Friday, September 3, 2010 1 comments ++[ CLICK TO COMMENT ]++

Better to use P/E ratio without any interest-rate adjustment

I ran across an interesting article by Mark Hulbert on using the P/E to forecast future returns. Mark Hulbert cites a study by Clifford Asness, "Fight the Fed Model: The Relationship Between Stock Market Yields, Bond Market Yields, and Future Returns" (Dec 2002), that suggests that using a raw P/E ratio is better than using one that is adjusted by the interest rate.  Recall how many look at the P/E ratio (or its inverse, the earnings yield) relative to bond yields (or interest rates). The Asness study appears to show that the absolute P/E value matters most.

The Result of the Study

Hulbert quotes an R-squared value of 0.349 if you simply look at the relationship between P/E ratio and the S&P 500 real returns over the next ten years; whereas if you look at the P/E ratio adjusted by interest rate, the R-squared value is only 0.097.

The raw P/E ratio R-squared value implies that 34.9% of the movement of the S&P 500 can be explained by the P/E ratio itself. Without digging into the statistical measures in the study I would say this is quite significant. You wouldn't expect the R-square to be much higher in this scenario since there are so many variables that influence stock price movement (such as wars, earnings growth, tax rates, investor psychology, and so on.)

The P/E ratio is essentially a measure of valuation and based on the quoted study it seems that 34.9% of the movement of the stock market is due to valuation. I like to think valuation is the most important driver of stock prices so this result is consistent with my expectation.


There are two implications that come to mind...

The first thing to note is that high P/Es (i.e. low earnings yield) is bad irrespective of interest rates. Some investors justify high P/E valuations if interest rates are low but this study contradicts that view. Basically, you don't want to invest when P/Es are high, period.

The other conclusion, and a real revelation for me, is how the raw P/E is a better predictor of future stock prices than the P/E ratio adjusted by interest rates. Stocks and bonds compete with each other and there is widespread thinking that adjusting the P/E ratio by interest rates is better than just looking at the raw P/E ratio. Surprisingly, doing so results in a weaker correlation to future stock returns.

One Caveat

I didn't read through the full study and can't vouch for the accuracy of the data. With investment studies it is always possible that there are big flaws in data selection. Picking a starting point of, say, 1933 can yield different results than, say, 1897. I don't know if this study is comprehensive enough and not sensitive to starting and ending points. This doesn't apply here but in many other studies we also have the problem of not enough data points (hence statistically insignificant conclusions). So don't blindly follow any study when it comes to investing.


1 Response to Better to use P/E ratio without any interest-rate adjustment

Parker Bohn
September 9, 2010 at 9:24 PM

It seems to me that stock market valuations should have a strong and rational relationship with interest rates.

However, its a bit different when you talk about expected returns, and not just rational current valuations.

If you're forecasting 10 year returns, then it is the interest rate 10 years from now that is important.  If interest rates are still this low in 10 years, then stocks are definitely not too pricey now (barring meltdown/catastrophy).
Interest rates are mean-reverting like everything else, and are quite likely to be a good bit higher in 10 years (they can't go much lower!).

Another way of saying this is, bonds and other interest rate based investments offer meager returns now.  Based on this, it is reasonable for stocks to also offer weak returns, without being overpriced.  That doesn't change the fact that expected returns are still meager, however.

Happy investing!

Post a Comment