By Mike McGinnis
DTN Staff Reporter
DES MOINES (DTN)--Because forecasters have estimated Brazil's 2005 soybean production within a few percentage points, nobody should be concerned about the U.S. Department of Agriculture's projected 66 million metric tons, which is at the top of end of a range of estimates, a Foreign Agriculture Service (FAS) production estimate said Sept. 14.
The 2005 South American soybean crop is forecast at 4.15 billion bushels, nearly 22 percent larger than the 2004 crop and 20 percent larger than the record crop of 2003. The boost in crop size would be a result of an increase in soybean acres in Brazil.
In its latest estimate, USDA forecast Brazil's 2005 soybean area at 23.5 million hectares (58 million acres), up 10 percent from last year.
"The estimate range is between 2 percent and 10 percent above last year," said Michael Shean, USDA's FAS production estimate and crop assessment division analyst. "Every forecaster is within the margin of error of between 5 percent and 15 percent.
But the estimates are for a crop that has yet to be planted. Coming up with estimates for seeds not yet sown involves using a number of sources providing anecdotal evidence, Shean said.
When determining a production estimate for Brazil, USDA and APHIS relies on state reports, data from the government of Brazil, private agriculture consultants in Brazil, farmers, co-operatives, Embrapa (Brazilian Agricultural Research Corporation), Cargill, Bunge, ADM, OilWorld and others.
"There could be between 12 to 20 sources we use to collect data," Shean said. "Some provide more than national estimates, they include state estimates.
"You can't make sense of a national estimate if you don't know how they made it up. You can't just say our estimate is lower due to good crop conditions in the south without knowing what the conditions in the north are like, so state estimates are needed."
Because Brazil doesn't have an agency similar to the U.S. National Agricultural Statistics Service (NASS) to provide ag statistics, there is no quantification in crop area or yield in Brazil.
This makes it difficult for FAS to rely on the accuracy of the government of Brazil estimates, Shean said.
"There is no way to determine which number is the best number," he said. "So well after harvest we will adopt official government statistics, but until that time we are forecasting what we think the final area will be."
Along with anecdotal source information, FAS estimates Brazil soybean production by using post-harvest final crushing figures, export totals, validating stocks and by estimating internal demand by the size and scope of poultry and hog industry in Brazil, Shean said.
"We then cross check the production number with our assumption and what can be proven," he said. "Most years the Brazil production number comes close to USDA number. Some years it doesn't. Some would say it's too much of a difference while others say it's a good fudge factor."
When it comes to forecasting a soybean yield for Brazil, people have to use historical data, Shean said.
"We use the last 20 years of yield data and study patterns in that data," he said. "The data is saying strongly the yields have increased. But, in the last three to four years, yields have stabilized because of rust and drought."
For 2005, because of a plateau-like pattern, USDA analysts have projected a Brazilian soybean yield 2 percent lower than the long-term trend.
Asian Soybean Rust disease is a factor when estimating Brazilian yield, Shean said, but how much a factor is unknown.
"We considered it in every state it was reported," he said. "We tried to decide among ourselves, will rust be controllable with correct management or will it be uncontrollable. We are going with it will be better managed but have a similar impact."
In 2004, Brazil had an estimated 4.4 million metric tons of soybeans not harvested because of soybean rust, said Tadashi Yorinori, an expert on Asian soybean rust.
"That could increase in 2004-2005 because farmers in southern Brazil that didn't have rust last season may be hesitant to spray like they should," he said.
Last season, grain losses in Brazil amounted to $1.2 million, or $275 for each metric ton.
Shean said the U.S.' FAS assigns several committees to debate Brazil's state yields so USDA officials feel comfortable with its crop estimate.
"A lot of information is reviewed to come up with a forecast long before the crop is planted," he said.
Once planting starts in Brazil, satellite and weather data become more important as do soil-moisture models. The post-planting information help the FAS assure accuracy in its data, Shean said.
"We keep collecting forecasts from various groups, monitor weather and compare it to the previous year," he said, "and use satellite data to determine the health of the crop."