Quote:
Originally Posted by Running_Man
Then for real world "proof" a chart is posted for 21 year period between Vanguard Equity and Vanguard total market. That shows Vanguard Equity income outperforms Vanguard Total Market from 1992 10 K becoming 80K invested in dividend stocks instead of 73 K invested in total market. The magic pants crowd then derides that difference as insignificant, due to the 2000 tech boom failure, or higher risk profile for the dividend stocks since they include value stocks and value stocks have a higher risk profile.
|
The most compelling argument I see on bogleheads that dividends make no difference is that they don't appear in factor models where one is performing regressions to explain equity returns. In general, if an obvious variable like dividend (or dividend yield, etc) doesn't appear in a regression model as a predictor it typically means one of the following:
(1) The variable is irrelevant and has no explanatory power
(2) The variable is relevant but the effect size is to small to make it worthy of including in the model
(3) The variable is relevant but once we consider other variables, the marginal benefit is too small to include in the model.
(4) The model is misspecified so that the term doesn't appear as significant (but might be significant if the model had a different form).
In this case I think many on BH might argue that dividends are subsumed by the value factor (reason #3). #4 seems unlikely as a reason that dividends don't appear in the regression models given the amount of academic interest.
Quote:
There have been several long term studies that have shown that dividend stocks do offer better returns. Jeremy Siegel in 2005 had a study and their is this good one on dividends.
|
Are you referring to this paper: “The Superior Risk and Return Characteristics of Dividend-Weighted Stock Portfolios,” (with Jeremy D. Schwartz and Luciano Siracusano), WisdomTree Investments, March 2006, 63 pages?
I wasn't able to find an online copy but given the publication time-frame Siegel would be well aware of the various factor models to explain equity returns. Do you know what his argument would be against the idea that the existing factors (e.g. value and perhaps low beta) already encompass the information in dividends?