Grain analysts are nervously expecting the U.S. Agriculture Department to drop another bombshell on Friday when it issues its U.S. quarterly corn stockpile estimate, a key number they have wildly missed forecasting the past three years.
The missed forecasts have resulted in wild price moves in the world's biggest grain market, Chicago Board of Trade futures.
"None of us has a magic bullet. I spend a lot of time trying to figure it out, and ultimately you're completely wrong," said Tim Emslie, grains research manager for CHS Inc, the largest U.S. farm cooperative and a major corn hedger.
Four times a year the USDA surveys commercial grain firms, country elevators, processors, and farmers to report on the amount of corn held in storage, a key economic indicator the grain trade uses to project the actual amount of corn on hand before the autumn harvest.
The United States is the world's largest grower and exporter of corn, which is also the largest cash grain for farmers. So USDA's grain stocks estimates for decades have been a bellwether for the futures market to adjust prices to market fundamentals.
But in recent years the calculus has grown trickier. Drought and other weather factors have jolted production and use. Export demand has surged as China tapped more U.S. supplies. Most importantly, a major new domestic corn consumer has appeared - ethanol makers, who last year ate up a full 40 percent of the corn crop from well below 10 percent a few years ago.
Corn itself is the main feed grain for U.S. livestock. But ethanol plants have muddied that picture by mammoth production of a byproduct called distiller's dried grain (DDG) - also now used as a feed and feed export. Cheaper competing grains like wheat have also thrown in another wild card to calculations.
The biggest miss in analysts' corn stocks estimates in the last 30 years occurred in March, the most recent quarterly stocks report. USDA reported March 1 corn stocks totaled 5.4 billion bushels - roughly 400 million bushels more than the average analyst estimate.
For the next report due at noon EDT (1600 GMT) on Friday, grain analysts polled by Reuters estimated June 1 U.S. corn stocks between 2.952-2.725 billion bushels, with the average at 2.854 billion - down nearly 10 percent from 2012. Some analysts, in giving their educated guesses, admitted that their estimate could be 250 to 300 million bushels above or below what USDA ends up reporting.
Complaints about the USDA report have escalated. USDA chief economist Joe Glauber told Reuters last week he had commissioned a study to research why private analyst expectations have diverged so much from USDA's official estimates. Glauber said that since there is a larger percentage of grain now being held away from farms, USDA surveys should actually be seen as more accurate, not less.
"I agree with that assessment," Darrel Good, a University of Illinois agricultural economist, said this week. "The problem is that unlike production estimates there is no way to know what is the correct stocks number. The fact that the methodology has not changed, and that significant issues with wheat and soybeans have not emerged leaves us without an explanation of the corn surprises."
Corn market analysts do their best to tout up a running tally of consumption. They use weekly U.S. government data on corn exports and weekly industry reports on ethanol production to gauge corn consumption and drawdown of stocks. But feed use remains the biggest guessing game.
"This feed usage number is all over the place. It's obviously a fudge factor for the USDA to make the balance table and the stocks report fit," said Rich Feltes, vice president of commodity research for brokerage RJ O'Brien, referring to USDA's benchmark monthly world supply/demand tables. "It just underscores how high risk this report is."
The USDA's corn marketing year begins Sept. 1, and ends August 31.
Over the past five years, corn for feed and residual use in the March-May quarter of the year has ranged from 718 million bushels in 2011 to 1.276 billion in 2010, analysts say. Last year, USDA put the number at 858 million bushels. The average trade estimate by analysts for third quarter feed use is near 840 million bushels going into Friday's report.
Analysts concoct their own formulas and ratios. Emslie said he factors in livestock herd numbers, the price ratio between wheat and corn - since wheat is commonly used as a feed substitute - and the availability of DDG's to come up with a number for feed use. They also try to factor in seasonal discrepancies, such as when new-crop corn may be harvested early in the autumn and fed in the fourth quarter.
Bill Lapp, head of Omaha, Nebraska, based consultancy Advanced Economic Solutions and a former chief economist at ConAgra, said analysts "have a limited number of tools to predict the corn stocks number" compared to straight crop production reports based on facts like planted acres, seed types, moisture and temperature, field surveys, and so on.
Lapp just completed a survey of dozens of active commercial users' USDA crop data. The bottom line? Lapp said that those responding gave the government passing marks on their grain reports - with grain stocks the exception.
"That one that stands out for the greatest room for improvement is the stocks report," said Lapp, noting that 90 percent of those surveyed said it needed to be improved.