Employment numbers and BS–or rather BLS

This morning markets cheered an upside surprise in April's US non-farm payroll report. Of 82 economists surveyed by Bloomberg for April 2008, the median expectation was for a job loss of -75,000 (Dow Jones had -85,000) for the month. Estimates ranged from -150,000 to -18,000, with no positive numbers excepted.
Then the big surprise hit at 8:30am, the economy had only lost 20,000 jobs in April. Great news??
Well unfortunately no. Once we understand how the job loss numbers are calculated, it is hard to jump for joy. Here is why.
The employment numbers are estimated each month by the US Dept of Labor, Bureau of Labor Statistics (BLS).
The Current Employment Statistics (CES) program, also known as the payroll survey, produces nonfarm employment, hours, and earnings series each month based on a monthly sample of nearly 400,000 business establishments nationwide. The CES is designed as a high volume, quick turnaround survey, collecting a few readily available data items from each sampled firm's payroll.
The current design has been fully implemented since 2003 and follows modern design principles for an establishment survey. Firms from all sizes, industries, and States are included in the sample. For more information on the sample design see www.bls.gov/web/cestn2.htm.
BLS estimates the level of payroll employment with the CES survey. They do this to try and account for new businesses “births” that may be creating new jobs not yet apparent in the sample survey and for businesses that may have gone out of business and did not respond to the sample survey. Unemployment Insurance (UI) records are the most accurate record of these companies but UI records only come available for updating the CES sample, 7-9 months after the report.
So the BLS uses an estimate of jobs created in various sectors to round out their report. The estimates significantly skew the employment numbers in a few important ways.
Firstly, the April number of -20,000 jobs assumes estimated net new jobs of 267K. So without this assumed +267K, the employment numbers would have actually been -287K for April. If only half the estimated net jobs were created then the payroll number would have still been triple digit negative. When we look at the sectors in which net new jobs were assumed to have been added we have good reason to pause. Thousands of jobs were assumed to have come from sectors like construction and finance where we know that jobs have actually been cut not added. In addition we know that the economy actually needs to add 150K new jobs a month in the US just to keep up with population growth. So even a reading of -20,000 is about 170,000 jobs short of what is needed to maintain the present level of employment.
Adding further doubt to the survey numbers is the fact that April had a low participation level with just 67% of businesses actually responding to the survey:
“As the economy slowed into the 2001 recession, companies did not fill out the BLS survey nearly as much as they had in the past, as the economy improved and expanded more companies started filling out surveys. And again as the economy cooled off from 2006 to 2008, the survey responses have dipped.
So why do I keep harping on this. Several Reasons.
1) It is an excellent leading indicator, which is being ignored by the BLS.
2) Lower response rates means the BLS fudge factor is larger. Last April the BLS had 69.3% of all firms responding, so the BLS had to estimate the employment at the remaining 30.7%. This year the BLS only has 67.3% of all companies’ data and has to estimate 32.7%
3) The combination of 1 and 2. Even worse, because the BLS is ignoring the first factor they are using even a larger plug rate for net new firms. In April of 2007 they estimated 262,000 new jobs. And in April of 2008 they are estimating 267,000 new jobs. (That's about 2 xs as much as in March). And because of factor #2, the plug rate becomes more important as the overall base is smaller.
So what if the BLS had used the March plug rate? Well today's -20,000 jobs would be -140,000. Thanks to an extra +120,000 “phantom” jobs, this April doesn't look too bad.”
See: CEO Economic Update.
Bottom line: the US economy almost certainly lost more than 100,000 jobs in April, despite the headline reported. And we will not see the truth of this for several months in retrospect. When we do, it is likely that April job losses will be more in line with an economy in early recession than with happy days.

This entry was posted in Main Page. Bookmark the permalink.

Leave a Reply

Your email address will not be published.