To start the discussion, you should read Jeremy Siegel's article that appeared in yesterday's Financial Times, "Don't put faith in Cape crusaders." One of his points is that the now-famous Shiller P/E ratio, which shows stocks to be relatively expensive today, might be overstating the valuation of stocks because it understates the value of earnings. He notes, for example, that accounting standards changed in the 1990s, and that had the effect of depressing reported earnings relative to the measure of corporate profits that comes from the NIPA (GDP) statistics, and making them more volatile. I illustrate that in the above chart, which directly compares NIPA profits to S&P 500 reported earnings (note that the y-axis of both series covers a similar span, with the top value being 100 times the bottom value). NIPA profits are clearly much less volatile, and the two series indeed began to diverge in the early 1990s. This backs up Siegel's assertion that Shiller's method could be improved by using NIPA profits instead of reported earnings.
John Authers argues in another recent FT piece, "From CAPE to CAPE," that reported earnings also suffer from changing taxation regimes, and he shows Alain Bokobsa's adjusted version of the Shiller P/E data, which suggests that stocks today are about fairly valued. I've long been intrigued by the Shiller P/E measure, which reportedly divides the S&P 500 index by a 10-year moving average of inflation-adjusted earnings. But either I'm obtuse, Excel-challenged, or possessed of the wrong data, but for the life of me I can't come close to reproducing Shiller's P/E ratios if I use inflation-adjusted earnings. Be that as it may, here are some interesting charts that use various different ways of calculating P/E ratios as suggested by Siegel:
The chart above uses Jeremy Siegel's suggestion that NIPA profits are better than reported earnings because they don't suffer from different accounting standards and because they are less volatile. Instead of dividing the S&P 500 index by a trailing 12 months measure of reported earnings, I've used the NIPA measure of after-tax corporate profits, and I've normalized the data so that the average P/E is again about 16. This also shows stocks to be very cheap in the early 1980s and very expensive in 2000, but it suggests that stocks today are very cheap.
This last chart uses a 10-year moving average of NIPA profits in a manner similar to Shiller's method. Again, the data are normalized so that the long-term average is about 16. This technique produces results that are almost identical to the previous chart, though the P/Es are less volatile as might be expected. Take your pick: according to these valuation exercises, stocks are either very cheap or about fairly valued.
I like the NIPA measure of profits, for the reasons cited above, and because NIPA profits have been calculated using actual data supplied by companies to the IRS and the same methodology throughout (this eliminates arbitrary write-offs and adjusts for inventory valuation and capital consumption). It's arguably the best measure of true "economic" corporate profits. So I'm inclined to view stocks as being very attractive today.