## Abstract

This article investigates the historical performance of the dividend-paying equity universe within the U.S. from 2000 through 2011 and compares it to the universe of stocks that did not pay dividends. To isolate the performance of the dividend factor, the analysis controls for the following fundamental factors: market risk, size, style, momentum, and sector. Finally, the article highlights the portfolio selection implications of this research for asset managers.

Dividend payments by a company can be considered as a tool implemented by management to reward the company’s shareholders in cash. Dividend payments generally are not related to daily changes in the price of the company stock but depend more on the long-term financial health of the company and thus are considered a “risk free” or “low risk” means of generating a desirable cash flow by shareholders. This makes companies paying out regular dividends very appealing to investors seeking income.

Theoretically, the dividend discount model relates the future dividend payments expected to be made by a company to the current value of the company’s stock. The simple version of the model states that the value of a company’s stock should be equal to the present (discounted) value of all future dividend payments. The model’s valuation of a company’s stock may make full intuitive sense, but rarely is a stock’s price at any given time accurately reflective of its value as predicted by the model. There is just too much fluctuation in any given stock’s price, and if the price was fully reflective of the stock’s value at each point in time then it would mean that the estimates of the expected future dividend payments were constantly changing, or the discount rate(s) used was constantly changing, or both. Whatever the case, it seems that there are many factors that contribute to determining the current stock price of a company.

The Fama–French three-factor model is widely applied by equity managers to construct portfolios with specific ex ante objectives (factor exposures) and to explain ex post portfolio returns. In addition to the market beta factor, which was originally proposed in the capital asset pricing model (CAPM), the other two factors used to explain stock performance are the relative size (market capitalization) and book-to-market ratio (book value) of stocks. Soon after the introduction of the three-factor model for asset pricing and portfolio selection, momentum as a screening factor for stock selection was suggested as a fourth factor. Empirical research seemed to indicate that because returns exhibited positive autocorrelation, an investment strategy based on selecting outperforming stocks over the immediate short run (say, 12 months) would continue outperforming over the immediate future. Although the three-factor model led to the creation of distinct asset classes within equities based on size and style, the momentum factor was not stable enough to justify the creation of an asset class.

Meanwhile, the original market beta factor was further fine tuned to capture the risks associated with the development status of the country the company belonged to, and the sector it was operating in. This distinction among stocks increased the ability to explain differences in their performance. This distinction also led to the creation of distinct equity asset classes based on the development status of a market (developed markets equity, emerging markets equity, and now, frontier markets equity) as well as sectors. More recently, a whole host of fundamental factors have been introduced that are believed to be correlated to performance. Factors such as the dividend yield (the ratio of the dividend paid and the price), earnings and sales growth, financial leverage (debt to equity) ratio, and liquidity (cash and equivalents divided by liabilities). Whatever the factor, the goal is always the same: Identify a stock-specific observable that not only correlates to the stock’s performance but that continues having predictive capabilities on a forward-looking basis. Then use that observable to screen stocks for portfolio selection and construction.

The goal of this article is similar: It seeks to investigate whether dividend-paying companies are different in terms of performance than companies that either cannot pay dividends (due to inadequate cash reserves) or choose not to pay dividend for other reasons. Because the goal is not to explain determinants of stock price, the approach used will be to treat dividend payments as an exogenous factor. In some respects, this is similar to treating the fundamental factors of a stock such as size, style, beta, or momentum as (given) risk factors when they are used to explain price and total performance. In reality, all of these factors for a given stock are endogenous and are related to price: Size as measured by the relative market capitalization depends on shares outstanding and the stock price; style depends on the company’s book value relative to its stock price; the beta of a stock is a measure of the sensitivity of a stock’s returns to the returns of a market portfolio; and momentum is measured by a stock’s return over a given period. Consequently, this article will treat dividend payments as a fundamental factor of a stock, similar to the treatment of the stock’s size, style, beta, or momentum in explaining the stock’s performance.

Even in a world where dividend payments are treated as an exogenous policy variable, from the perspective of the company they can be justified on many grounds. Dividend payments can be made to reward owners of the company’s stock. The company could use dividend payments to signal to the market that it has strong and stable earnings from a healthy business or that it has solid cash reserves. Investors may use dividend payments as an indicator of a good value for investing in the company’s stock. A company’s management may choose to make dividend payments instead of choosing to make low-return investments within the company. Or, they could also be hinting at a lack of a reasonably priced growth opportunity through an acquisition. Lastly, a company could choose to pay dividends to establish itself as leader, or top performer, within its peer group or sector.

The approach used in this article will be empirical in nature. The time period over which the analysis will be performed will be the most recent millennium, ending 2011. Based on a company’s dividend policy available at a point in time, the first step will be to create dividend-paying and non-paying portfolios using a U.S. stock universe. To keep these portfolios representative of their respective dividend policies, these portfolios will be reconstituted quarterly over the 12-year period under evaluation. In addition to whether the stock pays a dividend or not, the portfolios will also screen the stocks on the basis of traditional risk factors, such as the stock’s size, style, beta, and momentum, but only one factor at a time. At the end of the 12-year period, the performance characteristics of the portfolios will be compared.

**CONTRIBUTION OF DIVIDENDS TO U.S. EQUITY MARKET PERFORMANCE**

Given the poor performance of equities in the most recent millennium, it would not be an exaggeration to say that dividend payments have been the redeeming factor for equity investments. Without dividends, equity investment returns would have been not only dismal but, in many cases, negative. To demonstrate the increased importance of dividends to equity returns, Exhibit 1 presents the returns—with and without dividends—of two U.S. stock market benchmarks. The performance of a developed world ex U.S. stock market index is also included to drive home the point that this is just not a U.S. phenomenon; and the argument can be generalized to global equities. Exhibit 1 breaks down the performance by calendar year starting in 2000 and ending in 2011.

For the U.S., it is clear that if not for dividends, both the market indexes presented in Exhibit 1 would have posted flat to negative returns in the 12-year period analyzed. Dividends were successful in taking the negligible annualized price return of the Dow Jones Industrial Average (DJIA) from 0.5% to a remarkably better annualized total return of 2.9%. This means that on a cumulative basis, over the 12 years ending 2011, while the price of the DJIA increased by only 6.3%, dividends provided a positive cash flow of 32.3%, resulting in total cumulative gain of 40.6%.

Dividend cash flows were successful in turning the performance of the S&P 500 Index from a price return that was in the red to a total return in the black. Over the most recent 12-year period ending 2011, on a cumulative total return basis, the 24.8% dividend return of the companies making up the index were able to negate the -14.4% decline in their prices, thereby allowing the index to post a total cumulative of 6.8% over the time period.

Developed equity markets outside of the U.S. were also saved by dividends from posting negative returns. The cumulative decline of 11.3% in prices as measured by the Dow Jones World Developed ex U.S. Index was turned into a cumulative total gain of 19.6% due to a cumulative dividend return of 34.8%. On an annualized basis, this translated into a price return of -1.0% and a dividend return of 2.5%, for a total return of 1.5%.

Note also in Exhibit 1, that although the price return exhibits a high degree of volatility, the dividend flows are not only significantly stable but also, by definition, always positive. This is another desirable attribute of dividends. The calendar year price returns for the DJIA vary from a low of -33.8% in 2008 to a high of 25.3% in 2003 for a total spread of 59.1%. Meanwhile, the calendar year dividend returns vary from a low of 1.4% in 2000 to a high of 3.3% in 2009 for a total spread of 1.9%. It would be tempting to conclude that because the S&P 500 is composed of a much larger number of stocks than the DJIA, it would be more diversified and therefore it would have a lower volatility. The inclusion of smaller stocks actually increased the highs and the lows in returns, however, causing the S&P 500 to exhibit an even larger spread than the DJIA in the range of calendar year price returns (smaller stocks are more volatile). In 2008, the S&P 500 price fell by 38.5%, while in 2003 it rose by 26.4%, implying a total spread in price returns of 64.9%. In contrast, calendar year dividend returns exhibited a tighter spread of 1.2% (the lowest dividend was 1.2% in 2000, while the largest was 2.4% in the years 2008 and 2009).

The calendar year price returns for the index measuring the performance of broad developed markets excluding the U.S. also exhibited a wide range. The price of the Dow Jones World Developed ex U.S. Index declined by 45.9% in 2008, and the largest increase in price of 38.6% was in 2003—an enormous spread of 84.5 percentage points (pps). The annual dividend return ranged from a low of 1.5% in 2000 to a high of 3.4% in 2009 for a spread of 1.9 pps.

Finally, note that developed equity markets have been exhibiting a slow but steady increase in annual dividend returns over the first 12 years of this new millennium. As explained earlier, there may be a whole host of reasons that could provide an explanation for this observed trend. Irrespective of the explanations, the more important issue is the implications this has for the role of equities in a portfolio (for example, growth versus income) and consequently the impact on investors as revealed by their preference for investing in equities over other asset classes.

**THE U.S. 2500 EQUITY UNIVERSE**

It is clear that the dividend payments contributed significantly to the total performance of equities, and their relative contribution has been increasing over time. Therefore, the main goal of this article is to gain a better understanding of the dividend-paying equity universe. To keep the analysis manageable and to control for the country factor, the empirical research will focus on the U.S. equity universe. In addition, because the research will ultimately result in portfolio implications for investors, this article will use the Dow Jones U.S. 2500 Index as the equity universe.^{1}

The time period of analysis is the calendar year quarters. In other words, all portfolios will be reconstituted quarterly and the performance evaluated on a quarterly basis. Exhibit 2 presents a snapshot of the DJ U.S. 2500 stock universe at the end of the third quarter of 2011 or the start of the fourth quarter of the same year. This snapshot is used to calculate the quarterly return of the portfolios for the fourth quarter of 2011, which is the last time period included in the 48 quarters (starting in 2000 and ending in 2011) over which the analysis is conducted. The equity universe is broken down by the standard risk factors—size, style, market beta, momentum, and sector, as well as by dividend-paying and non-paying stocks.

Even in the last period, the counts are both interesting and insightful. About 1,061 or about 43% of all stocks in the universe paid dividends. This number increases to 1,284 or about 34% for the Dow Jones U.S. Total Stock Market Index universe, which was made up of the 3,809 publicly listed stocks as of the start of 4Q 2011. By style, about 63% of all the value stocks paid dividends, while less than a quarter of all growth stocks paid out dividends. Vice versa, of all the stocks that paid out dividends, about 70% were in the value category (733 of 1,061).

For the purposes of this analysis, we divided the equity universe into large and small sizes, using the median market capitalization of the companies, so that both the categories had an equal number of stocks of 1,245. A little more than half of the larger stocks paid out dividends as compared to only about 30% of smaller stocks. Irrespective of the size categories, of the largest 500 stocks, 66% were payers; of the largest 1,000 stocks, 59% were payers; of the largest 1,500 stocks, 52% were payers; of the largest 2,000, 47% were payers; and of the largest 2,500 stocks, 43% were dividend payers. This means that the larger a stock is, the more likely it pays a dividend.

The data by market risk (beta) were surprising. Of the stocks whose beta was greater than one (higher volatility than the market), about 38% paid out dividends. Of those stocks whose beta was less than one, more than half paid out a dividend. Conversely, of all dividend-paying stocks, more than half the stocks were more risky than the market (675 of 1,061 had a beta greater than one). But of all the stocks that did not pay a dividend and had enough data to calculate a beta, more than 77% were more risky than the market (1,117 of 1,406—the sum of 1,107 and 299). So, a stock that is riskier than the market is less likely to pay a dividend than a stock that is less risky than the market.

To categorize the universe by momentum, the median momentum was used to classify the universe in top and bottom segments. A significantly larger number of the top-momentum stocks paid out dividends than the bottom-momentum stocks (652 versus 409).

By sector, the largest is financials with 496 components (about a fifth of the total components), of which 350 components (70.6%) paid out dividends. The second largest sector is industrials with 460 components, of which 45.7% were dividend payers. Consumer services was the third largest sector with 341 components, of which 36.4% paid dividends.

Telecommunications was the smallest sector with 36 components, and only 11 (30.6%) of them paid out dividends). Utilities, the second smallest sector, included only 84 components, but 91.7% of the components paid out dividends, making it the sector with highest percentage of dividend payers—only seven utility companies did not pay out a dividend.

Having identified a total of 1,061 dividend payers from the Dow Jones U.S. 2500 universe at the start of the fourth quarter of 2011, Exhibit 3 presents the yield distribution of this group of companies using the dividend paid and the price at the start of the quarter. The average quarterly yield across all dividend-paying companies was 0.73%, which translates into an average annual yield of a little less than 3%. This is quite attractive considering that the yield on the 30-year Treasury bond during this time period was about 3.25% and the 10-year bond was yielding less than 2.25%. Half of the dividend-paying companies had a quarterly yield greater than 0.62% for an annual yield of more than 2.48%. One in every ten companies had a quarterly yield of 1.35% for an annual yield of 5.3%!

**PERFORMANCE BY FACTOR**

Exhibit 4 presents the historical performance of the various factor portfolios described in Exhibit 2, from the beginning of 2000 through the end of 2011. The components of the portfolios are reconstituted at the start of each quarter and are free-float, market-capitalization weighted at that time. In essence, these portfolios can be thought of as passive portfolios that are seeking exposure to a specific factor (or factors) at any given time.

The “All” columns can be interpreted as the performance (return and volatility) of portfolios controlling one factor at a time. In most cases, the performance across factors is as expected, as highlighted by the annualized returns in the “All” columns.

By style, value outperformed growth on both a price and a dividend-return basis and with a lower standard deviation. The annualized total return of the portfolio made up of all value stocks was 3.8% with a standard deviation of 16.5%. In comparison, the growth portfolio returned -1.2% with a volatility of 21.3%. The value portfolio also paid out a much higher dividend return than did the growth portfolio, 2.6% versus 0.9%, which contributed to increasing the difference in the price returns between the two portfolios. By size, small stocks outperformed large stocks on both a price and a dividend return basis but had a much greater volatility. The annualized total return of the portfolio made up of all small stocks was 7.1% with a standard deviation of 26.5%. In comparison, the large portfolio returned 1.0% with a standard deviation of 17.8%. In this case, the large portfolio paid out a higher divided return of 1.8% versus 1.3% for the small portfolio, thereby helping to lessen the difference in the price returns between the two portfolios. Low-beta stocks (those with volatility less than the market) outperformed high-beta stocks (those with volatility greater than the market) on both a price and a dividend return basis and, as expected, had a much smaller volatility. The annualized total return of the portfolio made up of low-beta stocks was 3.9% with a standard deviation of 13.2%. In comparison, the high-beta portfolio returned -4.2% with a standard deviation of 28.0%. The low-beta portfolio also paid out a higher dividend return of 2.1% versus 1.3% for the high-beta portfolio, thereby increasing the difference in the price returns between the two portfolios.

In terms of the standard factors, the only anomaly over this time period was that of momentum. The bottom momentum portfolio outperformed the top momentum portfolio on both a price and a total return basis but had a greater volatility. The annualized total return made up of low-momentum stocks was 1.5% with a standard deviation of 21.2%. In comparison, the high-momentum portfolio returned 0.9% with a standard deviation of 18.0%. The low-momentum portfolio also paid out a slightly higher dividend return of 1.8% versus 1.7% for the high-momentum portfolio, thereby marginally increasing the difference in price returns between the two portfolios.^{2}

For the purposes of this article, the results across dividend-paying portfolios and non-dividend-paying portfolios are more relevant. These are presented in the exhibit in the columns labeled “Dividend Payers” and “Non Payers.”

Comparing the results for dividend payers and non payers in the row labeled “All Equities” is equivalent to controlling solely for the dividend factor. Dividend payers outperformed non payers on both a price and dividend return basis and with a lower standard deviation. The annualized total return of the portfolio made up of all dividend-paying stocks was 3.0% with a standard deviation of 15.8%. In comparison, the non-payer portfolio returned -1.6% with a volatility of 24.5%. As expected, the dividend-paying portfolio also paid out a much higher dividend return of 2.4% versus 0.4% for the non-payer portfolio, which contributed to increasing the difference between the price returns of the two portfolios.

If more dividend-paying stocks tend to have a value bias, then clearly the results for dividend-paying stocks versus non-paying stocks would be biased. Consequently, the remaining rows of the exhibit control for one additional factor at a time in addition to dividend factors.

By the style and dividend factors, the dividend-paying value portfolio outperformed the non-paying value portfolio on both a price and a dividend return basis and with a lower standard deviation. The annualized total return of the former portfolio was 4.3% (price return 1.3%) with a standard deviation of 15.6%. In comparison, the latter portfolio returned 1.2% (price return 0.2%) with a volatility of 25.1%. Similarly, the dividend-paying growth portfolio outperformed the non-paying growth portfolio on both a price and a dividend return basis and with lower volatility. The annualized total return of the former portfolio was 0.9% (price return -0.6%) with a standard deviation of 17.9%. In comparison, the latter portfolio returned -2.8% (price return -3.0%) with a volatility of 25.0%.

By the size and the dividend factors, large dividend payers outperformed large non payers, and small dividend payers outperformed small non payers. They were able to accomplish this outperformance solely on the basis of the price return—the dividend return then magnified this difference, thereby increasing the difference between the total returns. In addition, for both the size categories, the dividend-paying portfolios had a lower standard deviation than the respective non-paying portfolios.

By the beta and the dividend factors, the high-beta dividend-paying portfolio outperformed the high-beta non-paying portfolio on the basis of return and volatility. However, this was not the case for the low-beta dividend-paying portfolio. That portfolio posted a total annualized return of 3.7% (1.0% price return and 2.6% dividend return), while the low-beta non-payer portfolio posted a total annualized return of 4.0% (3.5% price return and 0.5% dividend return).

By the momentum and the dividend factors, top momentum dividend payers outperformed top momentum non payers, and bottom momentum dividend payers outperformed bottom momentum non payers. They were able to accomplish this outperformance solely on the basis of the price return—the dividend return then magnified this difference, thereby increasing the difference between the total returns. In addition, for both the size categories, the dividend-paying portfolios had a lower standard deviation than the respective non-paying portfolios.

By the sector and dividend factor, the dividend-paying portfolios within industrials, consumer goods, telecommunications, utilities, and technology sectors outperformed the non-payer portfolios on both a price return basis and a total return basis. But within oil and gas, basic materials, healthcare, consumer services, and financials, the non payers outperformed on a price return basis. And within these sectors, the non-payer portfolios continued to outperform on a total return basis even after adjusting for higher dividend return for the dividend-paying portfolios.

In terms of volatility, across all of the sectors, except for finance, the dividend-paying portfolios have a lower standard deviation than the non-payer portfolios.

The annualized performance over the cumulative 12-year period presented in Exhibit 4 makes an extremely strong case for investing in dividend stocks. However, when the cumulative performance is broken down into smaller periods—over calendar years, for instance—the case starts to weaken. Exhibit 5 presents the performance of the dividend-paying and non-paying portfolios over the 12-year period broken down by calendar year. Exhibit 5 also includes the difference in the price and total returns of the two portfolios by calendar year. This difference can be thought of as the “premium” associated with the dividend factor. However, because these portfolios do not correct for other risk factors, such as style, size, market risk, and momentum, and because the dividend factor is correlated to these factors, this premium would not be a “pure” dividend factor premium.

Exhibit 5 reveals that the results between the dividend-paying and non-paying portfolios vary dramatically. On the one hand, in terms of total returns, the dividend-paying portfolio outperforms the non-paying portfolio by 30.4% in 2000 (4.5% versus -25.9%). On the other hand, it underperforms by -24.3% (24.4% versus 48.7%) in 2009. That is a huge range for this multi-factor premium. In terms of “batting averages,” the performance statistics between dividend-paying and non-paying portfolios is also fairly close. Let’s say that an absolute difference of 1% or more is required for the difference in performance to be statistically significant. Then, in terms of the price return, in exactly half the years the dividend-paying portfolio outperformed the non-paying portfolio. If the criteria for statistical significance were increased to an absolute difference of 2% or more, then on the basis of price return alone, the dividend-paying portfolio outperformed the non-paying portfolio in only 5 of the 12 years. Cumulatively, over the entire 12-year period, the buy-and-hold version of the spread portfolio (long the dividend-paying portfolio and short the non-paying portfolio) would have posted a price return of 29.7% (the difference between 8.1% and -21.6%), but the annually rebalanced version would have posted a price return of -4.4%, which translates into a difference of negative 40 basis points on an annual basis over the 12-year time period.

It is only when dividends are included that the batting averages for the dividend-paying portfolios start to look better. Irrespective of the criteria for statistical significance, on a total return basis, the dividend portfolio outperforms the non-paying portfolio in 6 of the 12 years, underperforms in 4 years, and the results are statistically identical in 2 years (2004 and 2005). Cumulatively, over the entire 12-year period, the buy-and-hold version of the spread portfolio (long the dividend-paying portfolio and short the non-paying portfolio) would have posted a total return of 60.9% (the difference between 43.1% and -17.8%, but the annually rebalanced version would have posted a total return of 22.3%, which translates into a difference of 1.7% on an annual basis over the 12-year time period.

**IMPLICATIONS FOR PORTFOLIO MANAGEMENT**

To illustrate the implications for portfolio management of using the dividend payment as factor, Exhibit 6 presents the performance of some selected passive portfolios. The primary investment objective of the portfolios is to generate dividend flows with a high yield. The secondary objective is capital appreciation. The portfolios are compared in terms of price and dividend return and in terms of volatility as measured by the standard deviation. All of the portfolios are reconstituted quarterly and are free-float, market-cap weighted at the reconstitution. The “All Dividend Portfolio” in the exhibit is made up of all dividend payers from the DJ U.S. 2500 (identical to the one presented in Exhibits 4 and 5) and can be thought of as the performance benchmark.

The “Top 90% Dividend Payers” portfolio excludes the lowest-paying 10% of the stocks from the benchmark and marginally improves the total return of the benchmark to 3.6% and reduces the volatility to 15.4%, thereby improving the return–risk trade-off to 24 bps for every percent of volatility. This trade-off can be improved even further to 43 bps by building a portfolio made up of the top 100 dividend-paying stocks. By selecting the top 100 yielding stocks (portfolio A), however, both the price and dividend return increase significantly to 7.1% and 7.0%, respectively, resulting in a total return of 14.6%. And even though the volatility of this portfolio goes up to 21.9%, the return–risk trade-off increases to 67 bps for every percent of volatility.

The high level of volatility may be a concern to investors with a low appetite for risk. The volatility of the dividend payers can be reduced by screening out stocks that have a higher volatility than the market (beta greater than one) from the top 90% of dividend payers. This reduces the volatility of top 90% dividend payers portfolio to 12.7%. Selecting the top 100 yielding stocks from this universe (portfolio B) results in a similar price return (7.2%) and slightly lower dividend return (6.2%) as portfolio A. But this portfolio has a substantially lower standard deviation of 15.8% for a return–risk trade-off of 88 bps for each percent of volatility.

Nevertheless, if portfolio volatility is not a concern and the volatility of portfolio A can be tolerated, then it may be possible to even further increase the yield of portfolio B by using leverage to exploit the low volatility. Exhibit 6 also presents the performance of a 150% leveraged portfolio B ignoring borrowing costs. The leveraged portfolio results in a yield of 9.5% and a portfolio standard deviation comparable to portfolio A. Exhibit 7 presents the time series performance of portfolios A and B and the 150% leveraged version of portfolio B. On a cumulative basis, portfolio A grew by about 414%, portfolio B by about 376%, and the leveraged version of portfolio B by 830%.

**CONCLUSIONS**

Treating dividend policy as a risk factor, this article analyzes the return performance of the dividend-paying stock universe for the U.S. and compares it to the non-paying universe. The approach used constructs passive dividend-paying and non-paying portfolios that are reconstituted quarterly over the most recent 12-year period, ending 2011. In addition to whether the stock pays a dividend or not, the portfolios also correct for the traditional risk factors, such as the stock’s size, style, beta and momentum, but only one factor at a time. The following conclusions can be drawn from the analysis.

First, for the most recent millennium, ending 2011, dividends played a key role in determining the performance of equities within developed markets. So much so, that if it were not for dividends, the asset class as a whole would probably be in the red. In addition, historical trends indicate that dividends will continue to be important for equity investments. Dividend-paying stocks outperformed non-paying stocks over the time period under consideration.

Second, even after controlling for traditional risk factors, dividend-paying stocks outperformed non-paying stocks in most cases. This was true for size, style, and momentum. The results were mixed when the universe was controlled for market risk (beta). The non-paying stocks outperformed the dividend-paying stocks for the category that included all stocks with volatilities greater than that of the market. But non-paying stocks did have a much greater standard deviation (meaning they were made up higher beta stocks).

By sector, the dividend-paying portfolios within industrials, consumer goods, telecommunications, utilities, and technology outperformed the non-payer portfolios on a price return basis and therefore on a total return basis. But within oil and gas, basic materials, healthcare, consumer services, and financials the non-payers outperformed on a price return basis. And within these sectors, the non-payer portfolios continued to outperform on a total return basis even after adjusting for higher dividend return for the dividend-paying portfolios. In terms of volatility, across all of the sectors, except for finance, the dividend-paying portfolios had a lower standard deviation than the non-payer portfolios.

Even though the cumulative results over the entire 12-year period make a strong case for dividend-paying stocks, the performance when analyzed over shorter time period (in this case calendar years) reveals a different picture. Even though the dividend-paying portfolio outperformed the non-paying portfolio in terms of price on a cumulative basis, the accumulation of the annual difference in the price return was insignificant to slightly in favor of the non-paying portfolio over the entire period. In other words, a strategy that would have gone long the dividend-paying portfolio and short the non-paying portfolio would have been flat to negative in terms of price appreciation but would have had a positive income stream as a result of the relatively smooth dividend payments. This result is based on a univariate analysis; that is, the only differentiating factor between the two portfolios was the dividend policy.

Third, the empirical performance of dividend-paying stocks has important implications for equity portfolio construction. In this millennium, portfolios whose objective was yield performed well. Furthermore, it was possible to construct portfolios that resulted in both price and dividend returns and at the same time controlled volatility.

In summary, it is clear that equity investments made up of dividend-paying stocks have performed very differently than their non-paying equity counterparts not only in terms of return but also in terms of volatility (and therefore most likely in terms of correlations). Even after controlling for risk factors such as size, style, beta, and momentum, the dividend-paying universe continued to perform differently from the non-paying universe. This might suggest that within the broad equity asset class, the dividend-paying equity universe might be a sub-asset class that is distinctly different from other sub-asset classes that have controlled for size, style, and the market.

**APPENDIX**

## ENDNOTES

This article was written when Francis Gupta was director of index research and design at Dow Jones Indexes in Princeton, NJ. Sarah Paretti contributed extensive and exceptional research for this article.

↵

^{1}The D J U.S. 2500 universe, which is composed of large and small stocks, is more liquid than the entire universe and, at the times of reconstitution in March and September of each year, is made up of the 2,500 largest stocks by market capitalization. These stocks cover around 98% of the total market capitalization of the U.S. stock market. The entire universe, which includes micro caps and covers 100% of the equity market in terms of capitalization and count, was made up of 3,809 stocks at the start of the fourth quarter of 2011.↵

^{2}This result was also true when momentum was calculated using total returns over a 12-month period versus the 3 months defined in the exhibit. Given that this is a broad universe, this result might not be such an anomaly. Within large-cap stocks, higher-momentum stocks could outperform lower-momentum stocks. But when small-cap stocks are included, that might not be the case. Over this time period, the higher volatility of the higher-momentum stocks may have enabled them to perform better than the lower-momentum stocks.

- © 2012 Pageant Media Ltd