Preliminary figures from the Quarterly Census of Employment and Wages (QCEW), which Governor Walker has previously touted as more accurate than the establishment survey , indicate a substantial downward revision in Wisconsin private employment.
Figure 1 shows the 12-month change (not growth rate) in QCEW (not seasonally adjusted) private employment and 12-month change in private employment reported in the BLS establishment survey.
Figure 1: Change in private nonfarm payroll employment as reported in establishment survey, s.a. (blue), in private employment as reported in QCEW n.s.a. (red), both in 000’s. Red square is preliminary 12-month change as reported by WI DWD. Source: BLS, DWD, and author’s calculations.
It is interesting to note that the preliminary QCEW figure is a full 11,700 less than the BLS establishment survey estimate. As the Governor’s administration has stressed  , the QCEW numbers are from a census, not a survey, and are hence much more accurate.
The QCEW figures are not seasonally adjusted, so a comparison with the establishment series in levels is a bit complicated. In order to draw out the implications, I use the 2001M01-2013M09 sample period to estimate the relationship between the two series:
nfpt = β0 + β1nt + monthly dummies + ut
Where nfp is the log employment series, and n is the log census series.
The coefficient on n is 0.96, t-stat using robust standard errors of 94. The adjusted R2 is 0.99, SER = 0.0017.
I use this equation and the 2013M12 preliminary figure to establish the 2013M12 implied revised level of the establishment series, and then iterate using the establishment series up to 2014M04. I plot this series, and the reported establishment series, in Figure 2.
Figure 2: Private nonfarm payroll employment as reported in establishment survey, (blue), estimated establishment series based on QCEW (red), both, s.a., in 000’s. Green shaded area is iteration period (see text). Source: BLS, WI DWD, and author’s calculations (see text).
The difference is 18,000 in 2014M04.
It is interesting to consider what this implies for Governor Walker’s promise to create a net new 250,000 jobs by January 2015, reiterated as recently as last November .
Figure 3: Private nonfarm payroll employment as reported in establishment survey, (blue), estimated establishment series based on QCEW (red), both, s.a., in 000’s, Wisconsin Economic Outlook forecast, interpolated from annual data (green), and Walker’s implied path. Source: BLS, WI DWD, Wisconsin Economic Outlook (March 2014), and author’s calculations (see text).
These calculations imply that, in order to attain the 250,000 new jobs goal, 22,078 jobs would have to be created each month, for the next 7 months (as opposed to 19,500 per month using the establishment series). Mean job creation has been 2,888/month, with standard deviation of 3,988, since January 2011. Even using the establishment series, the gap between the required monthly job creation rate and the average rate of job creation is 4.2 standard deviations. For 19,500 jobs to be created month after month for the next 7 months is therefore somewhat (!) unlikely.