Monday, October 15, 2007

Using Dividends to Predict Stock Prices

The academics have been telling us for years that the stock market is efficient. That is, it is not possible to use technical or fundamental analysis to predict the winners and losers in the stock market because each stock's current price already factors in everything that can be known about its future. Warren Buffett has said that the academics are off their rockers because he could not have amassed his $50 billion fortune if their concept of the efficient market were true. Our approach to investing is a bit different from either the efficient market approach or Warren Buffett's approach. We call our investment approach the Predictable Market Approach, and of course, the driver of the PMA is the dividend. We have found that approximately 100 of the 500 stocks in the S&P 500 have some correlation between price growth and dividend growth. By adding in interest rates, we can increase the stocks that have some predictable qualities to about 150. In the Dow Jones Industrials, we have found that about 15 of the 30 stocks are significantly correlated to their dividend growth and/or changes in earnings and interest rates. On the contrary, we can find little correlation in the S&P 500 between prices and dividends, earnings, or economic growth. As academia would say, the S&P 500 is random. On the contrary, we believe the Dow Jones is not as random in its movements as the S&P 500, mainly because the companies that comprise it have consistently paid a higher percentage of their earnings in dividends, which makes them more predictable. Here's a brief look at the PMA.
  1. Predictable: first we look for companies where some combination of fundamental data reaches an acceptable threshold of correlation to changes in their stock prices.
  2. Undervalued: second we look for companies where the predictability equation suggests a stock is undervalued, based on a simple one year projection of fundamental changes.
  3. Momentum: third we look for companies that pass the first two screens and have some validation of their undervaluation manifested in the upward movement of their stock prices.
Here's the bottom line--predictability of the potential total return for a stock is a tremendously under appreciated concept. Think about it this way, the price of United Technology (UTX) in the Dow Jones Industrials is over 90% correlated to changes in its annual dividend growth and changes in interest rates. IBM's price, on the other hand, is not correlated at any significant level to any company or economic data; it's random. You might be the world's greatest expert in predicting the dividend or earnings growth for IBM, but history tells us that there has been little opportunity to profit from this knowledge. On the other hand, if you have been skilled at predicting dividends or earnings for UTX, you have had a very high probability of predicting when UTX was over or undervalued, and profiting from this knowledge. Currently, our work shows that among the 150 companies in the S&P 500, which have some predictive qualities, 27 meet the three criteria listed above. That may seem like a low number, but that is up from 17 a month ago.