by Rob Bennett
Last week’s column argued that stock prices do not fall in the pattern of a random walk. Each time we have seen the P/E10 value rise to a dangerously high level (a P/E10 value of 24), we have seen a dramatic price drop within 10 years or so. Eventually prices fall to insanely low levels. Then they begin working their back upward again.
This pattern has played itself out four times now and there is not one time-period in the historical record in which stock prices played out according to some other pattern or according to no discernible pattern at all. The entire historical record shows that long-term stock prices are not random but cyclical.
I want to present more evidence.
We need to get this right. If stock prices really do play out in the pattern of a random walk, Buy-and-Hold is the ideal investing strategy. If stock prices are random, they are unpredictable. If stock prices are unpredictable, then the only way to obtain the high returns offered by stocks is to remain highly invested in them at all times.
However, if stock prices play out pursuant to discernible, repeating patterns, stock prices are predictable. To the extent that stock prices are predictable, it is possible to limit one’s participation in the stock market to times when long-term outcomes are likely to be appealing. That is, to the extent stock prices are cyclical rather than random, stock risk is optional and can be avoided by investors wishing to obtain the high returns associated with stock investing without taking on the high levels of risk commonly thought to be part of the stock investing experience.
The earlier column showed that the entire historical record indicates that stock returns are cyclical rather than random. What more proof could there be?
We cannot look at more data since we have already looked at the entire historical record. Still, we can justify a greater sense of confidence in our conclusions by looking at that same data from a different perspective.
Please take a look at The Stock-Return Predictor. This calculator uses a regression analysis of the historical return data to reveal the most likely 10-year return starting from any possible valuation level. If prices followed the pattern of a random walk, the likely return would be the same starting from any possible P/E10 value. The fact that the Predictor identifies different returns as most likely for index-fund purchases made at different valuation levels confirms the point made in last week’s column -- stock prices are not a random walk.
The calculator proves the point in a second way. Please look at the range of possible returns at 10 years out and compare the range of possible returns that applies at 20 years out and at 30 years out.
Do you see the change that takes place as the holding period for your index fund grows larger?
At 10 years out, the range of possible returns is broad. The lowest possible annualized return for a purchase made at a P/E10 level of 24 is a negative 4.20. The best possible annualized return is 7.80. That’s a spread of 12 percentage points
At 20 years out, the spread is smaller. The range is from a negative 0.93 to a positive 7.07. The spread is now 8 percentage points of return.
At 30 years out, the spread is smaller still. The range is from 4.06 to 8.06. The spread is only 4 percentage points of return.
At 60 years out, the spread is only two and one-half percentage points of return.
None of these findings are consistent with the idea that stock prices fall into the pattern of a random walk. If prices were random, nothing could be predicted. If returns were random it would be possible for there to be 20 years of negative returns in a row or 20 years of positive returns in a row. If prices were random, valuations would tell you nothing about future returns.
In fact, if returns were random, the concepts of overvaluation and undervaluation would be meaningless. There can be no mispricing in a world of random returns because it is not proper to think of some prices as being “wrong” unless it is proper to think of other prices as being “right” and no price is more right than another if all possible prices generate the same random return.
So -- the fact that the predictions generated by the calculator are statistically significant according to the tools used by statisticians for determining such things offers separate proof that stock prices are not random. That is only the beginning of our case, however. The calculator not only shows that stock prices are predictable, it shows that they become increasingly predictable over time. That cannot be in a world in which prices are a random walk. But what cannot be is! Prices do not follow a random walk.