The Numbers Racket

Harper's Magazine, May 2008: Numbers racket | Why the economy is worse than we know:

Transparency is the hallmark of democracy, but we now find ourselves with economic statistics every bit as opaque—and as vulnerable to double- dealing—as a subprime CDO. Of the “big three” statistics, let us start with unemployment. Most of the people tired of looking for work, as mentioned above, are no longer counted in the workforce, though they do still show up in one of the auxiliary unemployment numbers. The BLS has six different regular jobless measurements—U-1, U-2, U-3 (the one routinely cited), U-4, U-5, and U-6. In January 2008, the U-4 to U-6 series produced unemployment numbers ranging from 5.2 percent to 9.0 percent, all above the “official” number. The series nearest to real-world conditions is, not surprisingly, the highest: U-6, which includes part-timers looking for full-time employment as well as other members of the “marginally attached,” a new catchall meaning those not looking for a job but who say they want one. Yet this does not even include the Americans who (as Austan Goolsbee puts it) have been “bought off the unemployment rolls” by government programs such as Social Security disability, whose recipients are classified as outside the labor force.



Second is the Gross Domestic Product, which in itself represents something of a fudge: federal economists used the Gross National Product until 1991, when rising U.S. international debt costs made the narrower GDP assessment more palatable. The GDP has been subject to many further fiddles, the most manipulatable of which are the adjustments made for the presumed starting up and ending of businesses (the “birth/death of businesses” equation) and the amounts that the Bureau of Economic Analysis “imputes” to nationwide personal income data (known as phantom income boosters, or imputations; for example, the imputed income from living in one’s own home, or the benefit one receives from a free checking account, or the value of employer-paid health- and life-insurance premiums). During 2007, believe it or not, imputed income accounted for some 15 percent of GDP. John Williams, the economic statistician, is briskly contemptuous of GDP numbers over the past quarter century. “Upward growth biases built into GDP modeling since the early 1980s have rendered this important series nearly worthless,” he wrote in 2004. “[T]he recessions of 1990/1991 and 2001 were much longer and deeper than currently reported [and] lesser downturns in 1986 and 1995 were missed completely.”



Nothing, however, can match the tortured evolution of the third key number, the somewhat misnamed Consumer Price Index. Government economists themselves admit that the revisions during the Clinton years worked to reduce the current inflation figures by more than a percentage point, but the overall distortion has been considerably more severe. Just the 1983 manipulation, which substituted “owner equivalent rent” for home-ownership costs, served to understate or reduce inflation during the recent housing boom by 3 to 4 percentage points. Moreover, since the 1990s, the CPI has been subjected to three other adjustments, all downward and all dubious: product substitution (if flank steak gets too expensive, people are assumed to shift to hamburger, but nobody is assumed to move up to filet mignon), geometric weighting (goods and services in which costs are rising most rapidly get a lower weighting for a presumed reduction in consumption), and, most bizarrely, hedonic adjustment, an unusual computation by which additional quality is attributed to a product or service.