* Analysts grapple with WASDE after Q2 corn stocks surprise
* Cause of wide gap between USDA, private forecasts unclear
* Blame aimed at ethanol, early harvest, alternative feeds
By Karl Plume and Sam Nelson
CHICAGO, April 9 (Reuters) - Grain analysts continue to scratch their heads over a huge corn stocks estimate released by the U.S. government in late March, leading them to an unusually wide range of estimates for the next world agricultural supply and demand report (WASDE) from the U.S. Department of Agriculture report on Wednesday.
One division of USDA - the World Agricultural Outlook Board (WAOB) - will issue the monthly forecasts of how much grain will be produced and consumed worldwide in the coming year, including its view of how much corn will be left in the United States on Aug. 31, the end of the marketing year in the world's biggest grain exporting nation.
Corn, the biggest U.S. crop, is used mainly for livestock feed and ethanol production, but it is also utilized in dozens of industrial products from sweeteners to paint.
The WAOB uses the quarterly corn stocks estimate as a baseline to build subsequent monthly end-stocks forecasts. But quarterly stocks estimates are derived by another branch of the USDA - the National Agricultural Statistics Service (NASS).
To what degree the WAOB embraces the quarterly stocks data gathered by NASS is the issue. Analysts are uncertain, and it has led to the largest range of analysts' estimates ever ahead of a WASDE report - some 300 million bushels.
"This, in my career, happened in the early 1980s when stocks numbers were seen as less reliable," said Dan Basse, president of consultancy AgResource in Chicago. "I'm sure NASS will somehow, whether it's due to the ethanol industry and changes in the way corn is harvested in the Delta, work through it. But for the moment there's less reliability in these quarterly stocks estimates."
"Arriving at these stocks numbers and annual feed and residual (use) is more art than science," said Bill Lapp of Advanced Economic Solutions, the former top economist at food giant ConAgra.
Data on corn exports and ethanol production is released weekly. But feed consumption data is only released quarterly. So grain analysts try to work in factors such as livestock herd sizes and historical feeding trends to come to their usage estimates.
Analysts on average expect USDA on Wednesday to forecast U.S. corn ending stocks of 812 million bushels, a 17-year low due to last year's historic drought. But that number would also be up a shocking 180 million bushels from the March WASDE forecast.
"A year ago they came up with a surprisingly low March corn stocks number and WASDE didn't change their feed and residual in their April report because they didn't trust it. They were proven to be correct by the end of the crop year," said Basse.
NASS said U.S. corn stocks on March 1 were 5.399 billion bushels. That was 400 million bushels more than most traders expected, one of the largest misses in the past 30 years.
"This is still a mystery to everyone involved in the process. Disconnects between expectations and USDA stock estimates have occurred off and on in the past, but the frequency of the disconnect appears to have increased," University of Illinois economist Darrel Good told the Reuters Ags Forum, an online industry chat room.
Going back to March 2010, 11 of the past 13 NASS quarterly stocks figures have been well out of sync with private estimates, Good said, generating huge price swings in the Chicago Board of Trade corn futures market.
The latest instance sent prices down the 40-cent daily trading limit and nearly $1 a bushel since, although a strong corn basis in the cash market, an indicator of tight supplies, has raised doubts about stocks figure.
The recent history of NASS "losing" grain stocks in one quarter and then "finding" them the next added to uncertainty about its next quarterly report on June 28. Some think NASS may undo the March estimate with a large stocks cut in that report.
NASS says it stands by its quarterly estimates, gleaned from a survey of more than 83,000 farms and a survey of 8,800 off-farm storage facilities and reviewed for reasonableness and consistency with historical norms.
So grain analysts stay puzzled, though they offer theories for the discrepancies, such as difficulty in predicting corn use by ethanol makers or the use of cheaper alternative feeds by livestock and poultry producers. Some say the 2012 crop may have been undercounted or that larger corn crops outside of the traditional Midwest Corn Belt has skewed historical trends.
"The most obvious factor is that high prices lead to the most efficient use of corn possible, so we should have expected very low feeding rates per animal unit," said Tim Emslie, director of research for CHA Hedging.
Emslie, reflecting the complex calculus for corn analysts, also said "visibility" of corn stocks is greater when they are not in transit, as has been the case recently due to low corn exports and reduced use by ethanol makers. (Editing by Christine Stebbins, Jim Marshall and Bob Burgdorfer)