Our Picks for a Grim Employment Climate--Page 2

<< Previous Page
1 2 3

Birth/Death Adjustments
The monthly change in seasonally adjusted nonfarm employment--net new jobs created during the last month--is closely watched and can significantly sway markets because of the impact job creation and destruction eventually have on consumer spending. The problem is that it's currently impossible to gather data on all businesses every month, so the U.S. Department of Labor uses samples, which can't accurately account for businesses that were created or destroyed during the month.

The BLS' solution is the birth/death model. To put it very simply, this statistical forecasting tool fills the gaps with assumptions based on recent employment trends. However, as stated on the BLS Web site, "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 conclusion here isn't that the BLS' monthly employment numbers are useless or that the birth/death model is a deceptive guise. In steady years--most of the time between recessions--the birth/death adjustment leads to highly accurate estimations of current employment. However, when other data seem to be pointing down and the adjusted employment data keeps pointing up, we should heed the BLS' warning regarding the shortcomings of the model.

Before getting to a few charts, it's important to understand that the birth/death data provided by the BLS is not seasonally adjusted, so it isn't directly comparable to the seasonally adjusted data known to move markets every month. Unfortunately, the BLS doesn't provide the information necessary to make an accurate reconciliation. However, we can apply the birth/death contribution to the nonseasonally adjusted numbers to infer the approximate magnitude of impact on seasonally adjusted data. (For more details on this concept, read the birth/death model FAQ section on the BLS Web site).

The chart below depicts the share of the trailing 12 months' job creation that came from the birth/death model. As you can see, by December 2007, almost 90% of the jobs "created" in the previous 12 months were based on assumptions, not samples. This stands in stark contrast to the 39% average in 2005 and 2006. (The birth/death model wasn't fully implemented until a few years ago, so there isn't much data available.)

To put the significance of this model's assumptions into perspective, we did a scenario analysis to see what the reported job gains in 2007 may have been if the contribution from the birth/death model was curtailed. Our three scenarios assume that once the BLS obtains complete data (sometimes a year or more after the monthly data is released) only 60%, 40%, or 20% of the birth/death contribution will prove to be accurate. We then backed into the employment growth rates that would have been reported in each case.

Note that this approach doesn't work well with January data. The birth/death model contribution is normally negative during this month, which leads to a boost in job growth under our scenarios because we're reducing the negative impact. This is unlikely to have actually occurred. As shown in the line chart above, year-over-year employment growth in December would have slowed to about 0.4% under the base case scenario (40%) and 0.3% under our bear case scenario (20%), instead of the reported 0.6% growth. Although this may not seem like much, it translates into 40,000 and 53,000 extra job losses, respectively. (Nonseasonally adjusted employment normally falls the last month of the year--339,000 jobs were reported lost in December 2007.)

Next Page >>
1 2 3