The Business Week article points out the problems with using P/E multiples as an indicator of future stock price returns. In particular, the problem with forward (estimated) earnings is that they are often wrong, and can be distorted by a number of factors. A recent example is financial stocks. What investors thought were "earnings" quickly evaporated. Some investors try to compensate for this by using trailing (actual) earnings, but as stock prices are trying to discount the future, trailing earnings don't provide much information about market (stock price) turning points.
http://www.businessweek.com/investor/content/jul2008/pi20080718_837276.htm?chan=investing_investing+index+page_stocks+%2Bamp%3B+markets

In a Board of Governors, Federal Reserve System International Finance Discussion Paper (Predicting Global Stock Returns, June 2008), Erik Hjalmarsson did a statistical analysis of 20,000 monthly observations for 40 international markets, and concluded that traditional "fundamental" valuation measures such as the P/E multiple and Dividend/Price multiple (dividend yield) are of little use in predicting future stock returns, particularly in developed markets. Instead, his analysis showed that interest rate variables such as short-term rates and the term spread (rate spread between short- and long-term bond yields) are more robust predictors of stock returns. Thus all the hullabaloo about central bank actions and how they affect interest rates does appear to be an important factor for stock prices.