Wednesday, November 28, 2007

How Strong is the Labor Market Really?

From Barron's:

Charles Biderman, the proprietor of TrimTabs Investment Research, asserts the latest Labor Department numbers showing continued growth in payrolls is wildly overblown. That's without getting into the contentious argument over the Bureau of Labor Statistics' birth-death assumptions, which magically conjure phantom jobs.

Biderman, a Barron's alum who toiled for the magazine in the 1970s, tracks real-time data about the stock market and economy, including employment. Biderman's modest proposal -- follow real-time, real-world data, such as withholding and self-employment taxes, and ask payers of those taxes how many workers they employ. He also tracks employment indicators such as online help-wanted ads, and not just those on commercial sites.

This gives a better picture of the labor market because increasing numbers of people aren't on payrolls. It isn't just the eBay entrepreneurs touted by Dick Cheney. Lots of professionals have been laid off by corporations and then work for their old employers as "consultants," Charles observes. But, being hidebound bureaucrats and academics, the BLS resists utilizing these data.

In any case, nobody fudges their taxes to inflate the numbers. And based on those numbers, while the consensus was looking for a slowdown for the past couple of years, Biderman's real-world data showed the U.S. economy was cooking. That is, until a couple of months ago.

For instance, the official tally showed a 130,000 gain for nonfarm payrolls for October. By Biderman's reckoning, payrolls actually were down 30,000 for the month, which would jibe with the 250,000 decline in last month's household survey.

To be sure, the household numbers bounce around from month to month. But over the past three months, household employment has declined by 34,000 a month. Over the past 12 months, this measure has grown a paltry 56,000 a month, or about half what's needed to absorb new entrants into the labor force.

Part of the discrepancy between the BLS payroll numbers and reality is because a huge number of people in real-estate-related businesses are independent contractors. That includes construction workers, realtors, mortgage brokers, home inspectors and assessors, among others. They're not on an employer's payroll, so they don't get counted in that tally. Nor are they eligible for unemployment insurance, so they also don't get counted in the jobless claims data.

Models and statistics aren't my strong suit, so I have to rely on what other people are saying about the BLS' model. However, there has been a consistent refrain about the recent employment numbers and that is the birth/death model is distorting the picture.

For those of you who are unfamiliar with this BLS model, here's the basic deal (and for those of you who have a much better grasp of actual situation add it to the comments or email me at bonddad at prodigy dot net). The method the BLS uses to determine how many jobs are created every month does not include new businesses or businesses that have recently gone out of business. To compensate for this problem, the BLS has created a "birth /death" model that attempts to compensate for this problem.

The problem with the "birth/death" model adjustments is the adjustments are more and more responsible for the total number of jobs reported over the last year. In other words, the model and not the economy is creating jobs. I've also read the BLS' reported numbers are seasonally adjusted while the birth/death model isn't which makes comparisons difficult.

For a more detailed assessment of how this hit the latest report, see this article from Mish: Here is a great observation from the article:

The BLS has shown a net gain of jobs added to new businesses in both construction and financial activities nine consecutive months from February through October.

The BLS is assuming not only that jobs were added, but that new unaccounted construction businesses were created in this environment where business capex spending has been weak, housing has been horrid, and over 170 lenders have gone out of business or stopped writing loans since last December as per the Mortgage Lender Implode-O-Meter.

Clearly this is reporting from an alternate universe.

A note of caution: One cannot take the birth death adjustments and subtract them from the reported numbers because one set of numbers is seasonally adjusted and the other is not.

The bottom line is there appears to be some serious problems with the official numbers that we are seeing. As to the degree and depth of those problems, I will leave that to far more statistically oriented minds.