Will the Real Jobless Number Please Stand Up?

03/18/2010 05:12 am ET | Updated May 25, 2011

The brothers Grimm began spinning fairy tales in 1907. Now 102 years later, the Bureau of Labor Statistics, the government agency that doles out those closely watched monthly employment figures, seems to be giving the Grimms a run for their money.

In brief, the BLS is stacking the employment deck, a number of market professionals say. Or more specifically, they argue it is increasingly issuing misleading figures that paint a rosier picture of the country's biggest economic dilemma--mounting job losses.

In effect, the critics suggest that President Obama should factor in a lot more workers in his job-creating efforts because his labor analytical arm is woefully underestimating the number of unemployed workers.

The latest such BLS numbers-fudging is said to involve November's shockingly low jobs loss of 11,000, when the general expectation called for a much higher jobs reduction for the month of 125,000. That drop of 11,000 jobs, which was the best monthly performance since September 2008, knocked down the nation's unemployment rate to 10% from October's 10.2%. What had been widely expected for November was another 10.2% showing.

Other reported labor figures for the month, however, take issue with the accuracy of the BLS data, which, it should be noted, is very often later revised for the worse.

For example, Automatic Data Processing, the country's largest payroll service, which surveys 24 million employees a month, reported 169,000 jobs were actually lost in November. That's 158,000 higher than the BLS numbers.

Further, a survey by the Institute of Supply Management (non-manufacturing) shows an even bigger disparity, a contraction in November employment in the service sector alone of 192,000 lost jobs.

In addition, the latest weekly unemployment claims (for the week that ended November 27) showed 457,000 filings, which is indicative of a lot of employees being shed. Anything above 350,000 to 400,000 filings is generally viewed as a contracting employment environment.

Further casting doubt on the BLS' latest results, Monster Employment, an online jobs posting service, reported that job demand weakened from October to November.

Madeline Schnapp, the director of economics at TrimTabs Investment Research, a liquidity-tracking service out of Sausalito, Ca., says her analyses of the facts suggest that about 255,000 jobs were actually lost last month, which indicates an unemployment rate of 10.4%.

Schnapp says she has several problems with the BLS numbers, which she describes as "erroneous and which tend to inflate the results."

For starters, the BLS data is seasonally adjusted. In its model, the BLS assumes retailers will hire a pre-determined number of employees each year in October and November, roughly one million each month, and then let them go in January. So the BLS adjusts its survey every year to reflect such numbers in its seasonal change.

This year, though, reflecting lower inventories, expectations of less traffic and the belief that holiday sales will be lackluster, there won't be as many seasonal hirings as usual. The problem, Schnapp says, is BLS' seasonal rigid assessment doesn't factor in these changes and probably overestimated the seasonal hirings.

Further, she notes, only 40% to 60% of the BLS' survey is complete when the monthly jobs data is released.

The agency's evaluation of the significance of government hiring--which has been sizable this year--is also seen as a distortion of the jobs numbers even though government jobs represent just 17% of total employment. Yet, Schnapp notes, they account for 70% of the survey respondents. The BLS attempts to statistically adjust the responses, but past results, she points out, show it doesn't do a very good job.

In contrast, TrimTabs uses real-time daily income tax deposits to the U.S. Treasury to compute aggregate wages and salaries and then uses this data as input into an employment model to determine jobs gains or losses. Historically, Schnapp says, "our methodology has proven to be more accurate than the BLS' when all the revisions are in."

Schnapp, by the way, doesn't believe, as some do, that the government may be fudging the numbers. She blames the misleading numbers on what she says is "BLS' rigid econometric models." In the past, she notes, monthly revisions to its jobs data have run as high as 700%, and in September and October, for example, the revisions were 45% downward.

What does the BLS have to say in response to Schnapp's criticisms of its jobs-gathering methodology. Alas, nothing. "We won't comment on the caliber of our work," says spokesman Gary Schwartz.

Meanwhile, job losses, as Schnapp sees it, are in the process of getting less bad, although she expects the employment market to remain weak until the summer of 2010.

Her projections call for diminishing monthly losses, with 190,000 expected this month, 150,000 in January, 100,000 in February, 75,000 in March, 50,000 in April, 25,000 in May and no job losses in June.

Sounds good, a final end to the job losses. But, as Schnapp cautions, it's not an overnight happening, new hirings will be slow, and there will still be plenty of pain on the employment front before the end is in sight.

What do you think? E-mail me at