Seeking Alpha
Long/short equity, deep value, registered investment advisor, CFA
Profile| Send Message|
( followers)  

While you're reading for the nth time this year that "the job market is finally/really gaining traction," and how strong the recovery is at last, you might want to consider some of the unusual behavior in the employment data first, along with a couple of other items that paint a different picture than some of the enthusiasm on the airwaves.

First let's address the subject of the title. A couple of days ago, I wrote that the jobs report might see a print between 235,000 and 245,000 for November. The reason for this, I noted, was that the monthly year-on-year growth rates in 2013 payrolls had run extraordinarily close to those in 2012, a topic I have written about before.

In particular, I noticed that the year-to-date increases in unadjusted, total non-farm payrolls from the establishment survey had converged with 2012 in the last three months, something I partly attribute to the preliminary benchmarking revision (i.e., enhanced data fitting). The box below shows that convergence.

Total Non-Farm Payrolls

YTD chg

August

September

October

2012

0.35%

0.81%

1.46%

2013

0.34%

0.79%

1.46%

The October convergence especially caught my eye - after 10 months, the rate of increase was identical. How often does that happen? Well, it's happened the last three years - October was 1.47% in 2011. I have only calculated these comparisons going back to 1981, but nothing else comes close to this remarkable convergence.

To arrive at my November estimate, I simply took the 2012 year-to-date increase through November (1.76%) and applied it to the most recent October number. That was an accurate guess, because the latest numbers from BLS show - before revision - that the November 2013 year-to-date increase is 1.77%. Remarkable.

To get from there to the 235-245K seasonally adjusted (SA) range, I used the October year-year SA growth rate (1.74%) as a scaling factor, but it ended up being 1.71%, the same as September. I had guessed that the higher number might signify a step up from the preliminary benchmarking results that added a lot of jobs into the seasonally adjusted columns the previous month; perhaps it was a one-off effect the BLS used to compensate for the shutdown.

I may disappoint some at this point by saying that I nevertheless don't believe the BLS tries to cook the books. While I do believe that every administration tries to put the best face on jobs data, the Labor Department has to work with limited sample data and imperfect methods. For all the attention the payroll number gets, many veteran traders don't consider it to be all that reliable (except as proxies to trade around). Not every agency worker is perfect, of course, nor are they anywhere else - the latest bout of indignation making the rounds over possible fabrication is just the usual political hay. But it does seem that the agency is struggling mightily not to concede making another substantial upward adjustment in its annual revision - perhaps too mightily.

The unemployment rate is another matter. It's derived from the household survey, which extrapolates from a smaller, less reliable sample and is therefore known for bigger revisions. The October household number was a loss of 735,000 jobs, a comically low number that still stands, while the November number was a gain of 818,000 jobs, a comically high number that many enthusiasts felt free to boast about. It seems fairly obvious that the government layoffs distorted the small-sample count - the BLS said as much - and that combining the two months for a net gain of 83,000 ought to be far closer to reality than either of the two outliers.

That leaves us with the question not raised amid the enthusiastic jubilation I heard on radio and television the rest of the day - how did a net gain of 83,000 jobs over two months lower the unemployment rate from 7.2% in September to 7.0% in November? We're supposed to be doing 125K-150K every month just to stay flat.

The alternative U-6 rate also fell from 13.6% to 13.2% over the same two months - on 83,000 jobs? [I heard some "economist" exclaim that the drop from 13.8% in October to 13.2% - the result of the big government add-back, of course - was to be taken as some huge sign of strength, being the largest ever, or in 30 years or something - my expostulation drowned out the tail end of a statement that was either disingenuous or dishonest in the extreme].

It seems that the civilian labor force (CLF) - the denominator in the simple division process that calculates the unemployment rate - is actually smaller than it was year ago, or at least on the seasonally adjusted basis used to calculate the headline number. The "eligible" (not confined by law or disease) population grew by just under 2.4 million, but the "Not-in-labor-force" (NILF) designation grew by just over 2.4 million. You might object that a smaller denominator places upward pressure on the rate, but that's true only so far as the difference is excluded from the unemployment total. If we add half of the NILF total to the labor force and put the other half in the unemployed category, the rate would rise to 7.7%.

The weakness in the rate calculation can be seen in the chart below, which compares growth in the civilian labor force to NILF. The growth rate in the NILF category since the recession is spectacular.

(click to enlarge)

Before you reply that it's all demographics, it isn't just a matter of the baby boom retiring. At first glance it might seem that way, because the CLF starts to decline in late 2009. Since the baby boom started in 1946, the answer is easy - those born in '46 would be 65 in 2011, and some could have started taking social security (or disability) early. Problem solved.

It's a convenient answer, but a lazy one. The "baby boom" is justly named because of the surge in the birth rate, but the absolute numbers of people born in the 1980s and early 1990s are comparable. For example, there were 3.6 million births in 1950 and 3.6 million births in 1980, and a lot more people from 1980 are still around compared to 1950. Even if everyone in 1980 went to grad school and got a PhD, they're done by now.

If you haven't told the Labor Department you're looking for work in the last four weeks, you're just not counted in the labor force. But if you can't collect unemployment, there is no point in heading down to the Labor Department and waiting in line to tell them so. In other words, if you're unemployed but can't collect benefits, either because they're run out or you're ineligible for some other reason (like being an "independent contractor" who hasn't paid in enough) you're not really unemployed. You're Not in the Labor Force.

I'm sure it must be comforting to be in the fastest growing population category in the BLS: year-on-year eligible population grew by about 1%, year-on-year establishment payrolls grew by 1.7%, year-on-year household survey employment only grew by 0.8% (another sign that the job market is "really improving") and the civilian labor force didn't grow at all. But our beloved NILF is the champion: it grew by 2.6%, and that seems to keep economists happy, the administration happy, the Fed guardedly happy, the stock market happy, in fact just about everyone happy but the people in the NILF and the stores that can't sell them anything.

No doubt there are many explanations out there ready to do battle, but consider this: according to the latest data from the Bureau of Economic Analysis, year-on-year personal income growth through October stands at 3.4% (nominal), while four-quarter nominal GDP growth is at 3.3%, headed lower this quarter. A year ago, year-year personal income growth was 4.0%, four-quarter GDP was 4.8%, and yet since then the unemployment rate has fallen from 7.8% to 7.0%. There are other considerations, certainly, notably the sequester and the payroll tax, but we are getting close to having a full percent drop in unemployment with nominal GDP falling by almost the same amount. It's hard to escape the conclusions that the growth in NILF matters, and the fall in the unemployment rate is not all it appears to be.

Source: The Case Of The Identical Job Numbers