Valuation-Informed Indexing #323

by Rob Bennett

Shiller showed in 1981 that there is a correlation between the P/E10 level that applies in the stock market on a given day and the long-term return provided for purchases of that market. We now have 35 years of peer-reviewed research confirming the correlation. It definitely exists.

The finding shakes our understanding of how stock investing works to its roots. If valuations affect long-term returns, stock investing risk is not constant but variable. This means that investors seeking to keep their risk profile constant must change their stock allocations in response to big valuation shifts to be able to do so. The popular advice urging investors to avoid all forms of market timing has been discredited.

I have never heard anyone dispute Shiller’s finding. Nor can I recall anyone dismissing the finding as unimportant. But it is a rare thing to see an investing expert say that Shiller’s finding should cause investors to change their investing behavior in any way. The conventional advice is to advocate a Buy-and-Hold strategy, one in which the investor identifies the stock allocation best suited to his risk profile and sticks with it regardless of valuation shifts. Most seem to implicitly accept that Shiller changed the world of stock investing in a “revolutionary” (Shiller’s word) way. But most continue to offer the same advice that they offered before he came on the scene.

Odd.

I try to understand what people are thinking. I’ve had conversations with thousands of investors and with a small number of experts as well in an effort to figure out what is going on. The biggest factor, I believe, is that people are highly influenced by recent events. A giant bull market began shortly after Shiller published his research. Investors did not want to hear about the merits of going with low stock allocations at times of insanely high prices during years when the penalty for doing so was either non-existent or not too terribly great and, while Buy-and-Hold has not performed well from 2000 forward, most Buy-and-Holders remain to this day at least reasonably content with their overall experience with the strategy.

I don’t see that explanation as one possessing much intellectual content. Shiller did not say that Buy-and-Hold strategies could not pay off well for long periods of time. His research shows that market timing based on valuations works only in the long term, periods of ten years or more. Valuations did not get out of hand until 1996. So even Valuation-Informed Indexers see nothing odd in the market providing outstanding results for Buy-and-Holders through 2006. And, as noted above, market returns have been less than great since 2000. So nothing that happened in recent market history seems too out of place to those who believe that Shiller is on to something big. The presumption is that the big payoff for Valuation-Informed indexers will come following the next price crash, which should arrive within the next year or two or three if the market continues to perform in the future at least somewhat as it has always performed in the past.

There’s only one explanation of intellectual merit that I have been able to identify as the reason for why the last 35 years of research has had such limited influence on the advice offered in this field. I have come across a good bit of uncertainty re the strength of the correlation between valuations and long-term returns.

Some view the correlation as not terribly significant. Others view it as compelling and yet hold doubts as to whether it is wise to try to take advantage of the correlation by lowering one’s stock allocation in response to high valuations (the concern is that one will permanently miss out on gains in the years that one goes with a low stock allocation). A third group believes that the correlation is compelling and that it would be a terrible mistake to fail to take advantage of it. I am in this third group, which is by far the smallest group of the three.

There are statistical measures that I could cite to show the strength of the correlation. I am going to resist doing so because I think that in this case the statistical measures can cause confusion. According to the statistical measures, the correlation is reasonably strong but not compellingly strong. All sides of course agree re what the numbers say. So it is not differences over the numbers that are causing the differences over the practical value of Shiller’s finding. The problem is that evidence of a correlation of less than 100 percent will not persuade those who believe strongly that it just cannot be possible to predict stock returns.

I find that the correlation is compelling because I believe that it is the strongest possible correlation that could exist given what Shiller’s research tells us about how stock investing really works. Shiller’s findings show that it is investor emotions rather than economic developments that determine stock price changes. If that’s so, we should not expect today’s P/E10 level to tell us with precision the return that will apply 10 years in the future or 15 years in the future or 20 years in the future.

There is never one wave of emotion passing through investors at a point in time. There are always multiple waves, some pushing prices up and some pushing prices down. The price that applies is the product of the interaction of the various waves of emotion. The P/E10 level from an earlier time can inform us as to the probabilities of each of the various waves of emotion being dominant at a point in time but they can never specify with certainty which wave will be dominant or to what extent it will be dominant. Hence,the correlation between the P/E10 level that applies at the time a prediction is attempted and the return that ends up applying in reality can never be perfect.

Stock returns are as predictable as we should expect them to be if Shiller is right that it is investor emotion that plays the dominant role in setting stock prices. We need to learn how to think about stock investing in the new world ushered in by Shiller’s 1981 research. Once we become comfortable with a new way of thinking, I believe that we will come to worry less about the statistical tools showing how strong the correlation is and focus more on how big an advance it is for us to be able to predict long-term returns to the extent that is now for the first time in investing history possible.

Rob’s bio is here.