10-Year P/E Ratios: How Useful Is This Data?

| About: SPDR S&P (SPY)

The other day I wrote that the market's P/E ratio reached a four-year high. Instead of viewing it as overpriced, I said that this is an instance where the market's earnings multiple may not be a good valuation indicator. The reason is that stock prices are forward-looking and they're reacting to a brighter outlook a few months from now, even though earnings are still declining. Higher prices and lower earnings translate to a higher multiple.

In yesterday's Wall Street Journal, Mark Gongloff writes on higher multiples:

Benjamin Graham and David Dodd, the Romulus and Remus of value investing, suggest weighing prices against earnings averaged over a period of as long as 10 years. On that basis, the S&P 500 today trades at roughly 28 times reported earnings. That's lower than the peak of about 48 at the height of the dot-com bubble, but hardly a bargain; for the past 60 years, that P/E ratio has averaged about 21.

I need to add a few points. Remember that the Price/Earnings Ratio contains one major flaw. It compares two different sets of data. The price is a fixed-point number. You always know what it is at a given point in time. Earnings is a rate. That is, it can only be known between two fixed points in time.

I need to stress this point because it's often misunderstood. Just because price and earnings are different types of data doesn't mean that it's worthless to compare them. You can absolutely compare different data points, but you need to realize the limitations. (One recent commenter on my comparison of the Dow and S&P 500 said that the whole thing is worthless because one is market-weighted and the other is price-weighted. That's incorrect. It's certainly worthwhile to compare them, but it's what kinds of comparisons you can draw that is important.)

The other thing about the P/E ratio is that it's almost useless in timing the market, but it's very useful in making judgments about what stocks to buy and sell. If I have a chance, I'll post more evidence on this point later.

The last point is that I never understood the importance of looking at 10 years' worth of earnings. Graham and Dodd used that metric but it doesn't seem very useful to me. Note this recent chart I posted looking at trailing earnings and the S&P 500. What can I say, the relationship seems pretty strong to me. Why would I want to bring in data from 1998? When analyzing market data, the question to ask is how useful is this data? Not, is the data perfect?