I finally found something where I really disagree with Florian. I think there is no reliable indicator to spot bubbles in advance. And I don’t that we will ever get one.
Like Katharina, I’m a big fan of clear definitions. The reason is that, particularly in discussions of economic issues, words often appear to have a clear meaning, but when you look at them more carefully you realize that they are either completely meaningless or their meaning is very different from what everyone thinks.
So what does Florian, or anybody else, mean when they talk about a bubble? Well, the standard definition is something like this: An asset bubble is a prolonged period of time in which the market price of an asset exceeds the asset’s fundamental value. And the fundamental value is the present value of future cash flows which the asset is expected to generate. Fair enough. So how can we spot bubbles?
According to Flo, “bubbles are generally characterized by building up exponentially”. Hence one should be able to detect bubbles by comparing current asset prices to their long-run (linear) trend. Let’s think about this. I see a stock steadily rising from $1 in December 2010 to $2 in December 2012. Suddenly in January 2013 the stock goes to $4 deviating from the 2010-2012 trend. Would you seriously infer from this fact alone, that there’s a bubble? Come on. In an efficient market, if investors receive new information in January which they didn’t have in December, the price should react immediately to reflect that new information. If fundamentals have an exponential trend, so should prices. Where is the theory that suggests that fundamentals always change linearly with time?
But what really puzzled me is Flo’s remark further down in his post where he writes “…the price of a stock, which by definition is not much more than present value of (expected) future incomes…”. If you believe that stock prices are always equal to the present value of expected future earnings, the whole talk about bubbles becomes meaningless. In other words, the concept of a bubble only makes sense if you believe in some kind of market inefficiency that allows persistent deviations of asset prices from fundamentals.
The fact of the matter is that we can’t observe fundamental value, so there is no straightforward way to spot bubbles. And everyone who says otherwise should be regarded with suspicion. If there would be an easy and reliable bubble detection recipe (“just look at deviations from a linear trend”, “just look at price/earnings ratios”, etc.), you should be able to make lots of money following that recipe. And if a significantly large group of investors follow it, there would be no bubbles to be detected according to this recipe in the first place.
The recent experience with Robert Shiller’s cyclically adjusted price-to-earnings ratio (CAPE) nicely illustrates this problem. In 1996, Campbell and Shiller showed in a paper that the CAPE could predict the stock market returns in the past fairly well. They were able to spot both the dotcom bubble of the late 90s and the subprime bubble of the 00s before they burst by comparing the current CAPE to its long-run average of 16. Now, during the past two years, the CAPE was well above that value which apparently led a lot of people to stay out of stocks during this time. And guess what, they missed a pretty nice rally. So just because some indicator successfully predicted bubbles in the past does not mean it is of much use in the future.