest change over the past two decades was a 1.1 million-acre reduction in 2008.

The most aggressive trade estimates see corn harvested area falling 1.22 million acres and soybean area falling 1.07 million acres. Some 60% of the analysts polled by Reuters see USDA’s October corn number down less than 250,000 acres from the September figure. Some 69% see soybean area basically unchanged.

USDA METHODS

Regardless of how analysts respond to a poll, it is important to not lose sight of the underlying methodology of the numbers in question.

The yield estimates are driven by producer surveys, objective yield samples and satellite imagery. For Thursday’s update, the surveys were conducted between Sept. 28 and Oct. 4, and the field samples were collected between Sept. 24 and Oct. 1.

There is no set formula as to which of those components figures most prominently into the estimates as historically, the best predictors of yield have varied by crop, location and month. However, the farmer surveys and objective yield data are the primary informants, while satellite imagery is used to supplement.

The best way to evaluate outcomes for Thursday is to consider what has changed within the last month. Is USDA likely to revisit the sample plots and find materially different results than a month earlier? Are farmers likely to have significantly changed their minds on expected yields?

The second question is more open-ended than in the past because farmers have fewer actual results to draw upon. As of Oct. 6, some 15% of corn had been harvested, behind the five-year average of 27%. Some 14% of soybeans had been cut, below the average of 34%.

The warmer-than-normal temperatures across the Corn Belt last month were exactly what an extremely late-developing crop would need, although the excessive rains in the Upper Midwest and Northern Plains have been a nuisance, especially in North Dakota.