The Schiller cyclically adjusted P/E ratio, or P/E 10, commonly abbreviated as CAPE, is generally a worthless measure of how cheap or expensive the stock market is. While the reasoning behind the measure -- adjusting for the business cycle -- is sound, it suffers from several serious flaws when implemented. Specifically the measure suffers from a lack of distinct data points, inaccurate comparisons of the stock market in the 1800s and today, and inaccurate data due to changes in accounting methodology.
For reference, below is a chart (taken from multpl.com) of CAPE from 1881 to the present. The value of CAPE today is 26.10 versus a historical average 16.66.
While the chart makes the market seem overvalued, it suffers from some serious flaws.
Lack of data points
Perhaps the biggest flaw in CAPE is that it suffers from a lack of distinct data points. The data for CAPE goes back to 1871 (the first CAPE data point appearing for 1881), so at first you wouldn’t think this would be a big problem. However, CAPE measures 10 years worth of data, which means that each yearly data point for CAPE shares 90% of the same data with both the yearly data point ahead and behind it. In reality, there are only 14.5 distinct data points that make up the data for CAPE. This is in contrast to other valuation measures that often have 100 or more distinct data points that do not overlap. Right off the bat, we do not have sufficient distinct data points to draw any conclusions from.
According to the NBER, we’ve had 33 business cycles since 1854, and of those only 29 would appear in Shiller’s data series. Those business cycles lasted on average 56.4 months or 4.7 years. In the modern era (post WWII) the cycle has lengthened to 68.5 months or 5.7 years on average. Therefore, what we have is 14.5 distinct CAPE data points that are attempting to capture 29 business cycles. It also ends up undervaluing earnings during an expansion because the moving average will keep a contraction in data series for 10 years, while in reality it should have “aged out,” for lack of a better term, in about six years.
1881 is not 2016
The other problem we have is that the business and investing world of 1881 is not even remotely close to what we have now. In 1881 we had no Securities and Exchange Commission (we would have to wait until 1934 for that). One would think there would be vastly different risk premiums for a regulated market and an unregulated market. Up until the last few decades, 401(k) plans were not widespread, so the buyers of stocks in the 1800s were vastly different then the buyers of stocks today. Back in the 1970s, stocks made up 15% of household net worth. Today it’s 30%. Today about 60% of U.S. households have investments in the stock market. I don’t know what it was in 1881, but I can assure you that it was substantially less! The tax rules for dividend income and interest income (thus the relatively attractiveness of stocks versus bonds from some investors) and corporate tax rates have vastly changed over the years. We could fill an entire book going over the history of corporate and individual taxes from 1881 to today.
Additionally the types of companies that made up the market in the 1800s were quite different than today. Back then, a majority of the publicly traded companies were basic materials companies, food companies or utilities. Later on railroads would also make up a large portion of the market. The entire composition and structure of the market has vastly changed from 1881 until today. Today industries that weren’t around in 1881 such as consumer electronics, semiconductors and software (the technology sector) make up the largest market segment. Why would a bunch of data about railroad stocks from the 1900s be comparable to valuing Apple (AAPL, Financial) or Microsoft (MSFT, Financial) today?
Accounting changes
The final major issue with CAPE is that the data used across its data set is not consistent. There have been numerous major changes in accounting rules during the period of CAPE data set. Up until 1887, there wasn’t even a national accounting society. Indeed, in the early days accounting focused on the preparation of balance sheets and most accounting was done on a cash basis rather than the accrual basis used today. The accounting rules for public companies largely remained the Wild West until the passage of the Securities Act of 1933 and the Securities and Exchange Act of 1934, which created the SEC and allowed it to prescribe accounting methods to be used. The first general set of accounting rules wasn’t issued until 1939. Throughout the years, there have been numerous changes and clarifications to accounting rules. Modern accounting methods and rules can probably be traced back to just 1973, and the formation of the Financial Accounting Standards Board and some of the new accounting methods did not appear in accounting textbooks until the early 1980s.
The earnings data from the 1880s up until the World War II era is highly questionable (especially the 1800s era data), and is really not comparable to modern data.
Summary
It’s absolutely ludicrous to think that P/E multiples today will mean revert to a dataset that includes points from the 1880s. Why on earth would stock price multiples today bear any resemblance to an era in which stock markets were largely unregulated and accounting standards could be summed up using the modern slang “YOLO”? Why would a collection of utility companies, railroads and basic materials companies trade at the same multiple as a market that contains a high percentage of high tech information technology companies and patent protected pharmaceutical and medical device companies? Why would the stock market behave the same way in the 1800s when few people owned stocks as it would today when a majority of households own stocks?
Robert Shiller is a smart guy and CAPE is a valiant attempt to do something extremely useful. While CAPE has a lot going for it as a theory, it falls flat on its face when it comes to the real world. The stock market and the economy are just too dynamic and have undergone too many changes to make a valuation metric that includes data from 1881 worthwhile.
