I find myself in an interesting position for an investor from the value school. I recognize on one hand that this is one of the highest-priced markets in U.S. history. On the other hand, as a historian of the great equity bubbles, I also recognize that we are currently showing signs of entering the blow-off or melt-up phase of this very long bull market. The data on the high price of the market is clean and factual. We can be as certain as we ever get in stock market analysis that the current price is exceptionally high. In contrast, my judgment on the melt-up is based on a mish-mash of statistical and psychological factors based on previous eras, each one very different, so that much of the information available is not easily comparable. It also leans very heavily on a few U.S. examples. Yet, strangely, I find the less statistical data more compelling in this bubble context than the simple fact of overpricing. Whether you will also, dear reader, remains to be seen. In any case, my task in this note is to present the evidence, both statistical and touchy-feely, as clearly as I can.
So let’s start by looking very hard at all the great bubbles of the past, searching for useful guides to the future. The classic examples are not just characterized by higher-than-average prices. Price alone seems to me now to be by no means a sufficient sign of an impending bubble break. Among other factors, indicators of extremes of euphoria seem much more important than price. Ben Graham, as quoted two quarterly letters ago, said that as far as he could see no bubble had ever broken (by 1963) without being accompanied by signs of real excess such as those found in 1929. Two months ago, Robert Shiller also made the point (in the Sunday New York Times) – as I will do – that not nearly enough signs of euphoria were yet present to make this look like a late-stage bubble. (Although in my opinion they have finally begun to pick up in the last two or three months.) And Robert Shiller was one of a very small group predicting a future market collapse in 1999, and one of a few handfuls with us in 2006 focusing on the future risks from an unprecedented U.S. housing bubble breaking due to vulnerable mortgages. So when he held his fire this time on the issues of a market crash, as I have done, waiting for more evidence, I took considerable comfort. After all, for a major investment bubble to burst it must first form, and this one has been very slow to do that, at least until recently.
While I am attacking the concept of leaning on price alone I should add that I carry a lot of blame for exaggerating the significance of price alone as a bubble measure. After all, I helped pioneer the data-mined result that previous bubbles have been separated from ordinary bull markets by passing through two standard deviations on their price series, a level that statistically should occur every 44 years in a random series. It has been a very useful assumption for a broad-based study of global bubbles in a variety of asset classes and it had, of course, a good historical record, because that’s how we picked it. But this time, this statistical measure has been misleading because as the U.S. market hit a two-sigma level, it had almost none of the other more important bubble indicators of investor euphoria and even craziness. Similarly, early 1998 had none when we reduced our risk levels to a minimum based on price alone. In complete contrast, late 1999 and early 2000 had very many signs of bubbly, completely irrational behavior, just as mid, or even early 1929 had.