Buy the dip no longer sounds sufficient to calm fears, nor will forward guidance. Jerome Powell will...
A Look Behind ‘Improving’ Employment
02/07/2012 7:45 am EST
A lot of people are chirping about how much the US economy is improving, but don’t be too quick to jump on that bandwagon, writes John Mauldin in Thoughts from the Frontline.
The most recent non-farm employment report was good: 243,000 jobs, and they were not just in the health-care and food and beverage categories, but across the board. Unemployment dropped to 8.3%.
There were some early comments that the unemployment number was lower because another 1.1 million people dropped out of the work force, no longer looking for work. If you read just the simple number, you might think that. But there were asterisks all over this report, telling us we had to look deeper. A lot deeper.
First, this was the normal month for annual revisions, when the Bureau of Labor Statistics (BLS) makes adjustments to the prior year’s data, based on new information. And there were some extensive revisions. So the number in the workforce did not actually drop.
So this isn’t some new January phenomenon, but a result of the BLS using the 2010 census data to have more accurate data. In other words, the changes in the Household Survey to the various measures had taken place over the years prior to 2010, but for simplicity’s sake, the BLS incorporates these changes into one month (which they clearly point out)."
Spread out over ten years, 1 million people is not all that much on a per-month basis. If you just looked at the numbers in the actual release, it would also lead you to believe that somehow last month around 1.2 million working white men and women just disappeared, or that the number of working Hispanics rose by 800,000. There are a lot more of those types of anomalies.
But they are also explained by the fact that the BLS incorporated the recent 2010 census data into their formulas. Apparently, the Census Bureau found a lot more Hispanics and Asians in the country than they did in 2000, and that forced the BLS to make adjustments in their estimates, as they did with their numbers of people in the workforce.
All these numbers need to be taken with a large dose of salt, as they are subject to large revisions. This past year, the BLS adjusted the employment numbers on a monthly basis—mostly upward, as more jobs were created than they estimated, which is normal for a recovery. In the last recession, they had to go back and adjust the prior numbers downward.
It is simply the result of using models and making estimates. The BLS is very straightforward about how they make their models. You can re-create them if you want to. If you go through that process, you get a better understanding of the extent to which the monthly employment number is just an estimate.
For instance, last month—rather notoriously—the BLS found 42,000 new delivery jobs. No real surprise, as Fedex (FDX) and UPS (UPS) and other delivery companies hire more workers for the holiday season, and as more and more of us shop online. But those are temporary jobs, and the BLS likes to use seasonal adjustments to smooth out such anomalies.
A friend of mine talked with them today, and they said that they recognized the problem and had made adjustments to their models to take into account the new seasonality of holiday hiring. Next December, there will be no surprise of 40,000 temporary jobs showing up in the data. And did they back them out in this release? Yes, but in the revised December data.
If you subtracted 42,000 jobs from last month’s number, the non-farm payroll number would have been close to a loss. What would that have done to the stock market?
But if they used the current, revised data, it would have shown 207,000 new jobs, which is a good number—and much stronger than the first estimate. In fact, the last three months have averaged 200,000 new jobs a month, when we look at the revisions.
And that is the point. These are the best estimates the BLS can come up with. They are very clear about how they go about making the estimates. If you have a better way, then by all means propose it. (In fact, there are a lot of people who do just that. Clearly, they have more time on their hands than I do!)
But anyone who trades on this number is gambling. It can be revised up or down, even years later. I find the preoccupation of the market with that number amusing.
But what is not amusing is the reality that is masked by the joyful response of the stock market to the good news. This was a good employment number, not a great one.
It takes about 125,000 new jobs just to keep up with population growth each month. That means we created roughly 120,000 jobs that helped bring down the unemployment number. The US economy has created almost 3 million jobs in the last two years.
That means we only need another 7 million to get back to where we were in 2007! Even if we reclassify 1 million workers as Hispanic, Asian, or Black, we are still down 7 million jobs.
As I detailed about a year ago, even if we create 250,000 new jobs a month, it will take almost five years to get back to where we were in 2007. That is if we can avoid a recession in the meantime. Such a growth rate would require whole new industries and new types of work, much like computers and technology in the 1980s and 90s. (I think that could happen, but that is a story for another book.)
Is it any wonder that the Conference Board Consumer Sentiment number that came out last Monday dropped precipitously, falling to 61.1 from 64.8 (revised up from 64.5)? The present-situation component led the decline, falling from 46.5 (previously 46.7) to 38.4. The expectations component dropped slightly, from 77 (previously 76.4) to 76.2.
"The decline went against expectations of increasing confidence and is a sign of consumers’ uncertain views of the economic recovery." This in spite of the fact that the employment number was so much better than consensus expectations. Things may be getting statistically better, but we don’t feel all that content.
And while we should enjoy the better employment numbers, we need to take a peek at another, less sanguine, number in the BLS report, and that is wages and income.
There has been little growth in disposable income for five years. But it is worse than that: Government transfer payments have been an increasing share of disposable income since the beginning of 2008. Without that government spending, consumer spending would be much worse than it is.
But then so is the federal deficit. There is no free lunch.
It gets worse. Madeline Schnapps of TrimTabs shot me a note about her frustration with the employment numbers. TrimTabs tracks federal withholding taxes to give them an advance estimate of the employment number. In the past, the more taxes that were withheld, the more jobs there were.
For the past few months, their data has shown fewer jobs than the BLS estimates. I called her late tonight, and she answered (I know, neither of us has a life outside of numbers). She was still mystified. The last time their data was this different from the BLS numbers was in the last recession, when the BLS estimated too many jobs and later went back to revise their numbers, which were then more in line with TrimTabs tax data.
I suggested that the problem may be that even though more people are working, they are making less money and thus paying less in taxes. But that thought is not apparent in the data. Average hourly wages are not down all that much.
But self-employment income is not included in that figure. And many people have been forced into "self-employment," which can mean part-time contract work or consulting, and the drop in income is being missed by the data.
- The 10 Must-Know Market Forces of 2012
- Something Good Brewing at Home
- Don’t Fight the Fed (and Other Advice)
Related Articles on MARKETS
We initiated coverage on drug contract research company Icon PLC (ICLR) in May 2016; in 17 months, t...
Midstates Petroleum (MPO), based in Tulsa, Oklahoma, is a small-cap oil and gas exploration and prod...
While utilities aren’t exactly known for their entrepreneurialism, regulators can prompt them ...