by Calculated Risk on 1/21/2011 11:17:00 AM

Twice a month Gallup polls approximately 30,000 adults on unemployment. This survey is Not Seasonally Adjusted (NSA).

Before I get to the data, this is a reminder that the employment series really needs a seasonal adjustment. Every year, even in good employment years, the U.S. economy loses 2.5 to 3.0 million payroll jobs (net) in January. This is because of various seasonal factors, like seasonal retail layoffs, and the headline BLS report is seasonally adjusted (SA).

A good example is the strong employment period from 1995 through 2000. Over those 6 years, U.S. payroll jobs declined 2.6 million on average every January (NSA), and the BLS reported a gain of just under 200 thousand jobs on average (SA). Quite a seasonal factor!

The same pattern is evident in the unemployment rate; the NSA rate is always sharply higher than the SA rate in January. So keep that in mind when looking at the following from Gallup: U.S. Unemployment Steady at 9.6% in Mid-January

Unemployment, as measured by Gallup without seasonal adjustment, remained at 9.6% in mid-January, the same as at the end of December. This marks a one-percentage-point improvement from 10.6% in mid-January 2010.

This suggests a sharp drop in the seasonally adjusted unemployment rate in January (although the timing and methods are different between the two surveys).

In January 2010, the BLS reported the unemployment rate as 10.6% (NSA) and the headline was 9.7% (SA). Gallup also reported the NSA rate at 10.6% for the mid-January 2010.

If the Gallup poll is even close, then the headline seasonally adjusted BLS number might be well below the 9.4% reported last month. Of course if the unemployment rate declines because the participation rate falls further - that isn't really good news.

Use this with caution. There isn't much of a track record of using the Gallup survey to predict the BLS report. But it might be a hint of good news.