By Elaine Kub
DTN Contributing Analyst
I know to what exact degree 21 grain market analysts were surprised Monday morning by USDA's September prognostications about the size of the 2016 corn and soybean crop. I know whether they were surprised at those numbers being too high or too low, and I know whether they were more surprised by the soybean estimates than the corn estimates.
I know all of this because one of our nation's premier farm broadcasters, knowing how much I love to dig down into spreadsheets, shared with me his raw data of analysts' pre-report estimates. Last week before the report, Todd Gleason of Illinois Public Media (radio station WILL Ag) collected corn and soybean yield and production estimates from the most trusted research and brokerage firms in the business. I begged him for the data and asked him to keep everything anonymous, of course, so no hard feelings anywhere. I myself freely admit to being mightily surprised by just how large the USDA projects this year's crop performance will be.
The conventional wisdom since last month's 15-billion-bushel corn crop projection seemed to be, "Wow. That's a lot. We think it'll be a large crop, sure, but not THAT large." Therefore, going into the September report, most analysts expected the USDA to trim their projected national corn yield below August's number, and to only barely bump up their projected national soybean yield.
One hypothesis I held was blown to smithereens once I looked at the data. I thought analysts might consistently either overestimate or underestimate both corn and soybeans, but that wasn't the case, at least not this month. There was only one analyst who overestimated USDA's soybean yield projection, with a guess of 50.9 bushels per acre, but that analyst significantly underestimated USDA's corn projection, with a guess of 170.7 bpa. A couple of analysts overshot the USDA on the corn yields (with one guess as high as 175.6 bpa), but aside from them, everyone else was in for a big surprise when they looked at the official projections in this month's report.
Overall, prior to the report, the analysts from throughout the trade media expected the September projection to be 171.9 bushels per acre for corn (resulting in a 14.9 billion-bushel crop). USDA's actual yield projection turned out to be 2.5 bushels per acre higher than that, at 174.4, putting the crop size at 15.1 billion bushels. Soybeans were the special surprise; the trade media expected to see a 49.1 bpa yield producing 4.1 billion bushels, and USDA delivered numbers showing a record-large 50.6 bpa yield producing 4.2 billion bushels. That was a bearish 3% higher than the analyst pool's underestimate.
WILL Ag's stable of on-air analysts did better than the rest of the media at guessing what USDA would do. Their average pre-report estimate of corn yield was off by only 1.6 bushels per acre, and their average soybean yield estimate was off by only 1.4 bpa. This is a good time to talk about why these analysts and trading firms even bother to make their pre-report estimates available to the public, or why they don't lie to the journalists and shade their numbers in one direction or the other. Ultimately, they're trying to sell advice, consulting, and brokerage services to new customers. An on-air, verifiable reputation for accurate analysis of crop supply goes a long way.
I caught up with one of these mythical beasts, an actual "pre-report analyst," who was willing to share how he determined his numbers (which were among the most accurate, I should add). Matt Bennett from Bennett Consulting in central Illinois said his process starts with a "look at the state estimates from the USDA, then I figure which side of their guess I'm on and how far off we are, then I do the calculations based on the USDA's acreage numbers."
Bennett's ultimate feeling about each state's yield comes from consulting with agronomists on what they're seeing in the field, and as the Channel Seed grain marketing consultant, he has a special "in" with the Corn Belt's scouts. "My research is based on that communication," he said, adding, "I think there is no substitute for being in the field myself."
So these pre-report estimates aren't just a drive-by job. But there is some gamesmanship behind them. The analysts aren't actually publishing what they think the *final* yield will be. They're publishing what they think the USDA will *say* on this one report. Bennett, for instance, after predicting 172.8 and 49.4 for corn and soybean yields on the September report, actually believes final yields on corn and beans may be 170.9 and 50.4 bushels per acre. "I think the corn crop is phenomenal in places," he said, "but really has issues in others. Beans are just darn good, and if it wasn't for sudden death syndrome disease, I think we'd have a 52-type national yield on beans."
Grain farmers and other bona-fide commercial hedgers do make up 60% of the corn futures market, but not all of them are Matt's clients. So how much does any single analyst's pre-report guess really matter to the actual performance of the market? Even if we consider all the clients of all the 21 forecasters whose numbers I looked at this month, that still isn't representative of the whole market, is it?
Maybe not, but looking at how far off those pre-report guesses are on any given report day -- that is, how big the surprise from the USDA is on that day -- is quite a good predictor of the market's performance at the end of that day. These pre-report estimates should be considered a proxy for how the "smart money" may be positioned in the futures market ahead of a report, and therefore, they seem to imply how those traders will be suddenly motivated to change their positions in the hours after a report.
There are always other forces at work on the markets on any given day, even a report day, but by looking at the daily market performances during the past 12 report days, I found a noticeable relationship by looking at how far the analyst pool underestimated or overestimated the report numbers, and comparing that to how far the corn or soybean market shot upward or downward in response to that error. The poster child example of this phenomenon was the May 2016 WASDE report, when the USDA soybean ending stocks number came in 40% below the analyst pool's overestimate, and the new-crop soybean futures contract closed the day 51 3/4 cents higher. On the other end of that spectrum was this most recent September WASDE report. The USDA's soybean ending stocks projection was 9% above the analyst pool's underestimate, and the market dropped 16 cents on Monday.
I don't think that fall was inevitable based on the bearish numbers alone -- the 50.6 bushel-per-acre crop or the unusually comfortable 8.9% stocks-to-use ratio -- but rather it seems like a result of how surprised the market was to see those numbers. Maybe if those 21 analysts whose guesses I looked at this week, and the broader market they represent, had seen the big crop projection coming all along, our day-to-day volatility this week would have been softened. In any case, it was a fun dive into the data, estimating the market effects of all these estimates of estimates.
Elaine Kub is the author of "Mastering the Grain Markets: How Profits Are Really Made" and can be reached at elaine@masteringthegrainmarkets.com or on Twitter @elainekub.
(CZ/BAS)
© Copyright 2016 DTN/The Progressive Farmer. All rights reserved.