Robert Shiller (Yale Economist) computes a 10 year PE ratio which aims to smooth out short term earnings fluctuations to give a longer run view of market valuation. The following is a really great explanation of this metric and some its uses.
My only quibble with the article is the bit about the level of the PE and the return on Treasuries. This is basically the argument of the so-called Fed Model, which is on pretty shaky ground from a theoretical perspective. I won't rehash the issue now as I've talked about this before here and here.