Palm Beach, FL 2/7/12 (StreetBeat) -- The January reading of the Employment Situation Report was much better than expected. Total non-farm payrolls increased 243,000 during the month; the best result since last April. The unemployment rate fell again and, at 8.3%, is the lowest it has been since February 2009. A quibbler might point out the persistently low employed to population ratio, known as the employment rate, which at 58.5% is just three tenth off the cycle low. And there could rightly be some questions about why the labor force participation rate has fallen to a three decade low. If there is room for debate on the state of the labor market, please feel free to do it amongst yourselves. This essay, however, will have nothing to say about the labor market per se; there will not be a single comment about its relative strength or weakness. What follows is a tedious discussion of one small aspect of the plumbing of the employment report. I will investigate, from a Lilliputian’s eye view, employment report minutia. Importantly I don’t even know if this excruciating examination merits elucidation. It is entirely possible that reading further will be a complete waste of your time, if only because I am not sure that I will have a point to make when all is said and done. Consider yourself warned…back away from the screen and nobody gets bored…
Too late…
The US economy lost 2.689 million jobs in January. You heard what I said, non-farm payrolls dipped by almost 2.7 million last month, and that is a fact, or a reasonable facsimile of a fact anyway, pending future revisions. That is what happens every January. Lots of holiday workers hired in the closing months of the preceding year lose their short term jobs before the Bureau of Labor Statistics (BLS) surveys businesses and households about the labor market early in the month. The stronger than forecast increase of 243,000 non-farm payrolls didn’t actually occur, it is of course the result of mathematical calculation, called a seasonal adjustment, of the unadjusted jobs figure. There are other months during which several hundred thousand people land a job and the seasonal adjustment cuts the other way and shows a payroll number that is drastically below the unadjusted total.
This is not news to any of us. Nobody in the markets give the seasonal adjustment of the jobs data a second thought, it is just the way it is, why bother thinking about it. The seasonals exist to simplify things. “Over the course of a year, the size of the nation’s labor force and the levels of employment and unemployment undergo regularly occurring fluctuations,” says the BLS in a technical note included in the Employment Situation report. “These events may result from seasonal changes in weather, major holidays, and the opening and closing of schools. The effect of such seasonal variation can be very large. Because these seasonal events follow a more or less regular pattern each year, their influence on the level of a series can be tempered by adjusting for regular seasonal variation.”
The job report’s technical note suggests that the bottom line for the BLS is that “the seasonally adjusted figures provide a more useful tool with which to analyze changes in month-to-month economic activity.” Which is great, who doesn’t need a leg up when trying to figure out what is happening in the labor market in particular and the economy in general. Allow the seasonally adjusted data to smooth out the wild fluctuations of the unadjusted figures and it’s easier to spot the anomalies; great idea, widely accepted.
One of the key factors in analysts’ estimates for the January non-farm payrolls was the interesting pattern that has unfolded in recent years in one particular category called “Couriers and messengers” (c/m), a small part of the sector called “Transportation and warehousing”. Analysts noticed that in December 2009 and 2010 the seasonally adjusted increase in c/m jobs was +30.1k and +46.3k, respectively; in December 2011 the BLS reported a jump of 42.2k in these jobs. In the two decades that this data series has been around there had never even been an increase of 14k before now; in most years the December change for this category was well below +10k and a quarter of the time it was negative. So the last few years are an interesting new wrinkle in a long established pattern. The hiring appeared to be holiday related because in these two most recent January reports the c/m category saw employment decline by at least 40k each year. So it logically followed that the January 2012 jobs data would show a similar decline in these jobs. But, as it turns out the c/m payrolls declined by only 1.5k last month. The smaller than expected decline is primarily the result of an annual seasonal adjustment that took the December figure down from +42.2k to -7.5k. The net change of fifty thousand in the seasonally adjusted jobs was all the more interesting when you consider that the unadjusted figure was reduced by only 15k to +70.9k.
That’s quite an adjustment, one that did not seem to be justified by the change to the unadjusted report. But it’s not an outlier. The December 2010 jobs report originally stated that the c/m job category added 3.3k that month. However, one month later, after the annual seasonal adjustment worked its magic the net change for the December 2010 c/m category was +46.1k. As of the latest jobs report the December 2010 c/m results are now said to be -1.8k. It’s all down to the seasonal adjustment, because the unadjusted report, as of now, is only 0.1k different than it was before the forty-six thousand jobs were erased. The guy at the BLS who is in charge of figuring out the “Transporting and warehousing” sector told me that this is just how their special seasonal adjustment software works. They don’t have the funding to do an adjustment of their seasonal adjustment every month, he said, so they just do it once a year, with the January report.
To look back at the history of the seasonally adjusted c/m jobs data today there is not a notable change to the long term pattern. But something new is plainly afoot when you check out the unadjusted data, or if you still have a copy of the pre-January 2012 c/m data.
It is hard to figure out the value of the seasonal adjustments in this segment of the labor market, it seems to have added more complexity rather than simplicity to the analysis of the jobs data. If there has been a structural change to the c/m jobs sector, the result of the intersection of e-commerce and package delivery for instance, there is really no way to determine that by using the seasonally adjusted reports; who could make heads or tails of it with the way it has been twisted out of shape in the last couple of years. Just to review, in the last thirteen months the BLS has told us that the net change in c/m jobs in December 2010 was +3.3k, then +46.1k, +46.3k and now -1.8k; how useful.
And this, I think, begs the question of the amount of randomness that has become embedded in the BLS software, upon which its analysts depend. There you have it, but remember, you were warned.
StreetBeat Disclaimer
No comments:
Post a Comment