Over the past several months, the Arkansas employment reports have been providing mixed signals. On the one hand, the unemployment rate has declined dramatically as household employment has risen. Meanwhile, the payroll employment statistics have shown rather sluggish growth rates. The chart below compares these two measures of employment. Although they are compiled from different surveys and measure slightly different concept of employment, the overall patterns of the two series are quite similar.* The difference between growth rates over the past several months is apparent, however. From May 2011 through May 2012, household employment has increased by 2.6% while payroll employment has gained only 0.6%.
A third data source helps to reconcile these two divergent series. The Quarterly Census of Employment and Wages (QCEW) represents a more comprehensive and accurate version of the payroll employment statistics. In fact, the statistics from the QCEW represent the source-data for annual revisions of the payroll employment figures. These revisions are often substantial (see here and here, for example).
Although we are still a long way from the annual revisions, the latest data from the QCEW already suggest that there will be large revisions to payroll employment statistics for the second half of 2011. Using some simple statistical techniques to estimate the magnitude of the eventual data revisions, it appears that cumulative revisions to the end of 2011 will result in an employment level more than 11,000 higher than currently-published statistics. The upward revision for October alone is expected to be more than 18,000. As shown in figure below, the anticipated revision will raise the growth profile of payroll employment substantially. With these estimated revisions included, payroll employment growth from May 2010 to May 2011 amounts to 1.5%, narrowing the observed gap between houshold and payroll employment growth.
Because the actual benchmark revision will use QCEW data through the first quarter of 2012, the magnitude of the ultimate payroll data revision may end up being even larger. What we can say for sure at this point is this: The relatively weak growth of payroll employment observed over the past year is largely the result of preliminary data that under-measure the true extent of Arkansas employment growth.
# # #
*The household employment statistics are based on survey measures and local economic models, and are compiled under the Local Area Unemployment Statistics (LAUS) program. The payroll data are based on a survey of firms, and are published as part of the Current Employment Statistics (CES) program.