NEW YORK (CNN/Money) - Though U.S. jobs data have been ugly lately, some economists think the government's annual benchmark revisions, due in early February, will pretty the numbers up a bit.

But others say the effect could be akin to putting lipstick on a pig -- the job market's just not great, and the revisions won't disguise that.

On Feb. 6, the Labor Department is due to report January's unemployment rate and non-farm payroll growth. When it does, it will revise the payroll numbers for the past several months, possibly going as far back as January 1999.

Revisionist History Annual non-farm payroll revisions Year  Percentage change  1990  -0.2  1991  -0.6  1992  -0.1  1993  +0.2  1994  +0.7  1995  +0.5  1996  See note  1997  +0.4  1998  See note  1999  +0.2  2000  +0.4  2001  -0.1  2002  -0.2   Note: change was less than 0.05 percent  Source:  Bureau of Labor Statistics

Some economists believe the numbers for recent months -- including December, which showed a stunningly weak gain of just 1,000 new payroll jobs -- will be revised upward.

The theory is that the business survey generating the payroll numbers fails to count many start-up businesses, which grow in number in the early stages of a recovery. Revisions to the payroll data could belatedly capture some of the workers at such firms, pushing the total number higher.

"We're at the stage of the business cycle where statisticians can't keep up with rapidly changing developments, especially new businesses," said David Resler, chief economist at Nomura Securities. "That's where the greatest potential for an upward revision is."

For example, the payroll revisions were negative in 1990 through 1992, when the U.S. economy was in a recession and jobless recovery. In 1993 and 1994, however, when job growth returned, the revisions were upward, and in a fairly big way. 1993's payrolls were revised up by 0.2 percent, or more than 200,000 jobs. 1994's payrolls were revised up by 0.7 percent, or nearly 800,000 jobs.

The hope is that such revisions will happen again this time around, and that they will help close the head-scratching gap between the establishment survey and the household survey, which generates the unemployment rate.

The household survey showed a gain in total employment of nearly 2 million in 2003, while the establishment survey showed a loss of 60,000 jobs during the same period. The household survey says about 136.2 million people have non-farm jobs, while the establishment survey counts just 130.1 million non-farm jobs.

Related stories Jobless claims slip Jobs seen rising, barely IBM giveth, taketh away Economy gets B+ Payrolls disappoint again

What's more, the payroll numbers are seemingly at odds with a steady recent decline in the number of new weekly claims for unemployment benefits. Since the week of Sept. 13, 2003, the four-week moving average of claims has shrunk by a dramatic 66,500 -- consistent, many economists believe, with much stronger jobs growth.

"That kind of decline is a perfect fit with what we've seen in the household survey's measure of employment," said Steven Wieting, senior economist at Citigroup, saying he expected upward revisions to the payroll numbers, "if not in February, then in future benchmark revisions."

But many economists are skeptical the revisions will be significant enough to change anybody's mind about the labor market's health. For one thing, it's almost certain they will not be enough to close the six-million-job gap between the household and payroll surveys. 1994's revision was the biggest on record, and it added just 800,000 jobs.

And a big upward revision was expected the last time the Labor Department adjusted its benchmarks, in May of 2003. Instead, the numbers were revised down.

"I'm not sure we will see a big revision in February; I won't be surprised if we do not," said Sung Won Sohn, adding that the household survey's method of counting self-employed people and farm workers might add to the large discrepancy between it and the payroll survey.

What's more, some economists believe that structural changes in the economy are going to keep the payroll numbers weaker than usual. Jobless claims are down, they say, because businesses have stopped firing workers, while household employment is up because more people are calling themselves self-employed.

But the payroll numbers might not recover, they say, because businesses have learned to get more production out of fewer workers and are shipping some of their jobs overseas, where labor is cheaper.

YOUR E-MAIL ALERTS Department of Labor (DOL) Economic Cycle Research Institute (ECRI) Unemployment or Create your own Manage alerts | What is this?

"There's been a very real structural shift, primarily in manufacturing employment, as a result of outsourcing and earlier capital investment that has increased productivity. They're making employment act differently in this upturn," said Lakshman Achuthan, managing director of the Economic Cycle Research Institute (ECRI), another private research firm in New York.

"Benchmark revisions will make for less of a divergence between the household and payroll surveys, but they're not going to remove it," he added.

Still, Achuthan and other economists believe the payroll numbers will eventually look better. Most forecasters call for relatively strong economic growth in 2004 and say businesses will have no choice but to add more workers than they did in 2003.

The drop in jobless claims and an upturn in business hiring plans, according to several surveys, are among the leading indicators economists say prove job growth is on the way, and the pesky payroll report is just slow to catch up.

"Some of the other indicators are suggesting more strength than we see in payrolls," said Joshua Feinman, chief economist with Deutsche Bank Asset Management. "Whether that will be rectified with revisions, or simply with stronger numbers going forward, it's hard to say -- but I would anticipate that, at some point the payroll numbers, will be stronger."