John K Clark wrote:
> > Shiller then *tests* this expectation
> > by looking at real data. If you're going to then dismiss his analysis
> > as "technical" because it refers to real data, then you seem to me to
> > be saying that you are so sure that you are right about stocks that
> > is irrelevant.
>Yes, Shiller looked at data, he looked at a mountain of it until he
>found two variables that produced a straight line for a while.
I agree that if you look at enough variable pairs you can find spurious correlations. But I don't understand how you can be so confident that Shiller looked at that many variable pairs. Do you know something about him personally? Or do you just dismiss all data tests because someone *might* have looked at tons of pairs?
>He doesn't say why they give a better picture of reality than
>thousands of other variables, he doesn't explain why a deviation
>from that straight line should be a cause for worry. That's why I call it
Along with Hal, I don't find the variables Shiller looked odd or strange; they seem to me very straightforward variables to test the idea that post high P/E periods perform less well than low P/E periods. And I find the idea that "buy low, sell high" might work plausible enough to be worth testing.
If you were right about Shiller being very selective, then it would be easy for you and others to publish your own study showing that your favorite measure of P/E does not predict your favorite measure of long-term returns. Know of such a study?
>I have another even more fundamental objection, it hasn't worked.
>It has been nearly 4 years since the prediction, not a trivial fraction
>of a human lifetime, and in that time the market has exploded upward.
I agree that is an objection, though I don't think it especially fundamental or devastating.
Robin Hanson email@example.com http://hanson.gmu.edu
Asst. Prof. Economics, George Mason University
MSN 1D3, Carow Hall, Fairfax VA 22030
703-993-2326 FAX: 703-993-2323