The monthly Bureau of Labor Statistics (BLS) Employment Report is probably the most anticipated and watched economic indicator the U.S. government releases. Many industry decision makers base investment decisions on whether or not the employment numbers are better than expected or worse than expected. Yet more often than not, the BLS employment results are revised substantially from one month to the next.
For example, with the release of the October employment results, the BLS’ August and September employment results were revised higher by a whopping 154%. Two months after the BLS’ August employment release, employment went from 0, representing an economy on the verge of contraction, to 103,000 representing an economy in slow growth mode. Similarly, September’s results went from 103,000, representing an economy in slow growth mode, to 158,000 this past month representing an economy growing at a modest pace.
Based, in part, on the BLS’ dismal employment numbers in August and September, many industry decision makers expected the U.S. economy to drift into recession. Two months later, the BLS revisions point to an economy in slow growth mode, not in recession, a considerably different outcome.
Why are the BLS employment numbers frequently revised? We have identified four reasons:
1) The BLS Establishment Survey is not complete when the employment results are first released. In October, for example, the survey was 73.5% complete. Final numbers for October will not be released until January 2012. The preliminary BLS employment results then are skewed towards the first respondants.
2) Each year, from October through March the BLS applies huge seasonal adjustments. Seasonal adjustments for the October survey results were nearly 1.1 million jobs due to seasonal holiday hiring. A year ago November, for example, the seasonal adjustment was nearly 1.3 million jobs, and this past January the seasonal adjustment was 2.1 million jobs. When trying to measure relatively small employment changes from month to month, these huge seasonal adjustments tend to clobber subtle changes in employment.
3) While the BLS Establishment Survey is supposed to be a comprehensive, balanced, and unbiased survey representing all sectors of employment, across all sizes of companies, we believe it is biased towards employment activity in the government sector and large corporations. For example, while government employment represents about 17.6% of total employment, the BLS surveys 68.5% of all government employees. Government employees make up 39.9% of all respondents to the BLS survey. In our opinion, the heavy reliance on responses from government employment centers is a likely contributor to reporting errors.
4) Finally, the BLS applies a mysterious “birth/death” (BD) adjustment to the results to account for business openings and closings. This past month, the adjustment added 102,000 jobs to the non-seasonally adjusted results. According to our sources, the BD model appears to apply an upside shift to the monthly results of approximately 50,000 regardless of whether the economy is in expansion or contraction mode.
TrimTabs discovered about ten years ago that daily withholding tax deposits are a better gauge of real-time employment. Each and every day, withholding taxes from 131.5 million wage earners are deposited to U.S. Treasury. We developed a proprietary model that uses that data to compute wages and salaries and U.S. employment. Over the years, we have discovered that this is a much better measure of real-time employment than the frequently revised BLS numbers.
It is often said that employment is a lagging indicator. We believe nothing could be further from the truth. When someone loses their job, his or her consumption patterns change immediately. In our opinion, the only reason employment is a lagging indicator is because the government does not use real-time data. Trends in withholding taxes were an excellent early warning indicator to the 2008 recession.
- Madeline Schnapp
Director of Macroeconomic Research
TrimTabs Investment Research