The U.S. economy likely shed more jobs last year than previously thought, but analysts say the undercount by the government should prove less severe than it did during depths of the recession.

The Labor Department Friday will give an initial estimate of how far off its count of employment may have been in the 12 months through March. The government admitted earlier this year that its count through March 2009 had overstated employment by 902,000 jobs.

Analysts expect a much smaller miscount this time, given the economy's growth spurt in the second half of last year.

The department blamed its 902,000 miss on faulty estimates of how many companies were created or destroyed, and it has not yet made any changes to the so-called birth-death model that produces this projection.

Once a year, it compares payroll data from its monthly surveys of employers with unemployment insurance tax reports, which give it a much more comprehensive view of actual employment. It uses these tax records to produce a "benchmark revision" to adjust for discrepancies.

"That adjustment is probably overstating the employment gains because we are in a very subdued recovery and the likelihood is that the birth-death factor is making the data look better than it otherwise would be," said Neil Dutta, an economist at the Bank of America Merrill Lynch in New York.

Tax records will probably show more businesses closed than initially estimated by the Labor Department, analysts said.

"It's not going to be that severe (as last time). A lot of it is sort of aligned with the performance in the broader economy," Dutta said, noting that the economy picked up in the second half of 2009 and entered this year strongly.

Other economists shared that view, while some said it was even possible that employment would be revised upward, citing other data, including a separate Labor Department survey of households, that had outperformed the monthly payrolls count.

They also said that while the department had not changed the birth-death model, it had incorporated new data from a period in which business start-ups were weak.

"Potentially, the model could have underpredicted for a time. With the incorporation of this new data you may see an upward revision," said Zach Pandl, U.S. economist at Nomura Securities International in New York.

"In our view, the risks are tilted toward an upward revision."

Whatever the outcome, it will probably have little implication for U.S. monetary policy, given that it is backward-looking and the economic recovery is very weak by historical standards.

But it could shed more light on the nature of the unemployment problem confronting the economy, with opinion increasingly divided on whether it is cyclical or structural.

Analysts will be looking at the sectors where job losses are concentrated. Steeper job losses than already reported in manufacturing and construction could strengthen the argument of a structural unemployment problem.

"It's difficult to argue we don't have a structural unemployment problem because roughly half of the jobs lost were in manufacturing and construction," said Dutta.

"That also means more than half of job losses were in services. It argues that there is a decent amount of cyclical unemployment out there that we can take down by stimulating our economy more aggressively."