This excellent post from Neil Hume at Ft Alphaville points out some analysis from Andrew Smithers at Smithers & Co. Smithers believes the market is grossly overvalued:
As we have remarked before, valid approaches to value are disliked by many practitioners as they get in the way of business. As is inherently likely, and as a glance at Chart 1 will confirm, the stock market has been underpriced around 50% of the time. Those who sell shares would rather it were cheap 100% of the time and therefore prefer invalid metrics. The ones used vary from time to time, as those employed are restricted to those which currently give the desired answer that “stocks are cheap”.
As the valid measures of the US market show that it is currently around 40% overvalued, some ingenuity is needed to claim otherwise. The EPS for the past 12 months on the S&P 500 is $7.51 so, with the index at 1071, it is selling at a trailing PE of 142. This is far higher than it has ever been before, as the previous end month record is a PE of 47. But current multiples are no guide to value; when depressed, or elevated, they need to be adjusted to their cyclical norm.
This is how the cyclically adjusted PE (”CAPE”) is calculated and when its current value is compared with long-term average, using the geometric means of EPS and cyclically adjusted PEs,6 it shows that the market is 37.7% overpriced using 10 years of earnings’ data and 45% if 20 years are used. This method is therefore of no use to those who sell shares, or have made faulty claims about value in the past. The following are among the most common approaches to circumventing the problem this presents. Some produce relatively small distortions, but these can amount to a substantial degree of misinformation when combined.