April 13, 2015

Even the most well-meaning and rigorous attempt to come up with a single number (a price index) that reflects the change in the purchasing power (PP) of money is bound to fail. The main reason is that disparate items cannot be added together and/or averaged to arrive at a sensible result. For example, in one transaction a dollar might buy one potato, in another transaction it might buy 1/30,000 of a new car and in a third transaction it might buy 1/200 of a medical checkup. What’s the average of one potato, one-thirty-thousandth of a new car and one-two-hundredth of a medical checkup? The creators of price indices claim to know the answer, but obviously there is no sensible answer. However, in this post I’m going to ignore the conceptual problem with price indices and briefly explore the question: Which is probably closer to reality — the official CPI or the CPI calculated by Shadowstats.com?

Unlike many other members of the ‘sound money camp’, I’ve never been a fan of the Shadowstats CPI and I’ve only ever mentioned it in TSI commentaries to note that it is just as bogus as the official CPI. It always seemed to me that the Shadowstats number was derived by adding an approximately constant fudge-factor to the official (bogus) CPI to essentially arrive at another bogus number that, regardless of the message being sent by real-world experience, was always much larger than the official number. As illustrated by the following chart from the Shadowstats web site, since the late-1990s the growth-rate difference between the official and Shadowstats CPIs has consistently been about 7%/year.

From my perspective the Shadowstats CPI never appeared to be doing a better job than the official number of reflecting the dollar’s change in purchasing power. I therefore never paid any attention to it and never bothered to analyse why, given that the only differences between the Shadowstats calculation and the official calculation were the changes in calculation methodology that were implemented by the BLS (Bureau of Labor Statistics) since the early-1980s, there would be such a big difference between the official and the Shadowstats numbers. However, Ed Dolan has recently taken the time to analyse and explain the difference in a 31st March article at EconoMonitor.com.

The above-linked article starts by comparing the price changes of similar items that actually took place between 1980 and 2014 to show that the official CPI appears to under-estimate the change in the dollar’s PP and that the Shadowstats CPI appears to over-estimate the change in the dollar’s PP, with the magnitude of the Shadowstats over-estimation being vastly greater than the official under-estimation. It goes on to show that using the Shadowstats CPI to convert nominal GDP to real GDP leads to nonsensical results. For example, according to the real GDP calculation based on the Shadowstats CPI, the output of the US economy is no higher today than it was in 1990. This is a patently false result. Lastly, it attempts to answer the question: Has John Williams, the proprietor of Shadowstats.com, simply made a calculation error?

The answer, apparently, is yes. It seems that in the calculation of the Shadowstats CPI the effects of the same change to the official methodology are counted multiple times. Consequently, the rate of CPI growth estimated by Shadowstats has consistently been at least 4.5%/year too high over the past 15 years, even by Shadowstats’ own methodology.

I’ll be very interested to see whether John Williams can explain-away the apparent multiple-counting of the same BLS changes and, if not, whether the Shadowstats calculations are revised to remove this major error.

Print This Post Print This Post