USDA's Sept. 30 quarterly Grain Stocks report came in much lower than expected for both corn and soybeans. Since these numbers become the beginning stocks estimate for the 2019-20 balance sheet, it has potentially large consequences for the next World Agricultural Supply and Demand Estimates report that comes out Oct. 10.
On soybeans, USDA put stocks as of Sept. 1 at 913 million bushels (mb). Analysts had expected USDA would put stocks somewhere between 940 mb and 1.001 billion bushels (bb). But as is customary in the September report, USDA adjusted 2018's production estimates, lowering the national average yield by a bushel and trimming planted acreage, thus accounting for part of the lower-than-expected stocks figure.
On the soybean balance sheet, USDA gets detailed data on all of the major demand categories (crush, exports, seed use), which allows it to reconcile any errors with production at the end of the marketing year when it gets a finally tally of stocks as of Sept. 1.
USDA can't do the same thing for corn, largely because it doesn't have a statistically valid way to quantify feed usage. That's why it lumps feed and residual together on its supply and demand balance sheet, and why it never formally revises its production estimates like it does with soybeans.
So when corn stocks estimates on Sept. 1 came in at 2.114 billion bushels, well below the low end of forecast range of 2.37 bb, it raised questions. Did we feed more than anticipated last year? Or is it included in the "residual," a potential reflection of bushels lost due to last year's soggy harvest season?
Hopefully, USDA provides a better idea when it updates its supply and demand balance sheets next month. In the meantime, we're left to wonder at where the corn went, or if it was ever there in the first place.
It's worth remembering that it's not the first time residual use has raised questions. In 2014's July WASDE, USDA put the soybean residual at a -69 mb. At the time, it seemed like USDA created soybeans out of thin air. The National Agricultural Statistics Service eventually raised its production estimate for 2013, but at the time the report came out, it caused quite a stir.
Back then, I visited with a USDA economist who explained the role residual plays in USDA's statistical sausage making. I've included the bulk of an article I wrote back then below, editing out some old market commentary to avoid confusion. I hope it helps to put this Grain Stocks report into perspective.
-- Originally published July 21, 2014
Getting to the Root of Residual
USDA Analyst Explains the Change to a Poorly Understood S&D Category
USDA raised a simple question when it dropped the soybean residual to -69 mb in the July 11 supply and demand report: How do you create soybeans out of thin air? A USDA oilseed analyst explains.
OMAHA (DTN) -- USDA raised a simple question when it dropped the soybean residual to -69 million bushels in the July 11 supply and demand report: How do you create soybeans out of thin air?
The answer is anything but simple. It's a mixture of mathematics and the relationship between two USDA crop forecasting agencies.
Keith Menzie, an oilseed analyst with USDA's World Agricultural Outlook Board, walked DTN through USDA's thought process on dropping the residual into negative territory. He explained what residual is, its role on the balance sheet and how the soybean balance sheet reconciles at the end of the marketing year.
Residual is a statistical term. In layman's terms, it's a numeric value that represents how far off your original predictions were. USDA uses residual on the demand side of the balance sheet "to account for disappearance or usage that cannot be verified or cross-checked against another objective information source."
That includes uses USDA doesn't measure, such as food use of soybeans, Menzie said. But it also includes something more abstract: discrepancies in measurement.
"If you measure yourself on a scale three times, you might not get the same number each time. Instruments have a measurement error," Menzie said.
When you're trying to measure the volume of a large quantity of something, measurement error is more likely. For example, assume the Commerce Department's Census Bureau measures 1.600 billion bushels of soybean exports, but in reality, 1.605 bb were sent out on ships. Since the initial measurement was off, no one would ever know more beans were actually exported.
"To the degree that we don't correctly measure the true volume, the difference ends up in the residual," Menzie said.
There's also the possibility for error in USDA's soybean crush estimate. The Census Bureau stopped surveying every processor in the country due to budget pressure three years ago. Now, USDA analysts estimate total crush from the National Oilseed Processors Association crush report, which includes fewer processors.
"We're getting into estimating a number from another number, and there's room for that to not be precisely correct," Menzie said. "Again, the difference would become part of the residual."
RESIDUAL'S ROLE IN RECONCILIATION
"For soybeans, unlike any other crop, we have a pretty good way of truing up what the crop size is," said Bill Lapp, president of Advanced Economic Solutions. There's good data for the two main usage categories -- exports and crush -- unlike corn and wheat. There's no reliable survey of feed usage for those grains, so it's lumped with residual in the balance sheets. The National Agricultural Statistics Service's doesn't adjust production for those crops after the January final numbers come out. Recent research from a group of Illinois professors has suggested that the inability to reconcile the corn balance sheet contributed to large surprises in quarterly stocks figures in recent years.
"I like to say the corn crop estimate is more of a Protestant thing, where you pray and you go to church but you don't go to confession," Lapp said. "Soybeans are more like Roman Catholics. You have to confess."
As a result, it's not unusual for NASS to make sometimes large changes to its production estimate in conjunction with the September stocks report, he said.
On soybeans, the NASS's surveys establish beginning stocks, production and ending stocks. Soybean for seed use relates to acreage. Crush, export and import totals are tallied. The residual becomes the only variable in the balance sheet formula (supplies - use - x = ending stocks).
NASS reevaluates its production number in the September Grain Stocks report in light of its stock figure, disappearance data and farm program administrative data. Over the last 10 crop years, NASS left the production number unchanged twice. It revised production up and down over that time. The largest change was to the 2007 soybean crop when NASS increased production by 90.6 mb.
That was also the last time USDA reported a negative residual figure, Menzie said. The crop year started out with 84 mb designated for residual use. It dropped to 2 mb in April and -35 mb in July.
Menzie said the decision to drop to a -35 residual was made using an ending stocks estimate of 125 mb.
"If we had known all the facts, we would have gone a lot farther than" -35 mb on residual, Menzie said. NASS showed ending stocks of 205 mb. With an accurate prediction of the final ending stocks figure, the residual would have plunged to -120 mb.
Katie Dehlinger can be reached at Katie.firstname.lastname@example.org
Follow her on Twitter @KatieD_DTN
Copyright 2019 DTN/The Progressive Farmer. All rights reserved.