Why PE10?

Many people have been convinced that there is no meaningful way to predict stock market returns. Don’t fall into that trap.

I have taken this article from Professor Robert Shiller’s web site and edited it heavily. The article makes it clear that you can predict stock market returns in the intermediate-term. Professor Shiller used P/E30 in this article. Today, he uses P/E10.

P/E30 is the current price (index value) of the S&P500 divided by the average of the most recent thirty years of (trailing) earnings. P/E10 is the same as P/E30 except that it uses the most recent ten years of (trailing) earnings. All terms include adjustments for inflation.

These are statistical predictions. There is always the possibility of an unusual outcome. The bubble was such an outcome.

P/E10 is not the only indicator of value. There are many. P/E10 is the best measure that we have found for the purpose of calculating Safe Withdrawal Rates. Actually, we have found that its reciprocal, the percentage earnings yield 100/[P/E10], is the best measure.

This relates nicely with theory. The price of a company should be related to the amount of money that it returns to its owners. Rationally, this would be related to its dividends (and/or the equivalent income that it offers an acquiring company). Smoothed earnings turn out to be a better indicator than current dividend yields 1) because smoothed earnings are buffered from unpleasant surprises such as dividend cuts and 2) because dividends come out of earnings.

Price–Earnings Ratios as Forecasters of Returns: The Stock Market Outlook in 1996

by Robert J. Shiller

(and edited by John Walter Russell).

Professor Shiller's web site
Professor Shiller's Online Papers
Professor Shiller's P/E Ratio Paper

The theory that the stock market is approximately a random walk does not look right at all….If real stock prices were a random walk, they should be unforecastable….Looking at the diagram, it is hard to come away without a feeling that the market is quite likely to decline substantially in value over the succeeding ten years; it appears that long run investors should stay out of the market for the next decade.

Is this conclusion right? How can we reconcile it with the widespread public impression that the random walk hypothesis is at least approximately true?

Ratios as Indicators of Market Overpricing

The scatter diagram..is unusual, in that the measures shown on both axes relate to the long run. Ratios of stock market indices to measures of fundamental value (such as earnings) as indicators of the outlook for the market appear to be most useful when they relate properly to the long run….

The simplest and most widely used ratio used to predict the market is the price–earnings ratio. The use of one-year's earnings in the price–earnings ratio is an unfortunate convention, recommended by tradition and convenience rather than any logic. As long ago as 1934, Benjamin Graham and David Dodd, in their now famous textbook Security Analysis, said that for purposes of examining such ratios, one should use an average of earnings of "not less than five years, preferably seven or ten years." (p. 452) Earnings in any one year tend to be affected by short-run considerations, that cannot be expected to continue. In the present time, earnings have suddenly shot up in the last few years, bringing price–earnings ratios down dramatically, but it is doubtful that such sudden changes are meaningful….

We chose to represent long-horizon returns, of ten years, since that is what really matters to most investors, because there is so much interest today in long-term investing, and because there is recent evidence in the statistical literature that the long-horizon returns are more forecastable. This may be contrary to one's expectations; one might have thought that it is easier to forecast into the near future than into the distant future, but the data contradict such intuition. This forecastability of the market is not the kind of thing that will enable us to forecast that a crash is around the corner; it is forecasting gradual trends, analogous to forecasting the prospects for a city based on population trends, or forecasting the success of a university in terms of the number of young people who are enrolling.

Note that the apparent predictive relation is not really an artifact of the 1929 crash, as some might suspect….Neither is the 1987 crash of much importance to these results….

Figure 2 shows..the real (inflation corrected) gross return on the Standard and Poor Composite Stock Price index versus the same ratio of real price to the 30-year average of lagged real earnings. On this diagram, the relation looks even more striking….The reason for the better fit in this relation is returns are affected by the price–earnings ratio in two ways: by the effect on subsequent price changes..and also by their effect on dividend yields. Times of very high price earnings ratios tend to be times of low dividend yields. The low dividend yield in such circumstances tends to persist for years, thereby contributing further to the low returns...

Why Long Horizon Returns?

There is some popular confusion about..predictability in forecasting long-horizon returns….A related confusion concerns the apparent random-walk property of one-year returns. How, some will ask, can it be that one-year returns are so apparently random, and yet ten-year returns are mostly forecastable?..In looking at one-year returns, one sees a lot of noise, but over longer time intervals this noise effectively averages out, and is less important.

Warnings About the Above Analysis

The conclusion of this paper that the stock market is expected to decline over the next ten ears and to earn a total return of just about nothing has to be interpreted with great caution.

Have fun.

John Walter Russell
August 7, 2005