October 9, 2006

WSJ Notes Employment ‘Whoops,’ But Should Have Gone Further

Filed under: Economy,Taxes & Government — Tom @ 10:47 am

From a subscription-only editorial today:

The Labor Department released its September jobs report on Friday, and some wags are calling it the “whoops” report. The “whoops” is a reference to the upward revision of 810,000 previously undetected jobs that Labor now says were created in the U.S. economy in the 12 months through March 2006.

So instead of 5.8 million new jobs over the past three years, the U.S. economy has created 6.6 million. That’s a lot more than a rounding error, more than the number of workers in the entire state of New Hampshire. What’s going on here?

Our hypothesis has been that, due to the changing nature of the U.S. economy, the Labor Department’s business establishment survey has been undercounting job creation from small businesses and self-employed entrepreneurs. That job growth has been better captured in Labor’s companion household survey, which reported 271,000 new jobs in September after 250,000 new jobs in August, and a very healthy total of 2.54 million new jobs in the past year.

The Journal missed the fact that there is STILL a 3 million-plus job difference since January 2002 between the Household and Establishment reports (see this post showing a difference of about 4 million last month before the “whoops” adjustment). Labor’s “whoops” adjustment recognizes that the error is in the Establishment report, and raises at least these two current questions:

  • How many more jobs has the Establishment Survey missed in April through September of this year? It would appear, based on the 12-month revision of 810,000, that there may be 400,000 unreported Establishment Survey jobs created in the past six months.
  • Now that the Establishment Survey has been shown to be flawed, will the 527 Media give at least equal, if not greater, billing to the Household Survey’s usually-higher number of jobs created? I would suggest that they should; given that the Household Survey will ordinarily make the employment news look better, I’m going to “guess” that they’ll continue to ignore it, even though they have to look at it every month for the unemployment rate.


  1. We actually have an independent source on this issue — the Social Security Admnistration’s tax collections. If your adjust for changes in hours and wages, this had (through 2004) tracked the establishment survey very closely. So, I would be hard-pressed to take the results of a survey based on 60,000 households over two surveys effectively baed on two independent censuses of the labor market (UI filings and SS taxes).

    It is also worth noting the HH data substantially understated job growth compared to the establishment survey in the 90s. Does anyone want to conjecture as to the radical change in the labor market since 2000?

    Comment by Dean Baker — October 9, 2006 @ 5:28 pm

  2. Dean, thanks for the comment. I\’m glad you did, because I meant to blogroll you yesterday (under Money), and forgot.
    I was introduced to your work yesterday by a commenter at my NewsBusters post on the deficit, and I read a couple of your columns on the late 1990s bubble. It is worth noting that somebody in the left side of the economic aisle was smart enough to recognize the overhyped reality at the time.

    On the employment issue, your point is very well-taken. My response would be that the Establishment Survey has ALWAYS understated total employment compared to the Household Survey (i.e., the Establishment Survey has always been wrong), and that it is simply a matter of degree. The gap between the two now is about 8 million jobs and it was significantly narrower during the mid-late 1990s.

    I fail to come up with a reason why the Household Survey might overstate the total number of jobs that is as compelling as the reasons why the Establishment Survey always undertstates, which is simply this: it would seem to be a lot harder to locate a new business, especially a proprietorship or partnership without employees who contracts out all work, than it is to locate a new household. Overstatement of Household Survey workers would depend on people lying by saying they have jobs when they do not, which seems unlikely.

    As to your point about SocSec records — there are lot of people who are 1099 contractors and self-employed, probably more than there used to be. I question whether SocSec is as good at tracking these. Even if they are, people who have moved into contracting or self-employment will not get picked up as working by Social Security until they start making estimated payments. Many do not do this in their first year and get their April surprise when they fill out their 1040s; by that time they may have been working as long at 15 months. Additionally, many contractors NEVER make estimated payments because they have their spouses overwithhold at their employer. I would guess it might take SocSec a good two years before they pick up a self-employed person or contractor in that position, because they totally depend on the IRS to pass the data over.

    Comment by TBlumer — October 9, 2006 @ 6:47 pm

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.