Valuation and bubbles
Yesterday I wrote that “the historical correlation between very high valuations (and other frothy phenomena) and poor long-term returns is about as solid as any relationship in finance”. It strikes me that it is worth emphasising just how much work “long-term” is doing in that sentence.
Consider the equity risk premium. Robert Shiller of Yale’s version of the ERP, which he calls “excess Cape yield”, is the earnings yield on the S&P 500 (that is, earnings/price, using 10-year average earnings) minus the 10-year bond yield. It is an OK predictor of real excess returns (that is, real stock returns minus real Treasury bond returns) over the subsequent decade:
The trick is not to let that chart make you think that valuations have anything whatsoever to do with shorter return periods. Stock price movements are too volatile, and their relationship to fundamentals too loose, for that to work. Below is the excess cape yield and five-year returns. I’ve zoomed in on just the past 50 years, rather than the 140 in the chart above:
From 1994 to 1996, it is obvious that low valuations (ERP under 2 per cent) sent a terrible signal for subsequent five-year returns (mostly in the teens!). You will also notice, in the first chart, that the ERP also sent a false signal for 10-year returns a few years earlier. But the bad signal was “less wrong” and had a shorter duration for 10-year returns because, starting in 1991, 10-year returns began to include the dotcom crash. Five-year forward returns roared all the way until 1996.
In other words, the predictability of long-term returns using valuation is a function of the fact that financial asset prices do not separate from fundamentals — real cash flows — forever. Their relationship reverts to the mean. The period over which returns are predictable is the period that is highly likely to contain a moment of mean reversion (or, much more likely, a swing from one end of the valuation spectrum to the other). And there is no reason that, as the structure of the markets change, the length of this period might become longer or shorter. Ten is a nice round number, but it’s not magic.
For many readers this will be an obvious point. But it is worth emphasising that when we say “the market is expensive/cheap” we are saying something that has no meaning — none! — over the following few years. But Wall Street people (including, at moments of weakness, this newsletter) talk about valuation as if it is relevant to understanding short- and medium-term market action. It isn’t. We must all stop doing this.