That’s the concluding line from the release issued by the Wisconsin Department of Workforce Development’s Secretary Reginald Newson. The preceding sentence is “The job numbers are a lagging indicator for economic conditions, and we will continue to move forward.” Here are two graphs, with updated DWD data incorporating revisions, and preliminary data for November, that place those comments in context.
Figure 1: Wisconsin nonfarm payroll employment from BLS (blue), from DWD (green bold), and projections from October Wisconsin Economic Outlook, in 000’s. NBER defined recession dates shaded gray. Vertical line at 2011M01. Sources: BLS, WI DWD, Wisconsin Economic Outlook and NBER.
Figure 2: Wisconsin private nonfarm payroll employment from BLS (blue), from DWD (green bold), and projections from Wisconsin Economic Outlook, in 000’s. NBER defined recession dates shaded gray. Vertical line at 2011M01. Sources: BLS, WI DWD, October Wisconsin Economic Outlook and NBER.
Secretary Newson stressed the fact that the preliminary numbers, calculated in cooperation with BLS, had underestimated private employment.
“October was the fifth straight month and the eighth month this year in which the federal government
overestimated the preliminary job loss numbers or underestimated job gains for Wisconsin,” Secretary
Newson said. “I am particularly concerned by the disparity in the October preliminary numbers, which were
off by 7,300 for total jobs and 7,900 for private-sector jobs. These unreliable employment statistics out of
Washington misinform the public and create unnecessary anxiety for job seekers and job creators about the
shape of our state’s economy.”
On average, Wisconsin has lost 4,980 private-sector jobs each month for the past five months.
John Heywood, an economics professor at the University of Wisconsin-Milwaukee, said the state data is prone to sample error and revision. But after five months of consecutive declines, he said the decreases cannot all be attributed to statistical error.
“We all know there is random variation in the state estimate and that revisions often move that estimate. All true,” Heywood said. “It is very unlikely that the pattern simply results from sampling error. The survey design is imperfect, but typically not that imperfect.”