From an interview with Robert Shiller in January 2013:
Blodget: ….one frustration a lot of people have with the cyclically adjusted P/E and others is that it’s not particularly helpful for the timing mechanism, do you think it’s good to use as a sort of projected 10-year return, where when P/Es are high the return tends to be low, and vice versa?
Shiller: John Campbell, who’s now a professor at Harvard, and I presented our findings first to the Federal Reserve Board in 1996, and we had a regression, showing how the P/E ratio predicts returns. And we had scatter diagrams, showing 10-year subsequent returns against the CAPE, what we call the cyclically adjusted price earnings ratio. And that had a pretty good fit. So I think the bottom line that we were giving – and maybe we didn’t stress or emphasize it enough – was that it’s continual. It’s not a timing mechanism, it doesn’t tell you — and I had the same mistake in my mind, to some extent — Wait until it goes all the way down to a P/E of 7, or something.
Blodget: Right, perfectly safe, so then you can buy.
Shiller: But actually, the lesson there is that if you combine that with a good market diversification algorithm, the important thing is that you never get completely in or completely out of stocks. The lower CAPE is, as it gradually gets lower, you gradually move more and more in. So taking that lesson now, CAPE is high, but it’s not super high. I think it looks like stocks should be a substantial part of a portfolio.
Read more at Robert Shiller On Stocks – Business Insider.
Colin, This confuses me. Isn’t Shiller proposing a market timing technique here? He assumes some sort of reversion to the mean idea such that you adjust your stock holdings to take into account the current value of the CAPE in relation to the historical (mean?) level.
Shiller suggests that CAPE be used for asset allocation purposes, where you increase your exposure when CAPE is low and decrease when it is high. He makes the point that with CAPE you should never be completely out of the market.