One problem seems interesting:
It is well known that modern finance theory (NYSE:MFT) bases on the fact that stock prices are unpredictable. Exclusion is a secular market when it is possible to say that price will change in the market direction with more than 50% probability. But even in this case I doubt that couple years prediction can be robust.
On another hand, for CCC companies we can say that dividends will go up in 1 -2 years and David Fish even can say when with probability well above 50%.
So what to do with modern finance theory in this case? What do you think about?
I'm reading good academic book "The Equity Premium: Essays and Explorations" (editors Goetzmann, William N. and Roger G. Ibbotson). To my surprise the scholars admits that they ignored dividends in calculation of US stocks returns for 1926-1974 (p.31. In different calculations dividends cause from ~40 to 90+% of total returns, so MFT operated with incorrect "field" results.
Let me copy the relevant paper that eliminate lot of SA discussions and un-based opinions.
Stock Return Predictability: Is It There?
Andrew Ang and Geert Bekaert
Review of Financial Studies, Vol. 20, No. 3 (May 2007): 651-707
Previous research has found that dividend yields forecast stock returns at long time horizons. Using more appropriate statistical tests and international data, the authors do not find that dividend yields predict stock returns at long time horizons but do find predictability at shorter time horizons.
Previous research has found that dividend yields can predict stock returns and that the predictive power is stronger at longer time horizons. Dividend yields had been found to weakly forecast the dividend growth component of stock returns, so researchers had attributed changes in dividend yields to changes in expected return forecasts.
The authors use two sets of international data to examine and verify U.S. results. A long dataset includes data from the United States, the United Kingdom, and Germany from 1935 to 2001. A shorter dataset also includes data from France. In subsequent tests, the authors also examine results using data that start in 1952 (because interest rates were pegged in the 1930s and 1940s) as well as data that exclude the 1990s (because previous research has shown that the predictive power of the dividend yield disappears when 1990s data are added).
In their main tests, the authors use various econometrics to examine the relationship between dividend yields and stock returns 1 month, 4 months, and 20 months after the dividend yield is measured. They find that the predictive power of dividend yields at long time horizons generally holds when tests are estimated with typical standard errors. As soon as heteroscedasticity and serial correlation are accounted for, however, the predictive power of the dividend yield is weaker and largely nonexistent at 20-month horizons.
At short time horizons (one month and four months out), the authors do find that the dividend yield, together with the short-term interest rate, generally has predictive power for stock returns, although by itself, the dividend yield does not have predictive power when 1990s data are included.
Examining international data, the dividend yield has predictive power for stock returns only in the United Kingdom and only at very short time horizons. The coefficient for the short-term interest rate is negative at short time horizons and turns positive at longer time horizons for France, Germany, and the United States; the coefficients are statistically different from zero at short time horizons for the latter two countries. In another set of tests using pooled country data, the authors find evidence that both the dividend yield and the short-term interest rate have predictive power at short time horizons but not at longer time horizons.
In further tests, the authors find a positive relationship between dividend yields and future interest rates, with higher dividend yields predicting higher future interest rates, which is consistent with the relationship predicted by the authors' model and which is consistent across countries. The model and empirical tests also show that dividend yields have weak forecasting power for future cash flows that is not robust across countries.
Previous research has suggested that the earnings yield has forecasting power for cash flows and stock returns. The authors examine these issues and find that the earnings yield weakly predicts stock returns but has more power for predicting cash flows.
In summary, the authors find that dividend yields do not have predictive power for future stock returns at longer time horizons-once the appropriate statistical tests are used. This finding is contrary to previous research that has found predictability of stock returns at long time horizons. At short time horizons, evidence exists that dividend yields and short-term interest rates predict stock returns