b.s. from the BLS? – Adjustment Wonders [updated]

The ‘un/employment’ numbers for January were released today by the Bureau of Labor Statistics (BLS) to much fanfare from the Associated Press, and even the Wall Street Journal, as can be seen respectively here and here.  The AP gleefully exclaims that “Unemployment rate hits 8.3 pct. after hiring burst” while the gurus over at the WSJ quietly reassure that “Jobs Data Show Sustained Growth.”

Is that so? Well, let’s do a little constitutionalleysis, shall we?

For your convenience I’ve upload my latest handy-Jim-Dandy BLS stats spreadsheet which you can open here. I recommend you do so in a new browser tab as that will make the referencing process much simpler. Note that the spreadsheet consists of four [now five] sheets which can be individually selected by clicking on one of the tabs found near the bottom of your browser window. To start, select the “Historical” tab so as to afford ourselves a bit of perspective regarding the BLS unemployment numbers.

First, notice that the second column is titled “Civilian non-institutional population.” This is the starting point for all of the BLS computations. I’m not exactly sure as to who all is included in the “institutional  population,” but it’s a pretty big number. In July of 2011 the total population of the U.S. was estimated at 311.8 million. As shown in the BLS table, the non-institutional population was at that time 239.7 million, thus the institutional population must have been 72.1 million persons. More on ‘institutional’ vs. ‘non-institutional’ persons later in the program.

To get away from the issue of who is and is not included in the unemployment percentage figure at any given time, I suggest that we use the number “Percent of population not working.” This includes both the “unemployed” as well as persons not in the labor market, that is, actively looking for a job or whatever. This number can be found in the last column of the “Historical” sheet, is derived from the original BLS data, and would seem to be a better measure of economic activity than the narrowly defined BLS unemployment number. For the period shown in the table, the percent-not-working (PNW) has varied from a high of 42.2% in 1983 to a low of 35.6 in 2000. The average rate for 2011 was 41.6%, much closer to the historic high than low.

It’s at this point that the meaningfulness of the BLS stats become increasingly suspect. Is the Dec. ’10 to Dec. ’11 drop in the unemployment rate from 9.4% to 8.5%, nearly a full percentage point, a better indication of economic activity/recovery than the fact than the PNW number remained nearly constant? Dropping only 0.2% from 41.7% to 41.5%? These numbers are also affected by the number of persons counted as ‘in the labor force’ which actually dropped as a percentage of the ‘non-institutional’ population from 64.3% to 64.0%.

So, essentially what’s happened here is that as the non-institutional population grew, the percentage of persons counted as participating in the labor force dropped and the percentage of persons not working remained constant, but the unemployment rate went down by a percentage point. And that’s the number that everyone touts.

The “Jan ’12 orig” tab is “Table A-1. Employment status of the civilian population by sex and age” as it comes directly from the BLS factory. It’s included for reference purposes only. Let’s move on to “Jan ’12 mod”, shall we?

Starting out, it seems advisable to take a peek at the footnote at the bottom. There, BLS is kind enough to inform us that “The population figures are not adjusted for seasonal variation.” But if you look at the numbers, the only one that remains constant is the “Civilian non-institutional population” figure. Huh? Also noted is the advisory that “Updated population controls are introduced annually with the release of January data.” I don’t know about you, but “population controls” sounds a bit too fudgeable to me. All that aside, let’s move onto the results.

The yellow-highlighted rows are the official BLS unemployment rates for each of the given age and sex categories. Seasonal adjustment provides a 0.5% improvement in the overall unemployment figure, but men of all ages, and both sexes aged 16 to 19 years, did better by a full percentage point. Guess women 20 and over just don’t seasonally adjust as well. (Just kidding, ladies.)

Let’s look at the details of the “Men, 20 years and over” seasonal adjustment that resulted in the 0.3% improvement in their unemployment rate from Dec. ’11 to Jan. ’12, rather than what would otherwise be an unadjusted 0.5% increase in that one month span.

So, looking at the last column, for “Men, 20 years and over” the “Not in labor force” category dropped by 377 thousand, while the “Civilian labor force” for this group went up by almost exactly the same number. There were 633 thousand fewer unemployed and 377 thousand new jobs created for this group (1,010 – 633). But wait, isn’t that the same number as the reduction in the “Not in labor force”! Ah, so that’s from whence they came. How convenient.

I’m not so sure about that “Seasonally Adjusted” BLS unemployment number. Are you?


P.S. If you find an error please let me know and I will correct it. Thanks.

Update: Added a fifth sheet to the spreadsheet that is an expanded version of BLS Table A-8. “Employed persons by class of worker and part-time status.” This was done partially in response to Mr. Strzelczyk’s post at American Thinker, “January Unemployment Report: The Devil is in the Details” which is a sentiment I obviously agree with. In addition to the new spreadsheet page, much of the specific information found in the AT posting is available from the BLS here in the “Employment Situation Summary.”

If you wonder why I even bother with any of this nonsense, I do it primarily so that I might find some reasonable level of understand regarding the economic and political but mostly legalistic Kabuki dance that is the United States. After working at this for ten years, I’ve reached the conclusion that we are one of the worst-governed industrialized countries, as individuals occupy a much smaller area of free space than we think we do, and are distracting ourselves into oblivion with our national addictions to money, celebrity and self-indulgence of a licentious nature. But then, nobody’s perfect.

However, politics as usual is not going to change any of it and get us out of this existential mess. Well, I don’t think it will. And at least Mr. Steyn agrees with me.

Update II: You may wish to take a look at Mr. Kaminsky’s post on this subject over at the American Spectator blog. He is of a different opinion regarding the subversive nature of the Bureau of Labor Statistics, but I suspect that to some extent he’s conflated census adjustments with seasonal ones. But, however one looks at the numbers, the “Percent not working” measure doesn’t paint as optimistic a picture as does the Wall Street Journal and quite a few others.

This entry was posted in Uncategorized. Bookmark the permalink.