Lies and Damned Government Statistics

Essay, Economy, Lee Harding

“This is the best number I’ve ever seen in my life!” Jim Cramer told CNBC. The Bureau of Labour Statistics (BLS) had just reported a 3.5 percent unemployment figure in the United States. The November 2019 percentage was the lowest in fifty years. Others would say that might not be the case and that such figures misrepresent reality. Economists and other informed observers have more than a little evidence to back their claim. The stakes are high—so what is the truth?

That’s a question John Williams has tried to answer. The economic consultant graduated with his MBA in 1972 and has had a career that spans decades.1 One of his early clients was a large airplane manufacturer that found its econometric model to predict revenue passenger miles had lost its relevance. The model relied on the Gross National Product (GNP), and Williams could only solve the problem when he discovered flaws in the government’s calculation of GNP. He made adjustments that worked for a while—until the government changed its GNP calculations yet again.

Williams began to question not just GNP calculations, but all of the government’s economic statistics. He asked business economists to assess their quality, and most agreed it wasn’t very good. He interviewed key people involved in government reporting, past and present. In time, he had joint meetings with representatives of the statistical agencies and even testified to Congress. 

By now, Williams has made more than 1,000 presentations to government, business, the press, academia, banks, and investors. His conclusion is that the quality of government reporting has deteriorated “out of the realm of real-world or common experience.”2

Stephen D. Simpson agrees. His 2019 article, “5 Government Statistics You Can’t Trust”3 puts unemployment, inflation, gross domestic product, and deficit accounting on that list. He says, “The accuracy of reported economic data is a problem in virtually every country.” Sometimes this is intentional, as “countries engage in blatant manipulation to influence their obligations, manipulate markets (equity, bond and exchange) or influence capital flows.”

“If You Want To Know The Real Rate Of Inflation, Don’t Bother With The CPI.”4 Perianne Boring told Forbes readers in a 2014 article. The raw data the Bureau of Labour Statistics uses is unavailable. When she asked them why, she was told, “so companies can’t compare prices.” She refused to believe it, since such prices were available by a simple internet search.

Boring says the CPI didn’t jive with other government statistics either. The M2 money supply rose 4.9 percent in 2012, yet BLS said the cost of goods only went up 1.5 percent. The Department of Agriculture noted that beef prices rose 26 percent over the previous five years, but the BLS said beef and veal only rose 20 percent. When she asked a statistician at the BLS about the discrepancy, he could only reply, “I would expect those numbers to be a little closer together.”

Between 1979 and 1999, the BLS made more than 20 changes to the way it calculated inflation.5 Williams says the consumer price index has not reflected actual out-of-pocket costs since the early 1980s, when hedonic quality adjustments were introduced. 

The premise, as Simpson explains, is that, “at least some of the price difference between a good bought today and a good bought yesterday can be ascribed to significant quality improvements. Unfortunately, this is a highly subjective determination and one that does not always sync with reality.”

Another major change came in the 1990s when the government stopped measuring the price changes of a fixed basket of goods, and replaced it with a substitution-based basket of goods. The idea here is that when times get tough, a person might stop eating steak and have hamburger instead. Are they eating just as well as before? To you and I, no; but to the government, yes!6

Why would the government use these concepts to erode inflation calculations? Quite possibly because tax brackets, social programs, and some pension programs are indexed for inflation. A government that can lower those inflation numbers can get more and give less without the public recognizing it. Speaker of the House Newt Gingrich and Federal Reserve Chairman Alan Greenspan pushed for such changes in the ‘90s.7 They got their wish.

Williams laments, “Without the changes made to CPI calculations of the last several decades, social security payments would be more than double what they are today. Indeed, with the use of a substitution-based index . . . the resulting cost of living adjustments promise only a declining standard of living.”8

Lower inflation figures also mean that “real” (versus nominal) GDP figures will be skewed upwards. Every government wants to tell the public that the economy is burgeoning under their direction, and warped stats help them when reality can’t. Williams found the post 9-11 GDP stats to be impossibly optimistic. He concluded, “the GDP is heavily modeled, imputed, theorized and gimmicked.”9

As for employment, is the 3.5 percent unemployment figure accurate that Jim Cramer was raving about? And is today’s the best since 1969? The best answers to those questions are “sort of” and “probably,” and take some explanation to flesh out.

Starting in the 1960s, BLS unemployment figures began to exclude “discouraged workers.” These are defined as people who want a job but don’t have one and haven’t looked for one lately. (Canada has a similar methodology.)10

Under the U.3 data set (the main one used for employment), a discouraged worker hasn’t actively sought a job for the past four weeks. The U.6 data set (launched in 1994), includes unemployed people who tried to find a job within the previous twelve months.

To the common person, if a person is unemployed, they’re unemployed. But in the government stats, if someone hasn’t looked for a job, they don’t even exist. That difference is the main reason why a government announcement of 3.5 percent unemployment figure rings hollow—whether it’s 1969 or 2019. 

If the unemployment figure is not 3.5 percent, then what is it? The Labor Force Statistics from The Current Population Survey11 from November 2019 put the unemployment rate (U.3) at 3.3 percent and the U.6. at 6.5. Seasonal adjustments shifted these numbers to 3.5 and 6.9, respectively. 

That U.6 number of 6.9 is closer to the truth, but still ignores those people who are available to work but haven’t looked for a job in more than twelve months. Williams’ shadow stats claim if those people are included, real unemployment at 20.9 percent—the best in ten years, but not fifty.

Is Williams correct? Probably not. In a 2015 article, Ed Dolan examined Williams’ figures and found them less believable than the government’s. Dolan found that a careful examination of BLS survey data would render the unemployed people not counted by U.6. At that time, the number was 3.7 million—just one-seventh of the 26 million implied by Williams’ calculations.12 

What Dolan suggests is a new category called U.7. This would include everyone who is available for work but doesn’t have one. At the time Dolan proposed this, U.7 would have been 12.9 percent. That was 2.5 percent more than U.6, but ten points less than Williams’ alternative employment rate.

Percentages aside, can we at least say that because the “discouraged worker” category has existed for many decades, November 2019 really did give the U.S. its lowest unemployment rates in 50 years? Probably, but there’s a twist.

In the middle of that time, the BLS changed its employment survey question. Formerly, the BLS asked respondents if they currently wanted a job. But, starting in 1994, respondents were asked if they were available to work. 

Williams believes that the data collection was changed to hide job losses due to the North American Free Trade Agreement, but that is far from clear. The actual results were documented by Monica D. Castillo in her 1998 paper, “Persons Outside the Labour Force Who Want a Job.”

In addition to reducing the number of discouraged workers by half, the new measure of such workers, introduced in 1994, had the anticipated effect of showing somewhat stronger ties to the labor force than had been demonstrated under the old definition. . . [But] discouraged workers, even under the new, more rigorous definition, find it difficult to translate their desire for work into subsequent employment or even an active job search effort.13

Sir Charles Dilke once said, “There are three kinds of lies: lies, damned lies, and statistics.” Keeping accurate statistics regarding an entire nation is extraordinarily difficult. Furthermore, valid arguments can at least be proposed, if not upheld, that statistical adjustments could be useful to create a more accurate picture. However, these should be kept to a minimum. Given the billions of dollars and political agendas at stake, and ease with which hidden methodologies can obscure the truth, citizens should always take government stats with a measure of reserve. If someone’s gut tells them that things are worse than the government says, that hunch might well be right.

 

 

Lee Harding writes for the Frontier Centre for Public Policy.

”SeeEndnotes”

Endnotes:

  1. http://www.shadowstats.com/
  2. http://www.shadowstats.com/article/no-438-public-comment-on-inflation-measurement.
  3. https://www.investopedia.com/financial-edge/0411/5-government-statistics-you-cant-trust.aspx.
  4. https://www.forbes.com/sites/perianneboring/2014/02/03/if-you-want-to-know-the-real-rate-of-inflation-dont-bother-with-the-cpi/#b365ee2200b4.
  5. http://bls.gov/opub/mlr/1999/06/art4full.pdf.
  6. In addition, Williams writes, 24 percent of the CPI inflation reflects a category of “homeowners’ equivalent rent of residences.” Before 1983, the BLS calculation would accurately reflect home prices. Since then, the BLS envisions what homeowners would pay themselves to rent their own home. It’s yet another statistical warp.
  7. Ibid. pg,  4.
  8. Ibid.
  9. Ibid.
  10. At https://www150.statcan.gc.ca/n1/pub/75-005-m/75-005-m2015002-eng.htm, the organization says, “The fundamental concepts and methods used by Statistics Canada and the US Bureau of Labor Statistics are quite similar. Both countries use internationally recognized methods and indicators to measure the performance of the labour market. In some cases they integrate information obtained from surveys with information from administrative data files on individuals and on businesses.

“Studies have examined to what extent conceptual differences affect comparisons between the two countries. For example, at the national level, on average between 2007 and 2013, adjusting the Canadian unemployment rate to American concepts lowered the Canadian rate by about one percentage point.”

  1. https://www.bls.gov/web/empsit/cpseea38.htm, viewed December 10, 2019.
  2. In May of 2015, Dolan found that number to be 3.7 million, not the 26 million required in Williams’ numbers. Dolan talked to Williams and found his methodology somewhat opaque. But from the data, Dolan extrapolated that Williams only counted a low of 7 per cent attrition for LTDWs due to “death, entry into the labor force, or loss of the desire for a job.” Dolan says that statistics gathered by the BLS in 1995 and mentioned in Castillo’s study suggest that the attrition rate is actually 91.4 per cent. Dolan also cited a private sector poll (available at http://kff.org/report-section/kaiser-family-foundationnew-york-timescbs-news-non-employed-poll-methodology/), conducted by the Kaiser Family Foundation, the New York Times, and CBS News late in 2014, polled U.S. residents aged 25 to 54. The results of this survey suggested that those who wanted to work but hadn’t looked for a job in more than a year could be as high as 9 million, and of these a somewhat smaller number would be available for work.

13. http://www.bls.gov/mlr/1998/07/art3full.pdf, 41.

”Close