Global Economic
Trend Analysis

Recent Posts

Friday, March 09, 2007 2:13 PM

February Jobs / Employment Data

Mish Moved to MishTalk.Com Click to Visit.

On March 9th the Bureau of Labor Statistics (BLS) released the February Employment Report. Following is some data from that report.

Unemployment Rate

Household Data

The unemployment rate comes from the "Household Survey", a phone survey of individual households rather than actual numbers from weekly jobs claims and payroll data. The sample size is small and the methodology has been questioned by many. But for now it is what it is, and subject in theory to manipulation. So how did the unemployment rate manage to tick down in February? Look no further than the government reporting that 374,000 workers dropped out of the labor force (and are thus no longer considered unemployed). Even with that staggering drop, the unemployment rate did not make a lower low. Heading into a recession it is clear that the unemployment rate has bottomed this cycle.

One thing that is important to know about the phone survey is that anyone who says they have a job is counted no matter how few hours they work at it, and regardless of pay. Thus real estate agents who have not made a home sale for 4 months are still considered employed, whether or not they have commission checks coming in or not. The same goes for those selling trinkets on EBAY as well as those who consider themselves employed on the basis of some Multi Level Marketing venture they are in. Finally, the phone survey does not have a list of cell phone numbers so those without a "real line" are excluded from the survey.

Participation Rate

The following chart is thanks to VisionsFromSpace.
Data for the chart came from the BLS.

The above chart shows that a three month average growth in jobs is barely keeping up with population growth. Declining real wages for many of those with jobs makes the situation even worse.

Establishment Data


  • 62,000 construction jobs were lost
  • 14,000 manufacturing jobs were lost
  • 39,000 government jobs were added
Unlike the Household Survey, the Establishment Data is based on a sampling of real businesses. Unfortunately the data is seasonally adjusted as well as revised according to something called the Birth/Death model.

The above chart shows that 97,000 non-farm jobs were added in February. Government jobs provided a stunning 40% of those jobs.

Birth/Death Model

This was a very weak jobs report. But it is even worse when one looks at Birth/Death Model assumptions.

For those unfamiliar with the birth/death model, monthly jobs adjustments are made by the BLS based on economic assumptions about the birth of death of businesses (not individuals). Those assumptions are made according to estimates of where the BLS thinks we are in the economic cycle. The BLS will not disclose what their methodology is or even on what it is based. Essentially it is a "black box"with the BLS essentially saying "trust us, we know what we are doing".

The BLS has admitted however, that their model will be wrong at economic turning points. One more factoid on these adjustments: The months of January and July are typically revision months. Birth/death adjustments tend to be positive (adding jobs) February through June and August through December. Then as you can see in the above table, there are enormous negative revisions in January and July.

The latest birth/death numbers appear to be from Mars, Pluto, or France. With housing falling like a rock and homebuilders demanding concessions from contractors, the BLS is assuming that 11,000 new jobs were added in construction and 3,000 new jobs in manufacturing. With subprime lenders blowing up everywhere (going out of business) the BLS is assuming 11,000 new jobs were added in financial activities. The total number of jobs added by such assumptions for February was 118,000 jobs.

No doubt you will see some who will subtract 118,000 jobs from 97,000 jobs and conclude that 21,000 jobs were lost as opposed to gaining 97,000 jobs. Unfortunately such math is inaccurate because the establishment numbers are seasonally adjusted and the birth/death assumptions are not.

Here is the pertinent snip from the BLS on Birth/Death Methodology.
  • The net birth/death model component figures are unique to each month and exhibit a seasonal pattern that can result in negative adjustments in some months. These models do not attempt to correct for any other potential error sources in the CES estimates such as sampling error or design limitations.
  • Note that the the net birth/death figures are not seasonally adjusted, and are applied to not seasonally adjusted monthly employment links to determine the final estimate.
  • The most significant potential drawback to this or any model-based approach is that time series modeling assumes a predictable continuation of historical patterns and relationships and therefore is likely to have some difficulty producing reliable estimates at economic turning points or during periods when there are sudden changes in trend.
The important point in this mess is that both the job data and employment data are likely much worse than appears at first glance (and the first glance was did not look very good).

Two days ago, ahead of these reports from the BLS, ADP reported Jobs grow by 57,000 in February.
U.S. private-sector employment grew by 57,000 in February, the weakest job growth since July 2003, according to the revamped ADP national employment index released Wednesday. Job growth in February was about a third of the 166,000 averaged over the previous three months. The ADP index rose by 121,000 in January. Adding in some 20,000 government jobs created in a typical month, the ADP report would signal payroll growth of about 80,000 in February.

Beginning with February's report, the ADP index is being produced with an improved methodology that may reduce some of its past limitations. The sample size has been increased to 400,000 firms covering about 23 million workers, larger than the initial response to the government's survey. ADP is the one of the largest payroll providers in the world; it has more than 540,000 business clients worldwide.
In the past six month ADP has been at times wildly off the mark in their estimates (at least as compared to BLS numbers). To make their numbers more reliable, ADP not only changed their methodology, but increased its sampling size.

For February the numbers are very close. ADP estimated 57,000 private non-farm jobs and 20,000 government jobs. The BLS came in with 58,000 private jobs and 39,000 government jobs. It will be interesting to watch how close these numbers are going forward.

Mike Shedlock / Mish

Last 10 Posts

Copyright 2009 Mike Shedlock. All Rights Reserved.
View My Stats