Every year, one crop is chosen to be the focus of the National Agricultural Statistics Service (NASS) economic microscope. This year, it's rice.
USDA's NASS conducts three critical surveys yearly. One is the June survey when planted acres are first established. The December end-of-the-year survey is key for determining harvested acreage and yields. The third is the economic survey NASS does every February and March.
We go to farms and try to find out what it costs farmers to operate and how much money they make. We do that for the Economic Research Service. That information is summarized and presented to Congress in a July report, says Ben Klugh, Arkansas NASS state statistician.
The reason the survey is so important is this: it's a grower's chance to let Washington know how he's really faring down on the farm. It's especially critical now with new farm legislation peeking around the corner.
The survey is detailed. It takes at least an hour to complete and is very personal. It often takes some convincing from Klugh and colleagues to get a chosen farmer to do it. In Mississippi and Arkansas, the NASS normally has close to an 80 percent response rate among the highest in the nation.
The emphasis for the survey is different from year to year. This year's emphasis is rice. NASS won't get around to rice again for five or six years.
On Jan. 12, 2001, an article appeared in Delta Farm Press focusing on growers' and marketers' concerns with the data-gathering and number-crunching practices of NASS. Statisticians, it is fair to say, were more than slightly upset.
One reason stat folks were upset with the article was it came out just about the same time the (economic) survey was getting kicked off. We were going to farmers to get their economic readings and the article suggested we were trying to bury them, says Klugh.
Formed in 1863 by Abraham Lincoln, NASS goes to great lengths to distance itself from charges of cronyism or impropriety. Security surrounding NASS statistics is extremely tight.
We have no political appointees in the agency and are completely unbiased in performing our job. One of the major strengths of what we try to do is a release calendar. It may not sound too important. But we are mandated by Congress to put data out. Not only are we to come up with the numbers, but the calendar says we make them public on a certain time on a certain day.
The NASS calendar, adhered to strictly, is in stark contrast to the way agriculture statisticians work in many other countries. Concerns from American farmers about Chinese numbers, says Klugh, are justifiable because China is a major producer of so many commodities. Klugh has visited China in efforts to help that country develop a different statistical system.
In China, if the numbers come out close to the week it's scheduled for, they've done a good job. But collected data may sit for a long time before it is published, says Klugh.
Charges that NASS doesn't look at or use Farm Service Agency (FSA) certified acreage numbers are absurd, claim NASS statisticians.
If you go back and look at the track record, we're usually about 1 percent above their acreage numbers. This year, we're going to be pretty much right on top of the FSA numbers. Historically, though, there's about a 1 percent difference, says Klugh.
There are explanations
There are explanations for why NASS changes numbers. The last two years, NASS has received FSA acreage numbers at the end of August and beginning of September. They've never gotten the numbers so early before.
The first year we got (the August certified rice acreage numbers), we weren't sure exactly how to interpret them. The initial number we got was about 15,000 acres below the final December FSA number. The numbers were broken down by counties and we knew some of those numbers were not complete. This year's numbers were about 10,000 acres below the final number.
Some of the comments from farmers about us not contacting FSA are absurd. I have a hotline to the state FSA office to keep in touch and see what's going on. We want the certified acreage data as soon as we can get it, says Arkansas NASS deputy state statistician Doug Rundle.
Although FSA numbers are generally reliable, Statisticians also have occasional concerns with the numbers. In 1999, FSA's preliminary Arkansas number for soybeans in August had a substantial 111,000 acre bust in it. NASS caught the discrepancy and brought it to FSA's attention. Sure enough, when FSA's final report came out, the numbers were more accurate.
We use FSA data. Prior to 1996, we were always about 1 percent above. In 1996, with Freedom to Farm enacted, we were seeing more acres showing up than what FSA was indicating. We're not talking about a lot maybe going from 1 percent coverage to 3.5 percent.
Farmers certify their acres. That's true. But you're evaluating the difference between 10,000 acres over to 40,000 acres on a 1.4 million-acre crop. In 1997, it panned out. The census harvested acres for rice was 1.385 million. The FSA certified planted acres was 1.35 million. So there was a minimum 35,000 acres difference, says Klugh.
In1998 and 1999, NASS looked at the impact of this. We were looking at two consecutive years where we needed to be 35,000 to 40,000 acres above FSA. There had been new rice ground going in and some of it was on old soybean ground, which wouldn't have been part of a farmer's flexibility contract, says Rundle.
Up until 2000, for a farmer to participate in a program, the ground he farmed had to be in a production flexibility contract. For instance, any new rice ground on old soybean ground that hadn't previously been in rice wasn't eligible for an LDP.
Every March, NASS's projection of planted acres is eagerly awaited.
We don't try to second-guess a farmer. If he tells us he's planting 100 acres of a crop, we don't adjust his number to 95 acres.
In June, we go to about 4,000 Arkansas farmers a random sample of all kinds and types of farmers. This survey is broken down into two parts: farmer interviews and land area enumeration.
NASS looks at about 400 square miles in Arkansas and Mississippi. In each square mile, it does a complete census of what's being grown. From that NASS gets a fairly reliable estimate of what's planted in the state.
Arkansas' NASS has its most trouble with soybeans. When the survey is done around June 1, intentions might be to plant soybeans following wheat. But Arkansas may have about 35 to 40 percent of the crop not yet in the ground.
Combined with (the acreage census), we also sample from our (farmer list). On the list, we have 44,000 of the 49,000 farms in (Arkansas). If we want to sample corn, we have a list of farmers who grow corn. We draw a sample from that group. This is the cheapest way to get the most information with the fewest amount of contacts, says Klugh.
Lies, damned lies
Delta Farm Press has spoken with numerous farmers who claim they lie to statistical surveyors. How much trouble does NASS have getting farmers to answer questions truthfully?
What I throw back is we've been in existence since 1863. Most people use our numbers to make marketing decisions. If those numbers weren't good, if everyone lied, no one would use the numbers, says Klugh.
The other thing is when NASS collects data, it looks at relationships from year to year. For example, NASS produces rice yield estimates for each county.
Every year, we put together charts with weighted yields from counties. This takes a lot more observations than just setting a state average. Our final farmer yield that we put out in the state is based on about 600 reports. With our county survey, we got about 1,451 rice reports accounting for almost 700,000 harvested acres of rice. We use that to set the county levels.
By looking at data that way historical averages are easily seen. Certain counties historically produce better than others. What's interesting is if you looked at the same chart from years past, the pictures look the same. The yields change, but the distribution and relationship the ending data remains remarkably close.
So, if everyone is lying all the time, the pictures wouldn't look the same year after year. We'd all have to know ahead of time how everyone else is going to lie. Do some people lie? Probably, says Klugh.
Much has been made about why NASS doesn't simply do away with its stat-gathering in certain areas. Why not go to FSA to get the acreage numbers? Or the big commodity companies?
Simple, says Klugh: conflicts of interest and timing and scope of required estimates. I'm much more comfortable talking to a lot of people than to one or two folks that have market impact. One or two large market players can manipulate the situation. FSA and I'd never suggest they'd manipulate things has politically appointed leadership. NASS doesn't. Besides, when plantings estimates are required in March and June, certification by farmers at FSA is not complete. Those surveys are also used to estimate stocks of grain and GMO acreage, says Klugh.
And there have been instances where NASS bowed up and refused to accede to political pressure.
I was sitting in on a national board where we put out a stocks report. The numbers that came out were going to cost the USDA money. The secretary told (the NASS) administrator that he wasn't going to sign the report unless the numbers were changed. Our administrator told the secretary that he'd be the first to not sign such a report, says Klugh.
The report was published and signed with unchanged numbers.
Getting it right
The NASS sampling error nationally for June reports is 2 to 4 percent, depending on the crop. Would NASS like to be closer?
Of course. But in order to cut that percentage in half, we'd have to add four times to the sample size. Instead of going to 4,000 producers in Arkansas, we'd need to interview 16,000. That's impossible. Our budget is extremely limited and we do many crops, says Klugh.
Nothing is perfect and mistakes occasionally occur, says Klugh. One such is the oft-mentioned found rice acreage of 1998.
In December 1998, we got FSA acreage reports. The bottom four states are all below the FSA numbers. We can't have that so the revisions were to go slightly above, says Rundle.
That resulted in increases of 35,000 acres in Louisiana, 50,000 acres in Mississippi, 15,000 in Missouri, and 30,000 acres in Texas for a net of 130,000 acres.
Another thing that people need to remember is the crop production annual released in January. It's quite common for us to change acreage numbers if we need to. We have more data at the end of the year. When we ask someone about how many acres he planted and harvested at the end of the year, we're also trying to estimate harvested acres and abandonment.
We're accused of not using FSA data. But this particular change in acreage (the found acres) was made because of FSA numbers.
NASS also had to come back last September and change the numbers by about 2 percent. At that point, there was a perceived overshooting of acreage and production.
Unfortunately, as the 130,000 acres were announced, a price slide began. The two aren't linked, insist statisticians.
The slide occurred at the same time the World Board started adding stocks to world supplies. Our 130,000-acre revision never changed the stocks because we only produce 1.5 percent of the world's rice. The revision had little impact. When we make a 2 percent revision in U.S. rice numbers, that equates to three-hundredths of a percentage point in the world, says Klugh.
NASS points out that at the time, rice supplies had been growing worldwide for several years. Unfortunately for NASS, the huge stocks were coming into painful levels at the same time the 130,000-acre revision was made. I'm not downplaying the importance of what we do. But that 130,000-acre change did not cause the worldwide price slide, says Rundle.
The Arkansas NASS generally gets a 10 to 15 percent refusal rate on surveys. Farmers can't afford for the refusal rate to grow very much. We can't have a deterioration of response rates. The less response, the harder it is to make a good estimate. If farmers don't report to us, it makes the job much more difficult. By not answering surveys and the like, the farmers are doing a disservice to themselves. If we don't get good data, the quality of the estimates deteriorates, says Klugh.