The revelance of dividends : a test using an arbitrage investment strategy

This dissertation analyzes the relationship between dividend yields and stock returns. Black and Scholes (1974) test the Dividends are Irrelevant Theory and conclude that dividend yields are irrelevant because they find no relationship between stock returns and dividend yields. Conversely, Litzenberger and Ramaswamy (1979) test the Dividends Increase Returns Theory and conclude that dividend yields are relevant because they find a positive relationship between stock returns and dividend yields. Thus, the results of these studies are ambiguous and conflicting. This dissertation uses a multi-index model and a linear program to construct an Investment Strategy which consists of a long and short portfolio of high and low dividend yield stocks, respectively. The Investment Strategy is tested ex-ante by purchasing the high yield portfolio and by selling short the low yield portfolio. Several tests are employed to analyze the monthly returns on the Investment Strategy and test the relevance of dividend yields. The tests do not support the Dividends Increase Returns Hypothesis. The t-test concludes that the -0.0427 mean return on the Investment Strategy is not significantly greater than zero and the binomial t-test concludes that the 0.425 proportion of positive returns is not significantly greater than fifty percent. The tests of the Dividends are Irrelevant Hypothesis have conflicting results. The t-test concludes that the -0.0427 mean return on the Investment Strategy is not significantly different than zero at the five percent level, but is at the ten percent level. The binomial t-test concludes that the 0.425 proportion of positive returns is significantly different than fifty percent at both the five and ten percent levels. Three alternative asset pricing models are employed and none of the tests reject the Dividends are Irrelevant Theory. In summary, four asset pricing models are employed to test the relevance of dividend yields. The tests of all four models reject the Dividends Increase Returns Theory. However, the test results of the Dividends are Irrelevant Theory are not robust with respect to the particular multi-index model. The tests of three models support the Dividends are Irrelevant Theory, however the tests of one model does not. The conclusions of the Dividends are Irrelevant Theory depend upon the particular model selected.