The job market took another hit on today as the Bureau of Labor Statistics reported that the economy shed payrolls for the ninth consecutive month. And even though businesses have been cutting back since December—and the unemployment rate has risen 1.7 percentage points since last March to 6.1 percent, the highest in 4.5 years—a group of vocal analysts, bloggers, and journalists argue that the employment picture is even worse.

The reason, these critics say, is that the way government statisticians calculate how many jobs are created by new businesses and destroyed by dying firms is flawed and is no more than an educated guess.

The B.L.S. has two ways of calculating how many jobs are gained or lost each month. The method preferred by most economists is a survey of 150,000 businesses across the country—the preliminary count for August shows that 84,000 jobs were lost. The problem here is that the bureau does not have a way of detecting when an entrepreneur has started a business (which creates jobs), or when an old firm has closed (taking jobs with it).

As a way around this, the B.L.S. introduced what it calls a birth-death model in 1999. It uses historical data and fancy equations to estimate the number of jobs added and lost by new and dying businesses. But the model typically spits out a positive number, leading critics to charge that it doesn't know how to handle a downturn. (For August, the birth-death model added 125,000 jobs.)

This may well be true, but other data point to benefits from using the model.

Once a year, the B.L.S. releases a benchmark revision which is a much more accurate tabulation of the number of jobs in the country. This revision draws on records of the number of employees covered by unemployment insurance.

The B.L.S. says these records cover an impressive 98 percent of all nonfarm jobs. A revision based on this data is much more likely to reflect the real employment situation.

By 2003, the B.L.S. had increased the coverage of its birth-death model to all of the 10 major sectors (excluding government) that it tracks. In the five years before 2003, the average revision for the monthly employment figures from when they were first announced to the final version was 440,000. Between 2003 and 2007, that number fell to 334,880, according to data from the Philadelphia Federal Reserve. This suggests that, if anything, the addition of the birth-death model has actually improved the accuracy of the monthly numbers.

Now, there is another metric showing bigger job losses than the payroll survey, and that's the second of the two B.L.S. methods mentioned above. The Current Population Survey, a monthly survey of 50,000 households carried out by the Census Bureau for the B.L.S., indicates that economy has lost 1.17 million jobs during the current downturn compared with the 605,000 the payroll survey shows.

But you won't find too many of these critics pointing to the household survey to support their argument. The reason?

Back when the economy was growing, the household survey detected much bigger job gains than the payroll survey. Since most critics can be described as perma-bears, they spent a good bit of time arguing that the household survey was bunk. Perhaps they think it would be unseemly to flip-flop to now say it's good.

But all of this is not to say that government statistics don't need improvement; they surely do. And the emphasis put on the jobs numbers by the markets and the media aren't commensurate with the forward-looking information the data provides. Job growth is a lagging indicator, and the wide confidence interval around the reported job figures typically mean it's hard to know if the economy gained or added jobs in a given month. Even our current Federal Reserve chairman, Ben S. Bernanke, has thought so.

But his predecessor, former chairman Alan Greenspan, perhaps said it best in 1997, referring to another important data point, the gross domestic product:

"I think we are getting to the point...when we will have to move unless very clear evidence emerges that the expansion is easing significantly. I don't mean evidence that G.D.P. growth is moderating to 1¾ percent or whatever the forecast is. The G.D.P. is a nice number and it does have some relationship to the real world. I'm not terribly certain what it is, though everyone tells me it does."