Counting games: why I started tracking unemployment statistics my own way and why I’m quitting

I woke up yesterday morning, on the day each month that the Bureau of Labor Statistics (BLS) releases unemployment data, thinking I wasn’t going to–as been my habit each month–release my own analysis of the data. I’ve simply got too much reason to believe the data is being manipulated beyond my ability to correct for. I can’t replicate either Gallup’s underemployment survey (showing 19.1 percent underemployed today) or John Williams‘ Shadow Government Statistics alternate unemployment rate (checking in at just under 22 percent), but my feeling is that they’re able to come closer to the truth than I am, and I think I have good reason to deeply distrust even the numbers I have been getting. So I probably am going to discontinue this after this month.

I’m doing it this month only because as I read the many stories on unemployment today, I knew the U3 measure I have named after my cat would show an uptick (as it turns out, to 12.51 percent) in unemployment where everybody else shows a slight downtick (the BLS down to 9.70 percent). Why? I get different results from the BLS because I use the highest labor force participation rate to date as a proportion of current population to estimate the true labor market size. I assume that if that portion of the population worked during peak periods of employment (most recently the dot-com boom), they would work again if sufficient opportunity presented itself. By contrast, the BLS has devised several means of excluding people from the labor market.

Similarly, the Zero Hedge blog recently looked at how reindexing would affect the unemployment rate, concluding that a correct numbers both for labor force participation and for unemployment should be much higher. Their U3 comes close to mine and their U6 looks a lot like John Williams’. And I’m pretty sure that they know better what they’re doing than I do.

I take unemployment numbers personally, because I have been hit hard by every recession since I burned out as a computer programmer in 1985 and because by BLS rules, I would mostly not be counted as unemployed because even when I have been woefully underemployed, BLS methods would still count me as employed; because my own experience of unemployment has seemed far more severe than BLS statistics admit; and now because I have realized I don’t know a way of finding work that makes any sense, I’ve largely given up, so apart from a very tenuous connection to the labor market, the BLS might not count me at all, even in their U6. It’s hard to tell because BLS definitions are complicated. For one thing, they have six different definitions of unemployment (down from–if I follow correctly–a peak of eight). For instance, in a relatively lucid-seeming appendix to their attempt to explain 1994 changes to their methodology, published in October 1995, the BLS states,

The most marked definitional change in the CPS [Current Population Survey] dealt with persons classified as discouraged workers. In the old survey, persons out of the labor force who indicated a desire for work and a job-market-related for not currently looking for work were classified as discouraged workers, provided that no reasons to the contrary were also offered. This definition had been criticized in the 1979 presidential commission review as being too subjective. In the revised CPS, discouraged workers were redefined as persons who indicate explicitly in the survey that they want and are available for a job, have looked for work in the last year, and have given a job-market-related reason for not currently looking for work. [emphasis in original]

The plain English understanding of that definition suggests I should be counted, even without the substitute teaching gigs, but Williams either understands it differently or is relying on a different source: “In ’94, [the Bureau of Labor Statistics] changed the definition [of discouraged workers] so that in order to be discouraged, you had to have not looked for work in the last four weeks, but you had to have worked in the last year” [emphasis added]. By Williams’ understanding, only the five days I’ve worked (unimpressively, at about $100 each) as a substitute teacher since January keep me in the count.

[Maurine] Haver [chair of the statistics committee for the National Association for Business Economics] has little regard for Williams’ argument about discouraged workers. “If someone has done absolutely nothing in one year’s time, I frankly think they’re not very serious about working,” she said. “The U-6 gives you a measure of the very worst in unemployment, period. I don’t think there are any more people who truly are out of a job beyond that, people who won’t get their butt out of bed. If you have done nothing in a year, do you really think that person is a member of the labor force?”

Haver would apparently exclude me from the class of people who “have done nothing in a year” only because I did indeed send out a lot of resumés over the last year that produced absolutely zero results, because I did get an interview (by way of someone I know), and because of those five days of substitute teaching–which to me, in a year of unemployment without benefits, amount to nothing. What also strikes me is that given a rising portion of long-term unemployed being unemployed for longer than at any time since the Great Depression, a year seems an arbitrary and dubious–even capricious–time limit.

Something is clearly amiss when reputable organizations are offering alternative definitions to those used by the BLS, when obfuscatory BLS inclusions and exclusions of various categories of workers and would-be workers evoke the image of a shell game, when Williams has to scrutinize footnotes to figure out who’s being left out, when even publications like Barron’s regularly cite an economist (Williams) that a majority of mainstream economists won’t even discuss, and when so many other people are asking hard questions that never seem to be answered. Plainly, the government is hiding behind those mainstream economists, who like Haver, have an ideological bias in favor of the status quo and who have been complicit in these changes. Kevin Phillips, writing for Harper’s in May 2008 adopts Williams’ phrase, “Pollyanna Creep” to describe shifts in reporting methods:

This apt phrase originated with John Williams, a California-based economic analyst and statistician who “shadows,” as he puts it, the official Washington numbers. In a 2006 interview, Williams noted that although few Americans ever see the fine print, the government “always footnotes the changes and provides all the fine detail. Nonetheless, some of the changes are nothing short of remarkable, and the pattern over time is what I call Pollyanna Creep.” Williams is one of the small group of economists and analysts who have paid any attention to the phenomenon. A few have pointed out the understatement of the Consumer Price Index—the billionaire bond manager Bill Gross has described it as an “haute con job,” and Bloomberg columnist John Wasik has dismissed it as “a testament to the art of spin.” In 2003, a University of Chicago economist named Austan Goolsbee (now a senior economic adviser to Barack Obama’s presidential campaign) published an op-ed in the New York Times pointing out how the government had minimized the depth of the 2001–2002 U.S. recession, having “cooked the books” to misstate and minimize the unemployment numbers. Unfortunately, the critics have tended to train their axes on a single abuse, missing the broad forest of statistical misinformation that has grown up over the past four decades.

Phillips describes how successive presidential administrations have tweaked the methodologies for the major economic statistics by which we are told we should judge the economy. He doesn’t allege a conspiracy but rather “accumulating opportunisms” to, as Williams put it more recently, “protect the incumbents from voters who would surely rise up in anger, if only they knew the truth.” According to Phillips,

In 1994, the Bureau of Labor Statistics redefined the workforce to include only that small percentage of the discouraged who had been seeking work for less than a year. The longer-term discouraged—some 4 million U.S. adults—fell out of the main monthly tally. Some now call them the “hidden unemployed.”

As I said, it’s something I take personally. As of June 30, I will have been off the payroll at CSU East Bay for a year. If Phillips is right, I will be among those “hidden unemployed.” And if I could satisfy myself that the BLS definition of discouraged workers I quoted above wasn’t some kind of doublespeak, I could dismiss what Williams and Phillips say as a misunderstanding. Particularly where a trained (even if controversial) economist as Williams is involved, I can’t.

There is a simpler way of looking at this. Carlton Meyer, of Sanders Research Associates, wrote for Truthout simply that “an honest man would count anyone who would like to work as unemployed.” That clarity must surely weigh heavily in any evaluation of six different BLS measures defined with varying inclusions and exclusions that evoke the image of a shell game. Gallup seems to agree. For its daily survey, the organization says it “classifies Americans as underemployed if they are unemployed or working part-time but wanting full-time work.”

The facts that there is no reasonable way of finding employment; that the BLS may use that as a means to exclude people from those whom it counts; that the BLS can use multiple and complex measures of employment to limit scrutiny to that of a class of people (economists) who generally ideologically support policies of globalization, so-called “free” trade, and capitalism, and therefore seek to rationalize the status quo even when, as Riane Eisler has pointed out in The Real Wealth of Nations, measures of Gross Domestic Product count even destructive actions (such as catastrophic oil spills) as positives for the economy; and that there has been a persistent move to count fewer people as unemployed leads me to doubt anything official sources say about unemployment or, more generally, the economy. Politicians with vested interests in the results have had way too much to say about how the BLS and other economic statistics reporting agencies do their work. That this may affect the data on which I have based my calculations suggests that I may have been mistaken to offer those calculations. And that applies to this month’s result as well.