An implication of the Efficient Market Hypothesis (EMH) is that a stock's returns are commensurate with its level of risk (a good article on EMH is located here). However, several researchers have attempted to demonstrate that low P/E stocks have outperformed high P/E stocks over time. Proponents of EMH have responded, however, that low P/E stocks have carried higher risk levels, and therefore the returns are simply commensurate with the higher risk undertaken by the investor. With their own research, they have further attempted to demonstrate that low P/E outperformance has been at best temporary over time.
In order to get to the bottom of this argument and arrive at our own conclusions, my business partner and I decided to do our own first-hand research on whether P/E has any bearing on the expected returns of a group of stocks. To do this, however, we needed to find (in order to analyze) a data set containing historical annual (or more frequent) price and earnings information for a complete group of stocks belonging to indexes for which we were interested (Nasdaq, Russel, S&P etc.). This is easier said than done.
Financial databases like Bloomberg contain a plethora of information, but almost all of it is current (normally that's a good thing, but not in this case!). A search on the web results in several articles extolling or critiquing this apparent low P/E phenomenon, but few references to the actual data sources. I then stumbled across Yale Finance Professor Robert Schiller's website, which contains monthly prices and earnings over the last 130 years. Unfortunately, this data is only for the S&P 500 as a whole, not individual stocks.
Finally, I came across a resource called Compustat, at the Bud C. Johnson Library at the Richard Ivey School of Business. Through Compustat, one can input a list of stocks (symbols delimited by spaces) and a database query will return annual price and earnings data over a specified time period in a CSV file. We look forward to analyzing this data and sharing our results.