Wise men going back at least as far as the great George Orwell have pointed out that statists have little but contempt for objective facts, truth, or reality. In the present day, this contempt is proved most clearly by the parade of false reports and statistical manipulations issued by the Department of Labor. Headed by former congresswoman and current troglodyte Hilda L. Solis, the Labor Department has become — primarily through its Bureau of Labor Statistics (BLS) and Employment and Training Administration (ETA) — the real-world version of the Ministry of Truth. Every report is revised; every revision changes numbers to favor the state.
And reality goes down the memory hole.
Since the first of March alone, the Labor Department has revised three major reports on U.S. labor productivity, job creation and unemployment. There are several official explanations for these revisions; the main one is Solis’ intellectually dishonest commitment to making employment data “seasonally adjusted.”
In theory, seasonal adjustment is supposed to smooth statistical bumps like temporary retail hiring around the Christmas or summer holidays and make it easier chart trends. Fair enough. But, in practice, Solis’ staff of statistical manipulators uses perpetual revision to create the illusion that newer numbers are always an improvement over older ones.
When Labor released these cooked numbers, the mainstream financial press parroted the line that the number was the “lowest in four years.”
I’ve done enough statistical analysis to know that even (or especially) the most honest analyst will occasionally revise results. And, to be sure, Labor isn’t the only federal bureaucracy that does so. But few other agencies do so, so systematically . . . or with such partisan willfulness.
Here’s an example of the revision game from the ETA’s most recent (April 19) report on new unemployment insurance claims:
In the week ending April 14, the advance figure for seasonally adjusted initial claims was 386,000, a decrease of 2,000 from the previous week’s revised figure of 388,000. The 4-week moving average was 374,750, an increase of 5,500 from the previous week’s revised average of 369,250.
See what Solis’ hacks did there? They claimed that the number of initial claims for unemployment benefits was a “decrease” from the previous week’s “revised” number. What they don’t say is that the revision to the previous week’s number was to increase it from the 380,000 they reported at the time to the new number of 388,000. So, they raised the previous week’s number in order to claim the new number was a decrease.
Keeping the four-week moving average for unemployment claims below 375,000 is important because that’s a common inflection point between a rising and falling overall unemployment number. Given the accounting trickery in which Labor engages, a number so close to the inflection point has to be taken with a large grain of salt. The real number may be higher — and overall unemployment may be rising.
Solis’ subordinates play the same sorts of games with the so-called “jobs created” numbers. According to the BLS, “total nonfarm payroll employment” rose by 120,000 jobs in March. This was perceived as a bad number — significantly lower than the jobs created in the previous few months. If then-current numbers held, the trend for the first few months of 2012 was bad, indeed: January, 284,000; February, 227,000; March 122,000.
Focusing only on people who’ve recently been laid off, are actively looking for work, or are applying for unemployment benefits hides the backlog of unemployed people who’ve stopped looking for jobs.
But the Labor staff had a plan for making the downward line on a graph of those numbers less steep. They revised the January and February numbers, moving some of the January jobs into February. The “revised” decline: January, 275,000; February, 240,000; March 122,000.
The net effect isn’t much different; but the optic is better. The graph looks more like a plane gliding to a landing than crashing into flames.
Here’s an even more extreme example, from the April 19 ETA report I quoted above:
The advance number for seasonally adjusted insured unemployment during the week ending April 7 was 3,297,000, an increase of 26,000 from the preceding week’s revised level of 3,271,000. The 4-week moving average was 3,317,750, a decrease of 21,500 from the preceding week’s revised average of 3,339,250.
So, the current number of Americans receiving unemployment benefits was up. But the revised four-week trend was down. How could that be? Solis and crew used revisions to inflate the numbers at the back of the four-week chart.
They had used revisions to similar effect a few weeks earlier, on March 29, when it trumpeted a “decline” in initial unemployment insurance claims. It announced:
Initial jobless claims fell 5,000 in the week ended March 24 to 359,000, the lowest since April 2008 . . .
Later, in the fine print, the agency admitted that it had revised the previous week’s figure to 364,000 from an initially-reported 348,000. So, if not for the revision, the newer number would have increased by more than 10,000 new Americans on the dole.
Also in small type: the Department announced that it had revised weekly data on unemployment claims (and other key indicators) going back to 2007 — in part, to reflect seasonal adjustments. This caused all numbers in 2012 to rise about 4%. And cast the “since April 2008” part of the big announcement into doubt, too.
The guys from Enron went to jail or committed hara kiri because of tricks like this. Instead, when Labor released these cooked numbers, the mainstream financial press — including Bloomberg, Reuters, and the Associated Press — parroted the line that the number was the “lowest in four years.”
The numbers I’ve been considering so far aren’t the government’s formal unemployment numbers. Economists rightly consider the weekly unemployment insurance reports “additional data” and of less reliability than the BLS’ more formal unemployment surveys.
As you might already know, the BLS tracks six different measures of unemployment. These six unemployment statistics are:
- U1: the percentage of the U.S. labor force unemployed 15 weeks or longer;
- U2: the percentage of labor force comprised of people who lost jobs or completed temporary work;
- U3: the “official unemployment rate” — people without jobs who have actively looked for work within the past four weeks;
- U4: U3 plus “discouraged workers,” or those who have stopped looking for work because economic conditions make them believe that no work is available for them;
- U5: U4 plus other “marginally attached workers,” or “loosely attached workers,” or those who “would like” and are able to work, but have not looked for work recently; and
- U6: U5 plus part-time workers who want to work full time, but cannot due to economic reasons (underemployment).
These numbers give economists several different perspectives on the issue. Since all of the numbers are estimates, the combined perspectives are meant to offer a “three-dimensional” view of unemployment, more accurate than any single number. But even these more formal surveys are subject to manipulation and revision. During the Clinton administration, the BLS revised the formal stats — changing the “official” rate to U3 from a predecessor version of U5, which is always a higher number.
There were several problems with this cynical move.
The U3 statistic, with a methodology closest to the “additional data” numbers that Solis’ hacks manipulate these days, is the least reliable of the formal measures.
But there’s a more philosophical problem with the Clinton-era revision: focusing only on people who’ve recently been laid off, are actively looking for work, or are applying for unemployment benefits tends to downplay the number of able-bodied adults who are out of work. It hides the backlog of unemployed people who’ve stopped looking for jobs. A nearly-permanent underclass that includes the nation’s most incorrigibly unproductive people simply doesn’t appear in the U3 number.
Older unemployment numbers are revised upward; new jobs numbers are revised downward. Everything’s always getting better in this workers’ paradise!
So, Solis’ BLS can boast (as it recently did) that “the unemployment rate” fell to 8.2% in March 2012 from 9.1% in August 2011. But that headline ignores the fact that U5 unemployment was about 2 percentage points higher during that period — and U6 hovered above 15%.
This may be what statists want, though: to hide the economic effects . . . and very presence . . . of the hardcore unemployed. Who are, of course, the most enthusiastic supporters of big-government social welfare programs.
Throughout, the operatives continue to use seasonal adjustment to justify their manipulation of employment numbers. Here’s one example of a weasel-worded footnote explaining the spin:
Data in this release reflect the annual benchmark revision of BLS Current Employment Statistics program data on nonfarm employee hours, and revised seasonal adjustment of those data. . . . Quarterly and annual measures . . . for all sectors were revised back to 2007 to incorporate the annual benchmark adjustment and updated information on seasonal trends.
So much jargon, so little truth.
Solis’ hacks cling to seasonal adjustment because it serves as a blank check, an open-ended excuse for revising every employment report to show an illustrious victory for our valiant leaders. Under this administration, essentially every Labor Department employment survey or report that’s been revised has been so in a way that makes newer numbers look like improvements. Older unemployment numbers are revised upward; new jobs numbers are revised downward. So everything’s always getting better in this workers’ paradise!
Or, as one internet commenter noted: “We’ve got the Christmas season, summer season, and — most of all — we’ve got election season.”
Many politicians talk about abolishing the Department of Education; that’s become a kind of short-hand for commitment to limited government. But it might do more good to abolish the Department of Labor, whose truth-twisting under Hilda Solis has become so blatant.