Statisticians Share Crop Data Collecting Techniques
Thursday, March 15th, 2012
Where does the U.S. Department of Agriculture get its numbers for the reports it compiles and releases, such as the Prospective Planting report to be published at the end of this month?
The process is more labor-intensive and scientific than growers might realize, noted Greg Preston and Greg Matli, statisticians from the USDA’s National Agricultural Statistics Service, who spoke to a group of producers at the Indiana Livestock, Forage and Grain Forum in Indianapolis.
NASS releases crop production reports each month, including the August through November reports, which contain data from the agricultural yield surveys and the corn and soybean objective yield survey. They also release a crop acreage and stocks report each quarter.
“The quarterly crop and stocks report comes is different for each quarter — in March, we release the prospective plantings report; in June, we present the planted acreage report; in September, the small grains report; and in December, the row crop yields,” Matli explained.
“We’re also collecting on-farm stocks information to show the amount of corn, soybeans, wheat and oats stored on the farm so we can take the on-farm component and add it to the off-farm component for the total stocks issued in the grain stocks report at that same time,” he stressed.
Key to the data the statisticians collect for each report is sample size. A larger sample size ensures better statistics for data collection, Preston noted.
NASS collects agricultural yields from farmer-reported yields on fewer than 700 farms in Indiana, which differ from the objective yields derived from 170 sample fields with two plots per field.
Agricultural yield surveys are sent directly to farmers in May and August for wheat yields and in August through November for corn and soybean yields.
The agency objective yield survey samples are collected from the June enumerative survey, which consists of 254 area segments. Segments are areas of land about 640 acres in size.
“We interview farmers to find out what’s actually in those fields throughout the state,” Matli said. “The farmer already has been identified and the field identified — they just don’t find out until July that they’ve been sampled and have a chance to participate in objective yield process. We always get permission from them to go out into their fields.”
“We only have 170 corn objective yield samples,” he added.
During the growing season, the size of growers’ farms determines their participation in surveys. The statisticians randomly select a field that they think will be representative of all the fields across the state.
“You know you made it in Indiana when you made it into all our surveys,” Preston noted.
Each January, they take the data collected from the year before to compile final estimates, and publish the final acreage.
They compile the acreage from the quarterly surveys into annual surveys.
“We take all the information and stratify it by the size of the farm,” Preston said. “Some farmers may not have a lot of acreage to push into their area. We try to keep a handle on the areas around the state that are having problems.”
“We have a fairly high response rate in Indiana,” he said. “All we do is take the information that farmers provide to us and pull it together at the U.S. level.”
There are disadvantages to collecting agricultural yield data. Yields can be subjective to one side of a field where yields are higher, Matli acknowledged.
Time also has elapsed between the time data is collected and when it is released to the public.
“When we first release information Aug. 12, a lot of time goes by — we’re referencing based on Aug. 1,” he said. “We would still put out a report that stuck to that Aug. 1 reference date.”
Statisticians also must consider the probability expansion factor, since farms are selected for surveys based on size and almost all the large-sized farms are surveyed, he noted.
The objective yield forecasts — a harbinger of the coming year for many farmers — are derived from corn in 10 states, soybeans in 11 states and cover more than 84 percent of the corn crop in terms of production, Matli noted.
The surveyors collect field survey identification in June during a personal interview from land area samples.
The yield is the weight of the sample collected when the field is harvested, Matli said.
“April 25 is the average planting date for Indiana,” he said. “We also check the stage of maturity it is in and assign an actual number to that maturity level based on certain criteria.”
“We also count stocks per acre and ears per acre — we take a lot into account when we actually start to set out estimates,” the statistician said. “A lot of people think we use a magic eight ball or a dart board, but it’s pretty scientific work.”
“In our office, we come up with one number for Indiana,” Matli said. “The secretary comes in at about 8:15 a.m. to sign the reports, and it comes out at 8:30 a.m., 1 p.m., 3 p.m., or 4 p.m., depending on the report.”
NASS’s crop production report and world agricultural supply and demand estimates are released at 8:30 a.m. Eastern time between the ninth and 12th day of each month.
For their county estimates survey, published each year, the specialists sometimes use as many as 17,000 farms for the sample size. They set a state first, followed by estimates for each district.
County estimates are based solely on farm-reported data. Farmers must respond for the data to be accurate, Matli noted.
Individual counties must add to districts and districts must add up to state estimates, as well.
NASS began using satellite imagery to provide more accurate acreage and yield indications.
Matli stressed the importance of farmers responding to inquiries about field surveys and cooperation with numerators.
He said that none of the NASS employees are politically motivated — an issue that farmers often consider when setting stock and yield reports.
“Secretary Vilsack cannot talk to any reporters until an hour after the reports are released,” he said.