The Government{s} are very keen on amassing statistics -- they collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But what you must never forget is that every one of those figures comes ... from the village watchman, who just puts down what he damn pleases.

-- British economist Josiah Stamp, 1929 The figures sound as solid as a brick wall, as precise as a pinpoint.

The area's unemployment rate is 3.3 percent. Residents of the region work at 2,240,600 jobs. The federal government pays $11.2 billion to local companies that contract with it.

These and other statistics have become the mileposts of the local economy. They are assuming an even more crucial role these days as experts use them to debate the extent of the current slowdown.

But in fact, the numbers are about as solid as Jell-O, as precise as a paint stroke.

The figures that portray our economic condition are subject to error, long lag times, revision, seasonal factors, inflation and small sample sizes. Not to mention misinterpretation.

Few contend that the indicators could somehow be more accurate, given the vagaries of the data-collection business. Many of the factors that make the numbers look shakier to the layperson actually make them more precise. But all agree they should be consumed with some caution.

"I hate to say we shouldn't pay attention to information, but we shouldn't over-invest in a particular piece of information," said Stephen S. Fuller, chairman of the department of urban planning and real estate development at George Washington University.

Fuller should know. He learned first-hand the advisability of statistical skepticism earlier this month, when the latest installment of the figures on which he had based much of his claim to fame turned out to be extremely wrong.

For six years, Fuller had been compiling, analyzing and publishing statistics on federal contracting in the Washington metropolitan area for the Greater Washington Research Center. As the local economy boomed in the 1980s, federal contract spending here soared as well. It was clearly one of the underpinnings of the area's financial health and the numbers came to be among those closely watched by experts on the local economy.

But as Congress began to cut back on defense spending, the figures turned downward. Federal contract spending here fell almost $800 million in 1988, according to figures Fuller obtained from the Federal Procurement Data Center.

Thus, it was no surprise when the agency's figures showed that federal contracting had fallen again in 1989, by nearly $1 billion. The research center last April produced a 15-page report detailing where the cuts had occurred and Fuller held a press conference to explain the decline.

"Obviously, this reduction has potential negative impacts not only for the firms involved, but also for the area's overall economy," the report said, going on to note that so far the local economy still was showing signs of strength.

Unfortunately, the numbers supplied by the federal government were just plain wrong. The Federal Procurement Data Center somehow had given inaccurate figures to the Census Bureau, which had published them in a report. Nationwide, it turned out later, the center's data in the Census report were $12 billion off in the defense area alone. Fuller, meanwhile, had taken all his information from that report.

When the complete computer tape was made available to Fuller in September, it was clear the numbers didn't add up. After a frantic two weeks of investigation, the research center announced the error. Federal contracting actually had gone up by more than $500 million.

Research center president Atlee E. Shidler denounced the data center and the census bureau, both of which had known about the new figures since the spring, for allowing the area's residents to believe for four months that their economy was in worse shape than it actually was. Government officials responded that they had no mandate to notify all the subscribers to the data that it was in error.

Even today, the dispute is unresolved. Fuller's new figures do not agree with the new figures at the census bureau and it still is not entirely clear how the Federal Procurement Data Center error occurred. Explanations ranged from a new computer system to data-entry errors to late reporting of some contracts.

Embarrassed officials at the research center point out that they have been compiling economic statistics about the region for 30 years. This is the first time a mistake of this magnitude had occurred -- and it was someone else's fault.

"There was nothing wrong with anyone's analysis. What was wrong was the data," Shidler said.

Everyone agrees the problem with the contracting figures was a bit of a fluke. But experts also agree that even the official, regularly released government indicators about the local economy should come with warning labels.

By far the most important of them are the metropolitan jobs figures, which -- based on a survey of households -- report the number of people in the labor force, the number of people working and the number of people looking for work. The ratio of the unemployed to the labor force is the unemployment rate.

A separate survey of employers yields estimates, by industry, of how many jobs are on area payrolls. This information, growth in payroll employment, is perhaps the single best proxy for the performance of the local economy. Both sets of survey results are released by the D.C. Department of Employment Services, with input from the Maryland and Virginia employment agencies.

The data are scrutinized closely throughout the region after they are released by the D.C. Department of Employment Service each month and are reported in The Washington Post and elsewhere as if they were all-encompassing and final.

But Richard Groner, the department's labor market chief who is the maven behind the data, said: "Nobody should consider these numbers as absolute."

Consider the following:

The household-survey figures are based on a relatively small sample -- fewer than 1,000 in the District -- and are run through an econometric model to smooth out sampling and other vagaries before they are released.

The employer survey, in the District and the two states, is considerably broader and is not adjusted by a computer, as is the household survey. But its margin of error still is about 1 percent. With an area work force of 2.3 million, annualized changes of fewer than 23,000 jobs are suspect. Last month, the figures showed that 1,300 jobs had been added to area payrolls in the last 12 months.

Certain types of employment are not counted in one survey or the other, among them the self-employed and undocumented foreign workers.

The figures, unlike the national employment statistics, are not adjusted for seasonal variations because the sample size is too small. When economic data are subject to predictable seasonal swings -- such as the surge of young job seekers who enter the labor market every year when the school year ends -- economists generally prefer to adjust the numbers to eliminate such wide variations, thus making underlying changes easier to spot. The lack of adjustment for the local figures makes month-to-month comparisons difficult.

The monthly unemployment numbers are preliminary and often are revised a tenth of a point or so when the next set of figures are reported the following month.

Every year, the D.C. Department of Employment Service and the state agencies revise their sample based on federally reported changes in the population and other new information. When this is done each January, the monthly figures for the preceding year are usually revised.

The changes are not always large, but they can have an impact. In Virginia, for instance, officials first thought that payroll employment had grown 3.6 percent in 1989. When the annual "benchmark" revision was done, it turned out the increase was only 3.3 percent. It was not a major difference, but it was one of the reasons why Virginia state tax revenue forecasters underestimated the revenue they would collect that year, contributing to that state's budget crisis.

"There is a good reason for calling these figures estimates, even a year after the fact," laughed Roy Pearson, director of the bureau of business research at the School of Business Administration at William and Mary.

Other figures also are more ephemeral than the nonexpert might think. Retail sales figures for the metro area are not adjusted for inflation, so that it is impossible to separate price increases from increases in sales volume.

Figures on statewide economic activity are only available from the Commerce Department three years after the year in question. And expanding definitions of the Standard Metropolitan Statistical Area, the geographic base for the region, make historical comparisons difficult at times.

The situation gets even more complicated when the government figures are used as ingredients to create simulations or forecasts of the local economy.

The accounting firm Grant Thornton, for instance, every quarter publishes an economic index that combined seven economic indicators to summarizes the area's economic activity. It is widely used by businesses and published in the press.

Experts at Grant Thornton wait until all the figures are in for the three months of the quarter before completing their compilations. But results for the third month of retail sales figures, among others, are only preliminary. When they are revised, the index sometimes has to be revised too.

John Koegel, part of the Grant Thornton index team, said if he waited any longer the index would lose its timeliness. The one for the quarter ending June 30, for instance, was released Sept. 26. It showed that the index for this area fell 0.72 points, to a value of 110.8. But, Koegel cautioned, the margin of error is half a point -- so the index could have fallen 1.22 points or .22 point.

"I would caution anyone not to rely on our index or any measure alone. It should just be part of the picture," Koegel said.

Nonexperts can go wrong even if they fully understand how to read the statistics -- but don't know how to interpret them. Because this area has been so prosperous for so long, the economic experience of many local residents and business owners is confined to good times and they may not know what to make of the current slowdown.

Like past and future downturns, what's happening now is a normal phase of the business cycle and will be followed by faster growth, said Jay Langford of the Metropolitan Washington Council of Governments.

Home buyers, consumers, business persons and reporters would do well to remember, Langford said, that a slowing of frenzied growth is a long way from depression.

"There is a tendency to see short term trends as longer-term trends," Langford said. "It suggests a conclusion that the roller coaster will plunge straight to hell, that pretty soon we'll be down to the molten lava. Growth has started to slow down -- from the fastest period in history."

Federal Contract Spending in the Washington Metropolitan Area

----------- in Billions of Dollars; Fiscal Years ------------

................................Incorrect........latest

.........................1988..... 1989........... 1989

Washington...............$2.7......$2.3............$3.0

Maryland................. 3.5.......3.2............ 3.9

Virginia................. 4.4.......4.1............ 4.3

TOTAL.................. $10.6..... $9.6.......... $11.2

SOURCE: Greater Washington Research Center