Huzzah! We are saved: The economy is being rescued by manicurists!
At least, that is the proposition put forth by Virginia Postrel in the Sunday New York Times Magazine (Essay: “A Prettier Jobs Picture?“). Not just manicurists, but a whole slew of other positions that are bring under counted. Postrel argues, as have so many before her, that the economy is actually vibrantly creating many, many new jobs; Government statistics — at least the Payroll survey — is unfortunately under counting them.
The Times balances out this starry eyed view with a more sober questioning of the survey divergence issue in their Business section (Economic View: “Two Tales of American Jobs“). Edmund Andrews discusses the diverging employment counts between the Payroll and Household employment surveys.
In reality, it has long been known that these two surveys diverge, and often dramatically. Fed Chairman Alan Greenspan tried to put this issue to rest recently:
“Unfortunately for the optimists, the Federal Reserve has just thrown cold water on the household data. It concludes that the gloomy payroll data is essentially accurate and that the household survey is probably off base.
“I wish I could say the household survey were the more accurate,” Alan Greenspan, the Fed chairman, said in his testimony at a House hearing on Feb. 11. “Everything we’ve looked at suggests that it’s the payroll data which are the series which you have to follow.”
This is not the first time the Fed has looked at the dichotomy. The NY Fed analyzed the issue in 1999 (“Explaining the Recent Divergence in Payroll and Household Employment Growth“). Their conclusion was that the household survey can indeed under count employment numbers — but not necessarily in a good way. The champions of the Household survey (i.e., Postrel et. al.) overlook an important distinction between the two surveys — one measures jobs, while the other measures employed persons: The New York Fed noted why this matters:
“The employment increases reported in the Payroll survey are likely to exceed those reported in the Household survey because the payroll survey counts the number of jobs in the economy while the household survey counts the number of employed people. Since many workers hold more than one job, the payroll survey would be expected to yield higher estimates of employment” (emphasis added).
You read that correctly. It is not that the Payroll survey under counts employment — according to the NY Fed, it over counts it.
Proponents of the Household survey overlook several other statistical niceties. The Household survey is based upon 50,000 people, and then extrapolated to fit a nation with 293 million people. Tiny errors in the original data run get dramatically magnified in the process. Now, compare that methodology with the monthly Payroll records of half a million businesses, which employ 50 million people. Its not too hard to see why Greenie relies on the latter and not the former.
Then there’s the different data collection methodologies employed — the Household is a “self reporting survey,” whereas Payroll employs an objective corporate data collection method. Self reporting surveys are notoriously unreliable. Many people exaggerate, if not outright lie. Whenever you ask personal questions about matters which impact pride and self esteem, you create a naturally unreliable process. As Thomas Gilovich wrote in “How We Know What Isn’t So,”
“One of the most documented findings in psychology is that the average person purports to believe extremely flattering things about him or herself — beliefs that do not stand up to objective analysis. For example, a large majority of the general public think they are more intelligent, more fair minded, less prejudiced, and more skilled behind the wheel of an automobile than the average person. This phenomena is so reliable and ubiquitous that it has come to be known as the “Lake Wobegone effect,” after Garrison Keillor‘s fictional community where the women are strong, the men are good looking, and all the children are above average.”
This last issue ties into our own observations of the “so-called self-employed consultant.” Many people do start solo practices and consultancies which then scale up into legitimate and self sustaining profitable businesses. Indeed, many of the Big Picture’s friends in both programming and graphic design transitioned “between jobs” as consultants. I wish to emphasize how important these people are to both the entrepreneurial environment and the broader economy as a whole. Gregory Mankiw, chairman of the White House Council of Economic Advisers, has noted a rise in the “self-employed workers,” saying the “extent of self-employment has changed as the economy has changed.” However, we here at the Big Picture have noted the obvious discrepancy this creates: the sad tale of the so-called Self-Employed Work-at-Home Contractor. Simply stated, there are many so-called contractors who are simply unemployed. Its far more palatable to friends and families — and potential employers — to adopt the persona of the consultant.
I hasten to add that these consultants were not buying homes or cars or other big ticket items, simply because their cash flow was either insufficient or just too erratic. Try getting a mortgage when you tell the bank you are a “Self-Employed Work-at-Home Contractor.” (Good luck!)
Then, there is the small reality of starting a new business: They tend to fail, and in large numbers. Indeed, the majority of all new businesses are fighting an uphill battle. That’s not saying that people shouldn’t try, nor does it by any means denigrate those who create from scratch something vibrant and profitible. It is merely a recognition that many — indeed most — of these new business won”t be around a year from now. Hey, that’s capitalism, where only the fittest survive.
The Household survey “has not historically been a better leading indicator of job growth,” noted Mark Gongloff in an article last October (“The job market: a matter of opinion“). He cited a recent study by Anthony Chan, chief economist at Banc One Investment Advisors:
“According to his research, in the first 21 months of economic recoveries since the 1950s, the payroll survey showed average job growth of about 3.59 million, while the household survey showed job growth of just 3.09 million. In other words, Chan said, the payroll survey usually does a better job of picking up a labor-market rebound.
“This clearly shows that relying on the household survey’s employment measure as a barometer of labor market conditions is not only risky, but also an incorrect assumption,” Chan said.
What’s more, the household survey employment numbers include a “1 million-job jump from December 2002 to January 2003 that was simply the result of Labor Department adjustments in number-crunching. As a result, this year’s employment data aren’t comparable to last year’s.”
Finally, we must beware of those bearing an agenda. Gongloff noted that many of those advocating relying on the Household survey “seem to have a political agenda, since they’re tied to conservative causes or have been vocal proponents of President Bush’s economic policies.”
I have on several on several occasions noted my annoyance with the “massaging of data, the presenting of false projections, the politicizing of matters best left unpoliticized.” I do not read Postrel that often, so I will reserve my final judgement until I have seen a greater sample of her work. That said, this article does not bode well for either her economic insight or analytical acumen.
Her arguments that the BLS undercounts jobs in professions such as stone cutting (granite kitchen counter tops) and manicurists and massage therapists and gardeners and graphic designers, as well as many other categories, may even be right. The problem is that the BLS has always under counted these jobs. Always.
What is missing from the advocates of the Household survey is a persuasive explanation as to why this is a new phenomena. Show me that these are brand new jobs, created during the past 3 years. And to be fair, there is some small support for that view.
But consider the tortured path we must take to rely upon that conclusion: anecdotal evidence of historically under counted jobs, which we then must believe on faith are being created anew. This, despite the Fed chair saying this data source is not to be strongly relied upon. Not only must we use the old saw “the data must be bad” to prove the point — but in this case, its not just that the data which is bad — we must also ignore the data which the Fed notes as “more accurate”
OK, rely on the bad stuff, discard the good stuff, overly rely upon anecdotal evidence, and make several unsupported leaps of faith. Forgive my skepticism, but I tend to get suspicious when a flagellating argument runs counter to everyday ordinary experiences.
As we have been saying for some time now, this has been a very unusual recovery. Structural changes in the job markets caused by post bubble over capacity, productivity gains, and outsourcing are additional reasons for these changes. This is significant for the macro economy, the labor market, and quite possibly the 2004 Presidential Elections.
Pretty Job Picture? No, not really. Household survey advocates need to take off their beer goggles . . .
UPDATE: 02/22/04 2:46PM
Prof Brad DeLong notes the methods used to collect the data in the three (not two) surveys: Don’t forget the “Survey of Occupations” in addition to the Household and the Payroll Surveys.
“There is no reason to think that the totals of nationwide employment–which are derived from these second and third of these data sources–are substantial under counts because of any significant “bias against small enterprises and self-employment”
Two Tales of American Jobs
Edmund L. Andrews
N.Y. Times, February 22, 2004
How We Know What Isn’t So
Thomas Gilovich, 1991
pages 76 -77
The job market: a matter of opinion
CNN/Money, October 2, 2003: 3:01 PM EDT
Explaining the Recent Divergence in Payroll and Household Employment Growth