Who knew jobs data could be so exciting?

By Douglas Holtz-Eakin
October 8, 2012

The September jobs report ignited a firestorm when Jack Welch, former General Electric chief executive officer and Reuters contributor, asserted (or implied, or wondered if) the unemployment rate had been politically doctored to give President Barack Obama an electoral advantage. After all, how can the unemployment rate drop a full 0.3 percentage points to 7.8 percent when the economy is creating only 114,000 jobs?

More on that later. First, let’s dismiss the notion that the integrity of the data-collection process was undermined. Anyone at all familiar with the production of federal economic statistics – at the Bureau of Labor Statistics, the Bureau of Economic Analysis, the Federal Reserve, or elsewhere – can appreciate the firewalls that exist between the professional collection, analysis and publication of economic data and the remainder of the agencies’ missions – especially their political appointees. It is unfathomable that these would be breached.

It is even more unfathomable that they would be breached without the career civil servants getting on the telephone to, say, Reuters and reporting the political manipulation within a nanosecond of it occurring. And still more unfathomable that such a breach would be initiated and covered up successfully, while only lowering the rate to 7.8 percent. Why not 6.8?

For many, that simply makes the puzzle all the more baffling. How did this happen?

The employment report, as we now all know, has two measures of job creation: the payroll survey and the household survey.

In September, the payroll survey – derived from asking employers how many people they employed that month – showed that the economy created 114,000 jobs. This is consistent with an economy growing at 1 to 2 percent. The household survey – derived from asking households who in the house has a job – showed a stunning 873,000 new jobs. This is this highest that number has been since June of 1983. This makes no sense; it is out of line with any of the other data on the economy for September.

Even more amazing, more than 560,000 of those are part-time jobs. That is really stunning. One would expect that as extended unemployment benefits expire (and they are), some workers would migrate back to employment – that is a tried-and-true economic link. Some of those might first end up in a part-time setting. But why 560,000? And why in September?

The fact is that the household survey has only 55,000 to 60,000 households in the sample. To make it representative, it has to cover gender, age, race and education level. That means that there are likely relatively few people in any particular “cell” – how many female, Asian-American, 34-year-old, high-school-educated individuals will be in the September survey. So there is always the chance that the sample is simply not representative of the U.S. population and a misleading indicator of what is going on in the labor market. That appears to be the case here.

So in my first reading of the Bureau of Labor Statistics report, I throw out the household survey.

Digging a bit further, the September jobs report is solid – assuming you have bought into the new normal. Yes, the economy is creating jobs. But the pace remains far too slow to dig the U.S. out of its malaise quickly. At a rate of 114,000 jobs per month, unemployment would not get back to 6.6 percent (the average unemployment rate two years into the average postwar recovery) until September 2026. That pace is far, far too slow.

Average hourly earnings finally increased in September. One component of recovery that has largely been missing, and unreported, is increases in real, disposable income. So rising earnings is no doubt a positive change for the economy.

Labor force participation is up 0.1 percent, to 63.6 percent. Over the past several months we have seen labor force participation plummet – indicating that people are giving up all hope of finding work. It’s positive to see consumer confidence increase. Americans who have been struggling for years are looking for work again – and finding it.

What we learned Monday is that the economy is not falling, but it is not accelerating. The economy is moving sideways.

The broadest measure of unemployment remains unchanged at 14.7 percent. We are still facing a number of downside risks, still generating too little income, and well behind schedule in recovering.

So, on the whole, September’s report was modestly positive and no cause for either conspiracy theory or celebration.

But get ready for more finger-pointing and controversy. In October, the statistical anomalies will likely disappear, the 873,000 jobs will evaporate and the unemployment rate will jump north just prior to the election.

Who knew economic data could be so exciting?

PHOTO: People wait in line to enter a job fair in New York, April 18, 2012.  REUTERS/Shannon Stapleton

6 comments

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

Given the fact of your “unbiased” background, as well as the other BS buried in the article about “sample size” (no doubt to confuse anyone who knows nothing about statistics), the problem lies not with only this one report, but the whole process.

Your statistical methods belong back in the 1930s, when there might have been some excuse for the incredibly bad data produced on US unemployment.

The problem is there is no incentive for the government to “clean up its act”, because to do so would put them under enormous pressure to act in a fiscally responsible manner.

I notice that there are no other comments to this egregiously self-serving article, so I suppose Reuters is once again suppressing news that should be released.

So much for democracy and free speech.

Posted by Gordon2352 | Report as abusive

Well, I was wrong — about Reuters suppressing this article — but that means an even worse scenario, that the American people don’t give a damn about the numbers.

Failure to force the government to reveal the correct data in a timely manner means you deserve what you get.

Posted by Gordon2352 | Report as abusive

@Gordon2352,
“The problem is there is no incentive for the government to “clean up its act”, because to do so would put them under enormous pressure to act in a fiscally responsible manner.”
Bravo! Bravo!

The bigger problem is that the government in Washington is like “memory foam” without the memory. It shifts it’s shape constantly to pressure applied here and there, but the size, mass and trajectory towards fiscal chaos never changes.

Most young people don’t know that the United States never paid off the debt it ran up during WW II as the “armory of the free world”. Our politicians just inflated it over the years since to insignificance.

That’s the “good” news. The “bad” news is that they can just as easily do that to your pension, Medicare and savings too!

Remember the “doomsday clock” relentlessly counting down to the moment humanity destroys itself militarily? Well it certainly appears that world governments have created a comparable economic scenario.

As a visual aid we need another ominous clock ticking down to the moment that the combined idiocy of politicians and associated rates of government printing of money without meaningful collateral accelerate the a chart line consistent with the rate of inflation and concurrently accumulating debt to “straight up” infinity!

Posted by OneOfTheSheep | Report as abusive

Very good article explaining the workings of the BLS; the thing you have to wonder about is that the BLS would rely on such an ‘unscientific samplng’ methodology to extrapolate such an important economic metric, and that they wouldn’t themselves be ‘concerned’ with the results..
Don’t you think that they should have questioned the sampling results when it produces such a highly doubtful answer, before they published? I agree that the unemployment rate was probably not purposely skewed, but when the analysis produces an unreasonable answer, you don’t publish; you check and revise.. This has got to be an embarassment for the BLS, especialy if the rate jumps Nov 1st.

Posted by texas100 | Report as abusive

Updating the methods for these types of measurements presents a damned if you do and damned if you don’t situation.

Using the same flawed methods does allow comparing historical numbers for trending, even allowing for the numbers being less accurate than they could be.

Updating to a new methodology may get better numbers, but then you lose that historical comparison unless you still use the old method for comparison purposes and the new one for going forward, but that adds costs and complexities to an already complex process.

The same is true for all of our economic, productivity, and production measurements.

Posted by mikemm | Report as abusive

Many have said that it is impossible for the number to be manipulated – that to do so would require too large a conspiracy involving many statisticians in the BLS.

Bull. All it takes is one hacker. Are we to believe that the same country that was able to destroy Iran’s centrifuges by hacking the computers that control them isn’t also able to hack its own BLS database? Alternatively, isn’t it also conceivable that the federal government simply hired more people from the list of households on the survey? Is it mere coincidence that September marked the first month in a very long time that government jobs rose?

The establishment survey said that, as expected, relatively few jobs were added in September. That figure was corroborated by the ADP employment report a day earlier. Yet, somehow we’re to believe that the employment level (from the household survey, used in the unemployment calculation) rose by 873,000 – the highest since 1983 and the fourth highest since record-keeping began? In this economy? And it just happens to yield an unemployment rate that is exactly 0.1% less than when Obama took office? One month before the election?

You have to be extremely gullible to believe that this isn’t a manipulated number.

Posted by Pete_Murphy | Report as abusive