The Absurdity of ShadowStats Inflation Estimates
Lately I’ve seen John Williams and his “ShadowStats” website referenced repeatedly as if it’s a legitimate source for inflation data that’s in some way comparable to the data provided by the United States government. It’s not. And it’s very easy to see why.
The historical record provides an irrefutable set of data with which to determine inflation; we have actual sampled price data going back a number of years. But we don’t even need that. Each of us has recollections of what things cost at some point in the past. To assess the relative accuracy of official data and Williams’ data, we can discount today’s prices by the inflation rate claimed by each to determine what each says prices were at some time in the past. Simple enough, right?
However, Williams’ data are proprietary, and he generally charges for access to them. I’ll be damned if I’m paying for faulty data just to prove how wrong they are. Luckily, Williams has allowed use of his data in the inflation calculator at http://www.halfhill.com/inflation.html, so I’ve derived his data and can use this tool to prove the point.
The tool delivers cumulative inflation data based on beginning and end dates. Here are three comparison periods using CPI-U, the widely-used measure produced by the Bureau of Labor Statistics, and ShadowStats’ Alternate:
- For 2005-2010, CPI-U says cumulative inflation was 13.4%, ShadowStats claims it was 60%.
- For 2000-2010, CPI-U says 28.7%, Williams says 144.7%.
- For 1990-2010, CPI-U says 73%, Williams says 378.9%.
So lets go back 10 years. CPI-U indicates that what cost $100 in 2010 cost $77.71 in 2000. According to ShadowStats, what cost $100 last year cost just $40.86 in 2000.
Pretty dramatic difference. At least one of these figures is way off. Here’s where each person’s recollections come in (I haven’t bothered with an analysis of sample data since this giggle test is pretty damning). BLS says the price level’s increased by 29% since 2000, while Williams claims inflation of 145%. In 2000, was today’s $4 gallon of milk closer to $3.10 or $1.65? Was today’s $30,000 car closer to $23,300 or $12,250? Aside from gasoline and perhaps a house, there’s not much that’s debatable here.
Now let’s go back further. Changing the target to 1990, BLS says inflation over the last 20 years has been 73%. ShadowStats claims prices have increased 379%. John Williams apparently believes the price level is almost 5 times higher today than in 1990.
Can you think of anything that costs five times more today than it did in 1990? Examples:
- Did a gallon of milk cost $0.80?
- Did a pint of Ben & Jerry’s cost less than a dollar?
- Did a 12-pack of Coke cost less than a dollar?
- Did a case of Budweiser cost $5?
- Could you get your shirts washed for a quarter?
- Did Levi’s cost less than $10?
- Was a Big Mac, fries, and a drink on the dollar menu – combined?
- Did decent running shoes cost $20?
- Did an entry-level Lexus cost less than $10,000?
Not even gas, housing, or college tuition comes close.
Meanwhile, the BLS equivalents seem to be in the ballpark. Especially compared to ShadowStats.
Williams’ incentive to continue generating a bogus data set is clear: he sells his data to a self-selecting audience who might not pay if reality were reflected accurately.
But why he continues to be mentioned in the media as a source to take seriously, when his data are so obviously and significantly removed from reality, I have no idea.