Q+A: Jobless rate drops even as job growth disappoints

WASHINGTON (Reuters) - The unemployment rate dropped to a 1-1/2 year low of 9.4 percent last month from 9.8 percent in November. It was the largest monthly fall since April 1998.

Does that mean the U.S. economy is off to the races?

Far from it, analysts said.

Employers only added 103,000 new workers to their payrolls in December, a relatively tepid increase that suggests the economy is just treading water. Economists had expected a gain of 175,000.

Below are some questions and answers on why the unemployment rate and jobs count could send such seemingly differing signals, and how economists called it so wrong.

Why did the jobless rate fall so far when job growth was modest?

The employment report is compiled from two different surveys. Economists tend to look at the bigger and less-volatile survey of non-farm employers. That’s the survey that yielded the 103,000 increase in employment. The jobless rate is derived from a separate survey of households.

For its establishment survey, the Labor Department polls about 140,000 businesses and government agencies. In contrast, it polls only about 60,000 households for its household survey -- one reason economists rely more heavily on the poll of employers.

The household survey showed employment increased 297,000. At the same time, the size of the labor force declined 260,000. Both of those figures can fluctuate sharply from month to month; in December they both moved in a direction that combined to pull the unemployment rate down sharply.

Why were economists so far off in their forecasts?

The median forecast from a Reuters poll of economists conducted last week was for a gain in non-farm payrolls of 140,000. But many economists raised their projections after a report on Wednesday from payrolls processing company ADP Employer Services. That report said private employers added 297,000 new workers in December.

A quick repoll of economists came up with a new consensus for Friday’s government report: 175,000.

But the ADP data has a poor history as a predictor of the government employment count. In addition, there may have been a larger-than-usual discrepancy in December because of how ADP monitors employment and adjusts its data for seasonal fluctuations.

Regardless of their employment status, workers of ADP clients remain on payroll records until December when they are removed. Given the declining trend in layoffs, fewer workers probably fell off the records than the seasonal factors had anticipated, leading to a surge in the seasonally adjusted figure reported by the ADP.

Does the government data properly reflect the state of the labor market? Why are there often big revisions to prior months?

The government’s survey of employers is large, but even government statisticians agree they can’t capture all of the ongoing shifts in a 153-million strong labor force.

Data for the prior two months are subject to revision with each new month of data. These revisions incorporate additional responses from the employers that had been surveyed.

The department then has to extrapolate from its findings how many jobs were actually created or lost. It also has to rely on a statistical model to estimate how many new businesses are created or how many old ones had died in any given month.

Since the job count is simply based on a statistical sample, it is only a best guess at reality. The Labor Department says the margin of error in its payroll poll is around 100,000. That is, there is a 90 percent chance the true gain in employment fell between 3,000 and 203,000 in December.

That’s actually a pretty high confidence level given the size of the labor force. For example, the 72,000 difference between the Wall Street forecast and the December job creation figure represents a miss of just 0.05 percent of the total labor force.