Valuation-Informed Indexing #376
By Rob Bennett
The difference between Buy-and-Hold and Valuation-Informed Indexing can be portrayed visually. Buy-and-Hold is the strategy you follow if you believe that stock returns play out in the form of a random walk both in the short term and in the long term. Valuation-Informed Indexing is the strategy you follow if you have seen the graphic in the early pages of Robert Shiller’s book that shows valuations rising and falling in a hill-and-valley pattern going back for as far back as we have good records of stock returns.
It cannot happen that way if the market is efficient! It cannot.
If the market is efficient, stock price changes are determined by unforeseen economic developments. These should be random. There should be unforeseen positive developments mixed in with unforeseen negative developments. But no patterns that repeat themselves. The investing strategy followed by 90 percent of investors rests on a belief in the random walk but a world in which a high P/E10 level signals low long-term returns is not a world of random walks. It is a world in which investors are fooling themselves about the value of their portfolios when prices are high and in which the reality principle eventually steps in and pops the fantasy bubble.
I sometimes make the claim that 100 percent of the evidence available to us today supports Shiller’s view of how stock investing works and 0 percent supports Fama’s view. I of course understand that the statement strikes most Buy-and-Holders as extreme and absurd. But I advance the claim sincerely. It’s that hill-and-valley graphic showing how valuations play out in the long term that persuades me that the case is so strong. The hill-and-valley graphic applies for the entire history of the stock market. And it is logically incompatible with a belief in Buy-and-Hold.
A visitor to my website asked me the other day how many times the hill-and-valley pattern has played out. The answer is four times. We came to the end of one bull/bear cycle in the early years of the 20th Century. We came to the end of a second at the onset of the Great Depression. The third ended with the stagflation of the 1970s. And we are presumably nearing the end of the fourth bull/bear cycle in our nation’s history today. Four completions of the cycle was not enough evidence to make the case for my friend. He asked me to get back to him when I can say that the same basic cycle has played out not four times but thirty times.
It’s a fair point. If it rained four Tuesdays in a row, I wouldn’t conclude that I should carry an umbrella to work every Tuesday for the rest of my life. There’s a big difference between having a major league baseball player hit in four consecutive games (it happens all the time) and having a major league baseball player hit in 56 consecutive games (where have you gone, Joe Dimaggio?). If the case for Valuation-Informed Indexing rests on a phenomenon that has repeated only four times, perhaps the case for Valuation-Informed Indexing is not nearly as strong as I make it out to be.
I think the case is strong. The bull portion of a bull/bear cycle usually last for something in the neighborhood of 20 years and the bear portion usually lasts for something in the neighborhood of 15 years. The odds against it raining on any given Tuesday are not that long; so it is entirely possible that this could happen by chance four Tuesdays in a row. And the odds of a baseball player getting at least one hit in a game are not that long; so, again, it is not too shocking to see one run up a four-game hitting streak. But lots of things all have to happen the same way for a bull/bear cycle of the type that we have seen play out four times now in U.S. history to play out even once.
The bull/bear cycles that we have seen all possess the following four elements: (1) they all begin with P/E10 levels at rock-bottom lows, levels far below the fair-value P/E10 level of 15; (2) they all end with P/E10 levels at sky-high highs, levels of 25 or greater; (3) they all show gradually increasing P/E10 levels during the early years of the cycle (with occasional price drops mixed in, to be sure); and (4) they all show P/E10 levels remaining low for a number of years at the end of the cycle. That all these things would happen together on four occasions, and that there would not be a single time-period in the history of the market in which these things failed to happen, suggests that the hill-and-valley pattern has repeated not as the result of coincidence, like the rainy Tuesdays, but because there is a driver that causes the pattern to repeat.
The better comparison might be to the changing of the seasons. Winter follows Summer over and over and over again in parts of the country that experience all four seasons. It’s not coincidence when it happens one more time. There are conditions that cause Summer weather and there are reasons why Summer weather is always transformed in a largely predictable amount of time into Winter weather. No reasonable person would say that it is realistic to count on the weather that applies on August 1 to be a good indicator of the weather that will apply on February 1. An argument can be made that that’s a parallel to the claim that Buy-and-Holders make when they tell investors that it is okay to count the gains earned in the heat of a bull market to finance retirements that will be beginning in the cold of a long bear winter.
Shiller’s finding that valuations affect long-term returns suggests that it is investor emotions that drive stock prices. Our natural inclination is to push prices up, up, up regardless of the economic realities, Why not? But the market cannot continue to function unless prices are eventually pulled down to reflect fair value. So, when investors push prices too high, they force crashes. High prices always foretell years of poor stock performance. There is no random walk in the long term. Stocks are bought and sold by humans and the hill-and-valley pattern that has applied since the stock market opened for business reflects a basic reality of human nature.
Rob’s bio is here.