110 Office Park Way
Do Dividends Matter?
Pittsford, NY 14534-1755
585-385-6090 / 800-343-7151 January 2004
Michael E. Jones, CFA and Steven P. Greiner, Ph.D.
With the recent reduction in tax rate to 15% on corporate dividends, investors have renewed
their interest in stocks offering dividend income. However, popular academic theory holds that
investors should be indifferent to dividends. Also, growth stock proponents would argue that
equity returns are driven by capital gains and therefore buying dividend paying stocks is less
important than buying stocks with more rapid growth opportunities. With long term capital
gains tax rates equal to dividend tax rates, should investors ignore dividends? Do dividends
matter in the search for optimal equity returns? We believe that the proper answers to these
questions are found in the actual historic data on returns in the markets. This report is a brief
version of a study which we conducted to answer these questions.
We separated the universe of stocks into dividend paying stocks versus non-payers, and then
further segmented the dividend payers into dividend yield groups based on beginning period
stock price and reported current dividends. We then examined the historical returns of the
various groups to judge whether dividend yield had an impact on total returns.
Our study provides an analysis of the return characteristics of dividend paying stocks versus a
non-dividend paying universe over the time period of December 1985 through June of 2003. The
data shows that statistically significant excess returns are available by differentiating among
stock portfolios on the basis of dividend yield. We’ve found that dividend paying stocks
with yields between 1% to 4% consistently and significantly outperformed non-
dividend paying stocks over the study period with concomitant lower volatility. The
“sweet” spot to maximize excess return consists of a portfolio of stocks with dividend yields of
between 3% to 4%. We can also conclude that as a group, non-dividend paying stocks have poor
risk and return characteristics relative to the dividend payers in general.
We utilized the Factset/Computstat Research database without survivorship bias to examine the
returns of dividend paying and non-dividend paying stocks beginning in December 19851 to the
end of June 2002. Our beginning universe consisted of all the US domestic stocks greater than
$100 million in capitalization and with an average yearly price greater than $3 per share. For a
given month, we then downloaded the prevailing dividend yield of all stocks and separated out
the stocks with dividend yield from those with no dividend. We further segmented the dividend
paying stocks into separate portfolios based on their yield. Stocks with yields between 0% and
That is the extent of the data available in the Factset/Computstat Research database.
1%, 1% to 2%, 2% to 3%, etc. formed separate portfolios. We held these stocks (equal weighted)
for 12 months, while holding the non-dividend paying (equal weighted) universe for the same 12
months and monitored their performance. The time period would then move forward one month
and the process repeated. In this way 199 overlapping time periods were studied, the last being
July 1st of 2002 where the portfolios’ performance were monitored until June 30 of 2003. The
returns are gross total returns with no expenses or trading costs applied.
Total Return of Portfolios
We compiled total return of the varying subsets of dividend yielding stocks against the non-
dividend paying universe over time. Table I below shows the performance of the portfolios by
listing the 12 month rolling compound annual return for the dividend yielding portfolios, the non-
dividend paying universe and the Wilshire 5000 along with their standard deviation over the
time period of the study.
Non-dividend paying stocks posted poor results, gaining 5.85% versus a return of 9.15%
for the Wilshire 5000 Index. Stocks in the groups with dividend yields between 1% and
4% out-performed the Wilshire during this time period and showed less risk, as
measured by standard deviation of return.
The dividend yield range of 3% to 4% was the best performing portfolio, beating the non-dividend
paying universe by a hefty 4.11% per year. Returns worsen significantly for portfolios with
dividend yields above 5%. We believe this is due to these portfolios containing a few companies
in distress who are often about to eliminate their dividends. These stocks often suffer severe
price erosion when their dividends are reduced or eliminated. The effect of this minority group
drags down the total portfolio return.
Average 12 Month Rolling Total Return & Standard Deviation
December 1985 - June 2003
Avg Compound Excess Return over Std Deviation of
Annual Return Non-Div Universe Return
Div Yld Range %
No Dividend 5.85% 22.31%
0-1 7.88% 2.03% 15.08%
1-2 9.16% 3.31% 15.08%
2-3 9.27% 3.41% 14.80%
3-4 9.96% 4.11% 14.96%
4-5 8.74% 2.89% 15.52%
5-6 6.21% 0.35% 14.51%
6-7 5.96% 0.11% 13.63%
7-8 5.94% 0.08% 12.74%
8-9 5.41% -0.44% 13.76%
9-10 3.08% -2.78% 17.93%
Wilshire 5000 9.15% 16.28%
Clover Capital Management, Inc. – January 2004
Page 2 “Stock Selection: Do Dividends Matter?”
To dramatize the return differential, consider the impact of $1,000,000 invested at the non-
dividend paying portfolio return (5.85%) versus the return of the 3-4% dividend yield portfolio
(9.96%). During the 18 years of this study, the 3-4% dividend portfolio would have grown to
$5,523,637. The non-paying portfolio would total $2,782,502. If you invested in the dividend
paying portfolio, you would now have approximately twice as much money versus investment in
the non-dividend portfolio2.
The annualized return performance over the study period for dividend payers versus non-
dividend payers is significant. Equally impressive is the frequency of the excess return across all
the one year holding periods. Portfolios for ranges of dividend yield up to 4% out-perform a non-
dividend paying universe in more than 65% of the 199 periods in our study. This means that in 2
out of every 3 periods, the dividend payers out-performed non-dividend paying portfolios.
In any study of this nature, it is important to check the statistical significance of the data before
interpretation. For this reason, we calculate confidence limits on the means and variance of the
various dividend yield distributions of return using “bootstrapping”i methods. We can determine
quite accurately the 5% and 95% confidence limits on the means of the return distributions. The
statistical data is omitted from this summary for the purposes of brevity. However, the data
clearly demonstrates the statistical significance of excess return for dividend paying stocks up to
4% dividend yield, where they offer higher return with much less volatility and resultant higher
Sharpe ratios too.
Conclusion: Dividends Do Matter
We believe the results of this study clearly show that portfolios of stocks paying dividends are
superior investments versus non-dividend paying stocks. Both taxable and tax exempt investors
must consider the substantial long term return differentials.
The results from this study are representative of portfolios of stocks. Weighing the investment
merit of an individual security should not be based on dividend yield alone. However the data
suggests that dividend paying stocks as a whole offer superior long term returns. Therefore
investors would be wise to consider dividend yield in any equity investment strategy.
Clover Capital conducts studies of this type in a never-ending effort to improve our results
through rigorous research and well grounded disciplines. We continue to adapt our quantitative
and fundamental research techniques to utilize the fruits of these efforts.
This report provides an overview and highlights of our comprehensive study. The complete
version of the study is contained in a working paper which we are preparing for potential journal
publication. Please call if you wish to see the paper.
“Performance Evaluation using Fast Permutation Tests”, Tim Hesterberg, Research Dept, Insightful Inc., 1700 Westlake
Ave; N. Suite 500, Seattle, WA 98109 implemented in SPlus Software “Simulation and Bootstrapping for Teaching
Statistics” , Tim Hesterberg, Research Dept, Insightful Inc., 1700 Westlake Ave; N. Suite 500, Seattle, WA 98109
Assuming no transaction costs or fees.
Clover Capital Management, Inc. – January 2004
“Stock Selection: Do Dividends Matter?” Page 3
Providing Investment Management Services
For Institutions, Businesses and Individuals
110 Office Park Way
Pittsford, NY 14534