USDA Report Review

One Number to Rule Them All

Decreasing its domestic soybean ending stocks figure is what USDA does, with a pattern so consistent the constant overestimation seems systematic. (DTN chart by Darin Newsom)

In every set of monthly USDA Crop Production and Supply and Demand reports, including the October, there are hundreds of numbers for analysts and traders to look at. The spotlight is supposed to be on corn and soybean acres since USDA "might" use official Farm Service Agency data. More often than not, though, these acreage numbers change again in January. With all these choices of numbers for traders to get excited about, including the spotlighted acres, it's interesting that this time traders decided to focus on a number that was released at the end of September.

Watching the markets rather than the numbers as they come out, as I always do, I saw November soybeans shoot 14 cents higher, then 20 cents, then 30 cents, and finally 32 cents above Wednesday's close before taking a short break. Not knowing what numbers USDA had unsealed, my initial hunch was ending stocks had to be smaller than expected. Not that I have a particularly strong insight, but simple logic indicated that was the most likely candidate.

The average pre-report guess came in at 453 million bushels, down only 22 mb from USDA's mid-September guess of 475 mb. However, between the September and October round of guesses came the September Quarterly Stocks report. There USDA reported 44 mb less than its own September estimate, 301 mb, a number that not only became the new ending stocks figure for the 2016-2017 marketing year but also beginning stocks for 2017-2018. Remember, 44 mb less than what had been previously estimated.

P[L1] D[0x0] M[300x250] OOP[F] ADUNIT[] T[]

USDA's updated guess on 2017-2018 ending stocks on Thursday was 430 mb, certainly lower than the 453 mb average pre-report guess. But how did it relate to USDA's own previous new-crop ending stocks estimate? That's right, it was 45 mb less. So, accepting a couple rounding differences and the fact no other supply or demand guess changed for new-crop soybeans, the change in new-crop ending stocks was simply a function of the "final" ending stocks/beginning stocks number released on Sept. 29.

And that was the only number that mattered Thursday. Corn yield, production and ending stocks were all increased, and the market did indeed want to go down, but spillover buying from soybeans didn't allow that to happen. Wheat... well, it did go lower due to a 27 mb increase in domestic ending stocks and a 5 million metric ton bump in world carryover. But neither of those numbers were much of a surprise.

On the subject of surprise, there were some ag economists, post-report, left scratching their heads over USDA's decreased soybean ending stocks figure. How was it possible, given that national average yield continues to be guessed at roughly 50 bushels per acre? The answer is simple: Decreasing its domestic soybean ending stocks figure is what USDA does, with a pattern so consistent the constant overestimation seems systematic.

Though still early, only six months into a 15-month cycle of guesses, the 2017-2018 path is similar to the one USDA took during the 2015-2016 marketing year. That year, USDA started in May of 2015 with a guess of 500 mb, trimming it to 425 mb by October. This year, the high guess of 495 mb occurred in USDA's June report, before the 430 mb scribbled in the box for October. The September 2016 Quarterly Stocks report concluded that marketing year's journey with a figure of 197 mb. For now, using the average of the last four years' declines, next September could come in at 215 mb.

Based on USDA's history, everything that comes before next September's quarterly stocks number, in relation to soybean ending stocks, should be viewed as completely irrelevant.

Darin Newsom can be reached at darin.newsom@dtn.com

Follow Darin Newsom on Twitter @DarinNewsom

(AG/CZ)

P[L2] D[728x90] M[320x50] OOP[F] ADUNIT[] T[]
P[R1] D[300x250] M[300x250] OOP[F] ADUNIT[] T[]
P[R2] D[300x250] M[320x50] OOP[F] ADUNIT[] T[]
DIM[1x3] LBL[] SEL[] IDX[] TMPL[standalone] T[]
P[R3] D[300x250] M[0x0] OOP[F] ADUNIT[] T[]