Getting Behind Today’s Employment Report

Those who have been following the job creation story weren’t surprised at the weak BLS employment (the so-called “non-farm payroll”) report today, or by the fact that it revised the results for the last three months downward by 81,000. Various commentators (see here, here, and here) have observed that the rises reported in recent months have been due almost entirely to a “voodoo calculation, the “birth/death model” a plug to allow for business creation and failure (one we’ve commented on before). Note this feature was added in 2001.

And where had these jobs supposedly been created? In construction. Need we say more?

Barry Ritholtz offered his latest observation about the birth/death model before the BLS report was released:

Considering how much softer the economy has been much in 2007 than last year, it is simply unconscionable that the B/D model has actually created more jobs in 2007 than it created at this point in 2006.

Year Jan Feb Mar Apr May Jun Jul
2006 -193 116 135 271 211 175 -57
2007 -175 118 128 317 203 156 26

The Economic Policy Institute has a very good write up:

In an unexpectedly and extremely weak employment report from the Bureau of Labor Statistics, the nation’s payrolls shrunk last month, the first monthly decline since August 2003. Prior months’ job gains were revised down by 81,000, leading to an average monthly gain of just 44,000 over the past three months. Such a low monthly rate of job growth provides the first solid evidence that economic turmoil in financial markets has found its way into the job market.

The unemployment rate, calculated from the survey of households, was unchanged at 4.6%, but this was due not to job growth in that concurrent survey. In fact, employment fell 316,000 last month according to the more volatile (and thus less reliable on a monthly basis) household survey. Averaging over the year, that survey shows a monthly decline in jobs of 16,500 so far in 2007.

The reason August’s unemployment rate was unchanged was due to a large monthly fall off—down 340,000—in the labor force (those who leave the labor force are not counted among the unemployed). In other words, the unemployment rate was unchanged due to fewer job seekers, not more jobs. Had those who left the labor force instead been counted as unemployed, the rate would have jumped to 4.85%. (The BLS noted that part of the decline in the labor force was driven by teenagers going back to school, but the adult male labor force, which had been growing in general, also fell by over 100,000.)

Many industries shed jobs or grew slowly in August. The monthly diffusion index, a measure of how many industries are adding jobs, hit its lowest level since February 2004. In an acceleration of an already negative trend, factory employment was down 46,000, the largest monthly loss in that sector since 2003, dispelling hopes that strong exports in the second quarter might help stem these losses. Auto manufacturers made the largest cuts in the factory sector, down 11,000 jobs.

Reflecting the housing market turmoil—including sharply diminished home sales, rising inventories, and falling prices—construction employment fell 22,000, driven largely by a decline in residential contractors. Related losses also occurred among financial institutions that deal with mortgage lending. Putting these sectors together, EPI’s residential index fell by 28,000 jobs in August, and is down 127,000 since April 2006.

Local government was also a big job loser last month, driven by a 32,000 loss in public education. Given that the largest job losses were concentrated in factories, construction, and government, private service-providing industries were net gainers, adding 88,000 jobs last month. However, growth in this sector has also decelerated. On average, private services added 100,000 jobs per month over the past three months, compared to 150,000 per month over the prior 12 months.

The office sector (professional and business services), for example, which is a large, important engine of job growth in the service economy, added only 6,000 jobs in August, largely due to contractions in temporary help, another sign of cyclical hiring weakness.

Health care, however, was up 35,000 jobs and continues to buck any negative trends by steadily adding employment. Restaurants and bars also gained 24,000 jobs, and retail trade added 12,500.

These latter gains in retail and food services suggest consumers are still spending freely in some areas. In fact, on a yearly basis, both hourly and weekly earnings were up 3.9% in August, a solid rate that is likely to beat inflation, barring unforeseen price spikes.

However, if the pervasive weakness in today’s report continues, any wage pressure in the job market will shortly begin to let up, and wage growth will slow. This, in turn, will show up as weaker consumption and ultimately slower overall growth, reinforcing the trend in weak hiring of recent months.

A central question surrounding today’s report was whether it would provide clear evidence of a contagion effect from financial markets. Are the bursting housing bubble, the credit crunch, and recent financial market turmoil having a negative impact on the job market? The BLS report provides an unequivocal “yes” in response to that question. The fingerprints of these problems are all over today’s report. While we cannot draw a firm conclusion based on one month of jobs data, a host of factors point to a new and troubling job market: negative revisions of earlier months’ data, widespread losses and slowdowns across industries, and weak labor force growth.

Print Friendly, PDF & Email


  1. Laurent GUERBY

    I’m still why economists are still using the unemployment number instead of the employment to full population ratio (in particular male 25-54 where if you’re not working there’s a problem).

    Any idea?

    I wrote on wikipedia unemployment article a while ago:

    For the fourth quarter of 2004, according to OECD, (source Employment Outlook 2005 ISBN 92-64-01045-9), normalized unemployment for men aged 25 to 54 was 4.6% in the USA and 7.4% in France. At the same time and for the same population the employment rate (number of workers divided by population) was 86.3% in the USA and 86.7% in France.

    This example shows that the unemployment rate is 60% higher in France than in the USA, yet more people in this demographic are working in France than in the USA, which is counterintuitive if it is expected that the unemployment rate reflects the health of the labor market [10].

  2. Yves Smith


    I agree with your observation. It’s a much better measure and use of it would force more discussion of why the non-working population isn’t working.

    The word “economists” covers a lot of people. The profession has become very specialized. From what I can see, labor economists look mainly at policy issues, like the impact on employment of minimum wages. And frankly, I think their research is only selectively read by the mainstream press.

    So in my view, the real issue is the reporting by the financial press, plus Wall Street and other analysts. And unfortunately, governments all around the world publish unemployment stats. They are easy to find, and most governments define employment pretty consistently (how they define the pool of labor is more variable).

    So most people, including people who should know better, think the unemployment stats are pretty valid. Recall some of the comments on the post you mentioned earlier, where some people vigorously attacked questioning of the unemployment figures, and specifically, the notion that our unemployment rate might actually be comparable to Europe’s. One writer actually got emotional about it

    If you believe our unemployment levels are in the ballpark of Europe’s it says our Darwinian “employment at will” approach to workers might not be justified. It creates a lot of job losses and stress on families, yet the supposed benefits of higher production (and thus in the end higher employment) might not be there.

    That’s as close as you get to a religious issue in the world of commerce.

  3. Anonymous


    A note on payroll indicated errors.

    “The labor force participation rate decreased to 65.8 percent in August, largely reflecting a decline in participation among teens. The labor force participation rate of teenagers declined by 1.5 percentage points to 39.7 percent. The household survey reference period fell
    relatively late this August (covering the week from Sunday,
    August 12 through Saturday, August 18) and, as a result, a larger-than-usual number of teens had left the labor force to return to school when surveyed. While the movement in August may have been exaggerated by the timing of the survey week, the labor force participation rate of teenagers had
    been declining recently–from 43.4 percent in December 2006 to 41.2 percent in July 2007.”

    How has the late survey impacted numbers? Well they indicate it may exaggerate the decline in teen employment.

    The unadjusted fall in teen workers in August is -910k the largest in a decade except 2001.

    This translates to a seasonally adjusted fall of -275k the largest in a decade again apart from a blowout -428K in 2001.

    How about teen participation rates? Here the indicated potential error would lower the participation rate.

    The unadjusted fall in teen participation July-August is -6.6%, average for the decade.

    But the adjusted fall is -1.5% the largest in a decade (double the next largest)apart from a blowout -2.5% in 2001.

    These acknowedged potential errors have not been corrected for in the published seasonal adjustments.

    Given the large survey error +/-500K and the small result -4K

  4. Anonymous

    At this rate, who needs a college degree & education?

    Btw, Mish had some good points on the massive layoffs too.


  5. Anonymous

    It’s really too bad that this form of government doesn’t allow a wholesale change of the legislative body. Eight years of this is enough to do some serious damage.

    And btw, Robert Reich seems to have a new book out advocating (if I understand correctly from the article I read) getting rid of corporate taxes as tradeoff for their rights as citizens. Any comments?


  6. Anonymous

    My mistake, the household data relate to unemployment numbers and not payrolls.

    The errors might add 100K plus to each of the labor force and employment numbers and reduce the reduction in unemployment.

  7. Anonymous

    Negative impact on the job market:

    Bursting housing bubble: Yes
    The credit crunch- housing related: Yes
    Recent financial market turmoil: Not yet

Comments are closed.