It’s often difficult to sort through monthly press reports and associated partisan pronunciations about job numbers. Making it even more difficult this year is one source for job numbers says the state lost 33,900 jobs in 2011, while another says it gained 23,300. How do we reconcile the two reports?
The monthly numbers that get most of the press are from a survey of employers (approximately 5% or so). Because the reported numbers are extrapolated from the sample of employers, there are errors associated with the numbers. As more information becomes available, the numbers get revised (see here).
A second set of job numbers gets much less attention, primarily because there is a 6-7 month delay in reporting. All employers covered under the unemployment insurance (UI) system (about 90-95% of all jobs) report actual job numbers each quarter. Because these job numbers are actual counts and not extrapolations from surveys, they are the most accurate. In fact, each spring, the Bureau of Labor Statistics uses these numbers to adjust (benchmark) the survey numbers.
The chart below shows year-over-year changes in the two surveys from January 2009 through June 2011. The two move in sync because the numbers from the survey have been benchmarked to the UI numbers through June 2011 (the UI numbers for July-Dec were not available earlier this year when the benchmarking was completed.

We now have the numbers from UI and they diverge from the survey after June, as the chart below shows. As of December 2011, the actual job count from UI shows the state gaining 23,300 jobs; the survey shows a loss of 33,900.

Which job numbers are right? The figures from UI are the most accurate as they represent actual job counts from nearly all employers. And, early next year, the survey numbers will be adjusted to reflect the pattern found in the UI data.
One final question: Is this unique to Wisconsin? No. While we have UI data for Wisconsin through December 2011, we have it for other states through September 2011. We will have fourth quarter 2011 figures nationally at the end of June. However, when you look at July-September figures by state, you find Wisconsin is not alone in the survey numbers diverging from the UI numbers. The charts below show the divergence for four states, although you can identify at least 10 states with similar patterns. Unfortunately, we do not yet know why this is happening.




