Last week the BLS announced that 873,000 jobs had been created, dropping unemployment to 7.8% … or that maybe it had only created 114,000 jobs which would actually increase unemployment. The Obama and Romney camps went into overdrive spinning the numbers into all manner of interpretations – those which suited their purposes, of course. Pro-Romney Jack Welch was apoplectic … tweeting conspiracy theories and criminal accusations. Nobel Prize winning Obama shill Paul Krugman, while acknowledging the fact that the job creation numbers are shaky, asserts that anyone who does not see them on the high side is “deranged”.
Given that the source of the numbers is the Bureau of Labor Statistics; and in light of the fun we had six years ago blowing up the BLS productivity numbers – seems they don’t know the difference between offshore outsourcing and productivity – so they just call it all productivity – it seemed worth a look at these curious numbers.
The monthly unemployment figure – the one that said 873,000 jobs had been created and unemployment dropped to 7.8% – the one Krugman thinks one must be deranged to challenge a la Welch – is based on the results of phone calls to one half of one percent of American households. (They call about 60,000 homes out of about 120 million – they extrapolate the answers they get to the other 99.5%) When they get someone to answer they ask if the folks over the age of 16 who live there are working. It is pretty much a straight up ‘yes’ or ‘no’ deal – working 50 hours a week as a machinist counts the same as spending 4 hours a week cutting the neighbor’s grass. In fact, sitting at home sick, because the weather is bad, on strike, for personal reasons or even because you were laid off if you think you might get recalled counts as working. The bar is pretty low to count as being employed. A kid who lives on a farm with a father who makes him do chores counts as working. The other interesting point is that there is no relationship whatsoever between the official monthly unemployment number and the number of folks collecting unemployment compensation.
In the other corner we have Jack Welch – Jack who doesn’t need to actually know anything about the people working in an organization to know that 10% of them (based on numerical evaluations) are marginal workers who ought to go – not 9%, not 11% but exactly 10%. Welch really, really believes in numbers and his over-the -top reaction to suspect numbers from the BLS says a lot about what it must have been like working for him at GE. In his defense, given GE’s reputation for book cooking under his reign he is probably pretty well qualified to spot fiction in the figures,and as one well familiar with Chicago and its politics, I can’t say number juggling from Chicago folks is a wholly unreasonable possibility. Just the same, criminalizing bad data from the BLS is more likely criminalizing incompetence and I don’t think Jack really wants to go there.
At any rate, the low number – 114,000 new jobs – is the result of a survey of businesses, non-profits and government agencies. They basically send them a form asking how many folks are working and hope they send them back in. Surprisingly most of the organizations do send them back, but they do it in their own sweet time. As a result, this number typically gets revised each month for a couple of months until they ‘finalize’ it. According to the BLS, the error rate in this number is such that the real number has a 90% probability of being somewhere between 14,000 new jobs and 214,000 new jobs. The basis for this figure is pretty much the answer to the question, ‘How many people did you give some money to?’ Again, no difference between paying someone to work 4 hours a week or 50 – a job is a job. They acknowledge that, if you have two part time jobs, you count as two people working in this survey, while you would only count as one in the household survey. Another minor issue – if a business goes belly up, obviously there is no one there to send in the form so jobs lost to the tens of thousands of businesses that have failed recently are missed for the most part; and new businesses have better things to do than ring up the BLS and ask to be on the mailing list so they pretty much fall through the cracks, as well.
To sum up what the BLS told us last week, unemployment might have gone down, or it might have gone up, there is about a 90% probability the number of jobs created was somewhere between 14,000 and 873,000, but the low end of that might chage in a couple of months, and there is a one in ten chance it was something different all together, and whatever the number was it might be people working full time but it might be a bunch kids working part time.
And from this profound analysis we will elect a President, billions of dollars will change hands on stock markets, lionized economists and business leaders will reap headlines and the course of the greatest economy in the world will be charted.
Jeff says
Thanks for the thoughtful dose of reality Bill. While not into the conspiracy theory being tossed around I do wonder about the bias in the BLS numbers. For the past 4 years the weekly unemployment claims is consistenty 3,000-6,000 low. Every week but one thet have “adjusted” the previous week’s numbers upward. The net impact of this is to consistenty understate the number of unemployed and to give the impression of reductions because the headline always compares this week’s understated number with last week’s adjusted number. I’ve seen 4 week periods where the headline is a reduction in new claims when there has been a steady increase in new claims. An unbiased estimate would understate sometimes and overstate other times so the net average should be close. If you understate the actual number a hunderd times in a row you must be biased.
William Roger says
I would agree that the basis for the numbers may be flawed. But that flaw applies to the numbers that have not gone down as predicted or have gone up or have stayed the same. It wasn’t but a few weeks ago that the numbers were not as high as expected and one party defended and the other pounced.
I would be interested in a solution that would assure us all that the numbers are rational first and foremost. The political noise will continue no matter what but the information should be reliable.
What alternate means of determining the actual employment figure would someone suggest?
grahamc says
Why do you need to establish the employment figure? Why not the unemployed number or those who are in receipt of government support of some form.