DALLAS -- Before Federal Reserve governors nudge interest rates or the Council of Economic Advisers suggests government spending be cut, they look to people such as Ginni Ecklund for direction.
Not because Ecklund is an economist. She's not.
But she is one in a flock of 1,600 government interviewers around the nation who collect data each month for the unemployment rate -- a bellwether economic indicator -- by asking Americans whether they are working.
The answers to her questions can cause interest rates to move, the market to gyrate and politicians to stump. The results also show Washington where to send job-training funds.
Yet, government officials admit, the monthly ritual -- which also produces jobless rates for every state plus 270 metropolitan areas -- is prone to vagaries that can cause wild fluctuations in the unemployment rate. That, in turn, can affect funding for economically distressed areas and alter the count of employed people, which analysts watch most closely to judge overall economic strength.
"It's pretty good. We do the best we can," says Thomas Plewes, associate commissioner for the U.S. Bureau of Labor Statistics, referring to the monthly labor report that his agency prepares. "But we don't do a perfect job."
The government has reviewed procedures and implemented revisions since the first national employment report was published in 1948 by the Census Bureau.
But it struggles to keep up with today's more complex labor market and larger economy amid the higher expectations for the monthly report. The budget for the survey has been stagnant for several years, a plan to modernize the report has yet to be funded and the government is losing the battle to find and keep high-quality interviewers. The restrictions frustrate labor and census officials, who are responsible for the data, and private analysts, who depend on the data.
"It's skew or be skewed," says Robert Brusca, chief economist at Nikko Securities Co. in New York, of the government's statistics. "I'm not going to harangue the government for the labor report. But there are problems with the report, and I think they could spend a little more money to do it better."
The joint task of the Labor Statistics and Census bureaus begins the first Sunday before the 19th of each month with a question-and-answer routine called the Current Population Survey. It is conducted by census workers such as Ecklund. They are expected to contact about 60,000 households in person or by telephone.
A computer at the Census Bureau's headquarters in Suitland, Md., randomly selects the households. They may be single-family homes, apartment units or mobile homes, English- and non-English-speaking. Each is interviewed eight times during a 16-month period.
Ecklund records each interview in the 12-page Current Population Survey booklet, which contains dozens upon dozens of questions -- questions that contain potentially as many pitfalls.
For example. "The first question you ask them is what they did most of last week?" says Ecklund. "Most people say, 'working.' Very seldom do they say, 'nothing.'"
People are asked about the week preceding the 19th. If they worked then, they are recorded as employed -- even if they didn't work the rest of the month. Some critics say such questioning can inflate the employment count, but the government defends it.
"That week was chosen some time ago for a number of reasons," explains Plewes. "It contains the least holidays and is tucked right into the middle of the month."
The survey work must be completed in 10 days. Ecklund normally is assigned about 50 households where she is required to interview every occupant 16 years old or older.
If the residents move before a follow-up survey, Ecklund still must interview whoever answers at that address. Regardless of whether the former residents are working, it is the employment status of the new occupants that is recorded.
Any households Ecklund cannot reach before the survey period expires are recorded as missing. Each address, however, represents the presumed behavior of 1,500 households, so interviewers are pressed to complete every case.
"It's kind of a compromise between the money you have and the statistical reliability you seek," says Plewes. "If they miss a household or have an error, it can effect the statistics."
But finishing a caseload isn't easy, says Ecklund, because of obstacles such as language, distrust and the danger of darkness. Many of the interviews are conducted in the late evening when the likelihood of finding people at home is greatest.
"At one of the [apartment] complexes I work in, I know there's a lot of drug-dealing going on," says Ecklund. "But I've never had any problems."
At a predominantly Hispanic complex, she says she often pays bilingual residents to interpret for her. Sometimes, she says, the questions and answers appear to get mangled in the process.