Data from the Quarterly Census of Employment and Wages (QCEW) suggest that the trend in employment growth recently has been somewhat stronger than indicated by the current official figures on payroll employment.

• The QCEW measure, which encompasses almost all employees in the private, nonfarm sector, is used to benchmark payroll employment as of each March. However, the QCEW data are prepared with a considerable lag - data for March 2012, for example, will not be available until this coming September.
• Nevertheless, we can use data from the QCEW, which currently are available through September 2011, to make an educated guess as to the likely direction of the subsequent revision to payroll employment.
• Prior to seasonal adjustment, the average increase in the QCEW measure of private, nonfarm employment from March to September 2011 was 62 thousand per month larger than the average increase in private, nonfarm payroll employment. However, that is not the end of the story, because the two series have different seasonal properties.
• After seasonal adjustment, the differential in average monthly increases was smaller but still significant. Based on seasonal adjustment using the Census X-12 method, the average increase in the QCEW measure was 156 thousand per month, 26 thousand more than the average increase in the X-12 adjusted measure of private, nonfarm payroll employment.
• In short, these data suggest that the growth of private, nonfarm employment over the March-to-September period was somewhat stronger than indicated by the current official data on payrolls. They hint at the possibility that gains in payroll employment over the period between March 2011 and March 2012 could be revised up in early 2013.

Our readers are no doubt familiar with the regular monthly report on payroll employment from the BLS, which is derived from a sample of 400 thousand business establishments. In addition, with a considerable lag of about six months, once per quarter the BLS issues a separate and more comprehensive estimate of employment that is based on reports from virtually all employers (the QCEW).[1] Data from QCEW are not seasonally adjusted by the source agency.

After adjusting for differences in coverage, the BLS "benchmarks" the more widely followed data on payroll employment to counts from the QCEW. Benchmarking is performed on the data for each March. We expect the next benchmarking will occur in February 2013 with the release of the January 2013 data on employment. The BLS will likely provide a preview of the benchmark revision after the QCEW data for March 2012 become available in September 2012.
Presently we have data from the QCEW only through September 2011. We estimate that, prior to seasonal adjustment, private nonfarm employment as measured in the QCEW rose by slightly more than 3 million from March 2011 to September 2011. This exceeds by approximately 370 thousand the cumulative increase in private nonfarm payroll employment over the same period (also before seasonal adjustment). However, we should not translate this difference directly into an estimate of the possible revision to payroll employment later in the year, for several reasons. First, we only have 1/2 of the data that will be used in the next benchmark - we won't have the March 2012 QCEW data until this coming September - and it is possible that subsequent months could show a weaker, not stronger trend in the QCEW than in the payroll data. Second, the QCEW measure is similar to, but does not correspond exactly, to the payroll figure, even after benchmarking. (For example, in March 2011, the QCEW level was approximately 2% smaller than the level of private, nonfarm payroll employment.) Third, seasonal patterns in the two series are different, so allowance for this fact must be made prior to making inferences about the revision to payroll employment. Nevertheless, we can make an educated guess about the possible benchmarking after carefully considering the two series.

To facilitate a direct comparison, we applied the same "X-12" method for seasonal adjustment to both the QCEW and payroll measures of private, nonfarm employment. After adjustment, the QCEW measure rose by an average of 156 thousand per month from March to September 2011, somewhat larger than the average increase of 130 thousand in the X-12-adjusted measure of private, nonfarm payroll employment. The difference of 26 thousand per month is noticeable, but smaller than based on the raw, unadjusted data.[2]

Subject to the caveats enumerated above, the QCEW indicates a somewhat stronger trend to employment growth recently than in the currently published data on payroll employment. As such, it hints at the possibility that payroll employment through March 2012 could be revised up early next year.[3],[4]

[1] Excluded from the QCEW are self-employed workers, employees of certain nonprofit organizations, and a handful of other categories. A full description is available at http://www.bls.gov/cew/cewfaq.htm#Q14.
[2] We chose to use the same X-12 procedure for both series to facilitate a direct comparison that is more informative. A less informative alternative would have been to compare the published, seasonally adjusted data on payroll employment to the X-12-adjusted data from the QCEW. Because the published estimate on (seasonally adjusted) payroll employment is based on a bottom-up aggregation of seasonally adjusted data for many categories, and because it is not possible to replicate the full seasonal adjustment by BLS for each component, it is more appropriate to perform the comparison on series adjusted in the same fashion with X-12.
[3] Each benchmark revision is wedged back to the preceding March. As a hypothetical example, if the QCEW data (after appropriate adjustment) suggest a revision of, say, +120 thousand to the March 2012 level of payroll employment, this would be reflected in upward revisions to the employment changes each month from April 2011 to March 2012 of approximately 1/12 of this amount, or about 10 thousand per month. The effect is to leave unaffected the level of employment as of the previous benchmark, in March 2011, and in all months prior to March 2011.
[4] The same exercise performed over the past five years has been reasonably informative about subsequent revisions to payroll employment. In all five years the sign of the suggested revision correctly anticipated the actual revision to the 12-month change in payroll employment, and in three of those years, the size of the suggested revision was close to the actual revision.

This is from a commentary that was published on March 30, 2012.

Contact Macroeconomic Advisers
TweetFollow @macroadvisers