Statistical Deceptions

Paul Craig Roberts, July 28, 2009

Last week on NPR a professor in the Sloan School of Management at MIT explained that what is really at stake in the health care bill is the US government’s ability to borrow. In other words, the bill is about cutting health care costs, not about providing hard-pressed Americans with health care.

The professor said that if we didn’t get health care costs under control, in 30 years the US government would not be able to sell Treasury bonds.

It is not at all clear that the Treasury will be able to sell its debt instruments in 30 months, and it has nothing to do with health care costs. The Treasury debt marketing problem has to do with two back-to-back US fiscal year budgets, each with a $2 trillion deficit. The size of the US deficit exceeds in these troubled times the supply of world savings available to fund the US government’s wars, bailouts and stimulus plans. If the Federal Reserve has to monetize the Treasury’s new borrowings by creating demand deposits for the Treasury (printing money), America’s foreign creditors might flee the dollar.

The professor didn’t seem to know anything about this and gave Washington 30 more years before the proverbial hits the fan.

One looks in vain to the US financial media for accurate economic information. Currently, Wall Street, the White House, and the media are hyping a new sign of economic recovery--« surging » June home sales. John Williams at shadowstats.com predicted this latest reporting deception.

Here is the way Williams explains how statistics can produce false signs of recovery.

The economy has been contracting for so long that a plateauing of the falloff in home sales compared to the previous time period’s more rapid contraction can appear like a gain.

The Census Bureau itself notes that the reported 11% increase in June home sales might be illusory. The reporting agency says that the gain is not statistically meaningful at a 90% confidence interval and that « the Census Bureau does not have sufficient statistical evidence to conclude that the actual change is different from zero. » [PDF]

Williams explains other data distortions likely to create false hopes and lead to investment losses. Financial stresses from the current state of the economy have changed behavior. This means that normal seasonal adjustments to statistical data can result in misleading information.

For example, the recent decline that was reported in seasonally-adjusted new unemployment claims was a result of the normal adjustments for the retooling of auto lines that did not, in fact, take place to the normal extent due to the bankruptcies and uncertainties. Adding in seasonal adjustments that did not in fact take place artificially reduced the unemployment claims.

Williams warns that after a period of contraction, new monthly or quarterly figures are being compared to prior periods of collapsing activity. « Improvements » are thus artifacts of the prior collapse and not signs of economic rebound.

The « Birth-Death Model » is used by the Bureau of Labor Statistics to estimate the net of the non-reported jobs lost by failed businesses (deaths) and new jobs created by start-up companies (births). Williams explains why the model understates job loss during periods of contraction. The modeling on which the birth-death adjustment is based consists primarily of periods of economic growth when there are more non-reported start-up jobs than non-reported job losses from business failures. The BLS model came up with a monthly adjustment of 75,000 new jobs added to the reported number. That means an adjustment factor of 900,000 new jobs added to the reported payroll jobs number each year.

However, during economic contraction, such as the current one, it is wrong to assume that new start-ups are creating 75,000 jobs each month more than are being lost to business failures. Thus the job losses are understated by the 900,000 upside birth-death adjustment and by the absence of a downside adjustment to estimate the jobs lost as a result of failed companies that cease to report.

The reported unemployment rate is itself deceptive as it no longer includes discouraged workers who have been unemployed for more than a year. These long-term discouraged workers are simply erased from the rolls of the unemployed.

The Consumer Price Index no longer measures a constant standard of living and is not comparable to pre-Clinton periods. During the 1990s, the CPI ceased to be based on a weighted fixed assortment. The principle of substitution was introduced. For example, under the old measure, if the price of steak rose, the CPI rose. Under the new measure, if the price of steak rises, the index switches to hamburger on the assumption that consumers substitute hamburger for steak.

Consumer confidence typically is swayed by « good news » hype. The drops in the Conference Board’s and the University of Michigan’s measures of consumer confidence in July suggest that Americans are becoming inured to recovery hype and are realizing that the government and the media lie about the economy just as they lie about everything else.