Regression to Trend: Debunking the Alternate CPI
March 25, 2013
by Doug Short
For the past few years we have been updating four market valuation indicators on a monthly basis, one of which is a regression to trend of the S&P Composite, available here. The chart that illustrates the historic pattern of oscillation above and below the regression trendline is based on the inflation-adjusted spliced S&P index popularized by Yale Professor Robert Shiller and now widely used as a long-term gauge of US equities. The inflation adjustment is based on the Consumer Price Index, which dates from 1913, and the earlier Warren and Pearson's price index for the earlier years.
Here is the latest version.
Occasionally in the past we've included a curious version of this data series, one that is adjusted with the Alternate CPI published monthly by John Williams at his Shadowstats.com website. We received a request yesterday from reader Mike to update Shadowstats deflated version of the regression chart. Mike explained that his curiosity was triggered by one of John Mauldin's Outside the Box guest commentaries with the provocative title Is The Government Lying To Us About Inflation? Yes! (a commentary we reprinted here).
In response to Mike's request, we've updated the chart in question.
This Alternate CPI version of the regression analysis suggests that today's market is radically undervalued, comparable to the levels in 1949 and 1982. Our view, of course, is that this perspective is complete nonsense. It is based on an argument that inflation has, for example, averaged 9.5% per year since the turn of the century versus the Consumer Price Index annualized average of 2.7%. The official government series would put the cost of living up 37.9% since New Year's Eve of 1999. The Shadowstats Alternate CPI puts the increase over six times higher at 234.2%.
On a personal note, my wife and I raised a family during the stagflation of the 1970s and early 1980s. The annual average was 8.8% from 1973, the year of the Oil Embargo (when gasoline prices tripled) to the end of 1982, after Chairman Volcker broke the back of inflation by raising the Fed Funds Rate above 20%. I was a personal budget hawk then, the same as I am now. The Shadowstats claim the today's annualized inflation (9.62% in the latest Shadowstats update) is higher than the BLS calculated average of 8.60% in 1973-1974 is a complete delusion, albeit one that many people for various reasons are eager to embrace.
For independent evidence that the Consumer Price Index is a reasonably accurate representation of the prices we pay, see the MIT Billion Prices Project US Daily Index.
But for those of you who remain skeptics of BLS's inflation data and think the Shadowstats Alternate CPI is accurate, let's consider what real GDP would look like if it were deflated with the Alternate CPI. The thumbnail below on the left shows real GDP as published by the Bureau of Economic Analysis, which is adjusted by the BEA's GDP deflator. The one on the right is nominal GDP deflated with the Alternate CPI. That alternate version essentially says the US has been in a devastating recession for the better part of the last twenty years.
I'll close with one more example of why I completely reject the Alternate CPI as a valid metric. Let's consider real median household incomes since 1967, the year of the earliest Census Bureau data. Adjusted for inflation using the government's CPI, the $7,143 median household income has only risen 4% from 1967 to 2011, the latest full year of Census Bureau data. If we make the inflation adjustment using the Shadowstats Alternate CPI, the median household income has declined by 70%.
As I commented when I originally posted this household income chart last year, the Alternate CPI is a completely bizarre outlier. What this deflator is telling us translates into something like this: If we chain the 1967 median household income of $7,143 in 2011 dollars, it would have had the purchasing power back then of $166,683.
On a personal note, my first full-time university faculty job in 1969 paid me $9,000 ($7,500 base plus $1,500 for teaching the summer session). Our household income that year was approximately $13,000, which would have been about 55% above the median $8,389. According to the Shadowstats CPI, our 1969 income had the purchasing power of $275,860 in 2011 dollars. That does not, even with the most rosy colored glasses of memory, remotely resemble the reality of our 1969 financial situation.
Footnote on the S&P Composite : For readers unfamiliar with this index, see this article for some background information.