Facts, as founding father John Adams put it, are stubborn things, but our minds are even more stubborn. Doubt isn’t always resolved in the face of facts for even the most enlightened among us, however credible and convincing those facts might be. One reason is that, due to the well-documented phenomenon of confirmation bias, we undervalue evidence that contradicts our beliefs and overvalue evidence that confirms them. We filter out inconvenient truths and arguments on the side opposite our own. As a result, our opinions solidify, and it becomes increasingly harder to disrupt established patterns of thinking.
This is especially true when it comes to investors who believe that dividend-based strategies are an efficient way to invest (even though the research suggests many investors don’t understand how dividends actually work). No matter the amount of evidence presented, or the logic supporting it, investors who believe in dividend strategies will ignore the facts and data.
At least for tax-advantaged investors, dividends are irrelevant: They are neither good nor bad in terms of forward-looking return expectations. Therefore, while there is no reason to exclude dividend-paying stocks, focusing solely on them leads to less diversified (less efficient) portfolios.
The individual investor’s preference for cash dividends has long been known. However, from the perspective of classical financial theory, this behavior is an anomaly.
In their 1961 paper, “Dividend Policy, Growth, and the Valuation of Shares,” Merton Miller and Franco Modigliani famously established that dividend policy should be irrelevant to stock returns. As they explained it, at least before frictions like trading costs and taxes, investors should be indifferent to $1 in the form of a dividend (causing the stock price to drop by $1) and $1 received by selling shares.
This theory has never been challenged, and for good reason – the historical evidence supports it. Stocks with the same exposure to common factors (such as size, value, momentum and profitability/quality) have the same returns whether they pay a dividend or not. Yet, many investors ignore this information and express a preference for dividend-paying stocks.
Even worse is that high-dividend strategies – which, as opposed to growth-in-dividend strategies, are basically value strategies – have provided the smallest value premium compared to standalone value strategies based on metrics such as price-to-book value, earnings, cash flow, sales and EBITDA.
Among the more commonly cited explanations for investors’ preference for dividends is that they offer a safe hedge against the large fluctuations in price that stocks experience. Another is that investors don’t have to sell any stock to generate cash flow. But these explanations ignore the fact that the dividend is offset by a fall in the stock price, and is thus a virtual substitute for (or equivalent to) the sale of stock in the same amount as the dividend. It’s what can be called the fallacy of the free dividend – the only free lunch in investing is diversification, not dividends. Further explanations are provided by behavioral finance professors Hersh Shefrin and Meir Statman, which I have discussed in detail elsewhere.
What is particularly puzzling about the preference for dividends is that taxable investors should favor the self-dividend (accomplished by selling shares) over dividends if cash flow is required. Unlike with dividends, where taxes are paid on the full distribution amount, when shares are sold, taxes are due only on the portion of the sale representing a gain. And specific lots can be designated to minimize taxes.
Thanks to Mebane Faber of Cambria Investment Management and Wes Gray and Jack Vogel of Alpha Architect, we can see just how inefficient a high-dividend strategy is for taxable investors, especially high-tax-bracket investors. Before we get to that tax impact, however, it’s worth noting Faber, Gray and Vogel examined the performance of a high-dividend strategy relative to a simple composite value strategy (using an equal-weighted ranking of four value metrics: book-to-market, EBITDA-to-total enterprise value, earnings-to-price and sales-to-price ratio) over the period from 1974 through 2015. In the tables that follow, EW refers to equal-weighting.
|
1975-2015
|
S&P 500
|
Largest 2,000 Stocks EW
|
Top 100 Dividend Payers EW
|
Top 100 Value Composite
EW
|
Top 100 Value Composite EW ex-Top 25% Dividend Payers
|
Top 100 Value Composite EW ex-Top 50% Dividend Payers
|
Top 100 Value Composite EW ex-All Dividend Payers
|
|
Annualized Return (%)
|
10.8
|
12.8
|
13.9
|
17.4
|
17.2
|
15.9
|
15.6
|
|
Standard Deviation (%)
|
15.4
|
18.9
|
15.1
|
19.0
|
19.9
|
21.8
|
23.1
|
|
Maximum Drawdown (%)
|
-51.0
|
-51.4
|
-55.1
|
-56.5
|
-52.6
|
-52.3
|
-52.0
|
As you can see, while a high-dividend strategy provided higher returns than either the S&P 500 or an equal-weighting of the largest 2,000 stocks, its return was 3.5 percentage points per year below that of the value-composite strategy. And while the volatility of the high-dividend strategy was lower (15.1%versus 19.0%), its maximum drawdown was only 1.4 percentage points less than that of the composite-value strategy. Interestingly, whether we exclude the top 25% of dividend-payers, the top 50% of dividend-payers or all dividend-payers from the value-composite strategy, its returns were higher and the worst-case drawdowns were less than those of the high-dividend strategy, despite higher volatilities. We’ll now turn to the impact of taxes.
The following table shows the impact of taxes on dividends for taxable investors. Faber, Grey and Vogel ran three simulations using historical tax rates at the lowest and highest tax brackets. First, they looked at an investor who paid no dividend taxes or liquidation tax upon sale. This may be a person who invested in a retirement account who then donated the shares to charity. Next, they examined an investor in a taxable account and again assumed he donated his shares. In this case, they considered both a low and a high tax rate. Lastly, they analyzed two more simulations, namely an investor taxed upon liquidation at rates of 15% and 35%, respectfully, and then both those simulations with low-dividend and high-dividend tax rates.
They found that the after-tax benefit of avoiding dividend stocks, while adding a value tilt, reaped huge rewards over time.
Unfortunately, their analysis doesn’t also include the impact of realized capital gains from the annual reconstitution of the portfolios. However, it does show the impact of taxes on liquidation at the end of the period. I pointed this out, and Faber, Grey and Vogel plan to incorporate that in a future analysis. With that said, given the large differences in the table below, taxes from realized capital gains due to portfolio turnover won’t make much difference, especially for passive strategies with relatively low turnover. Also, taxable investors can use ETFs; broadly diversified ETFs have a history of almost no capital gains distributions. In addition, investors can use tax-managed mutual funds, which minimize the impact of realized capital gains by avoiding all but forced realizations of short-term gains due to actions such as mergers and acquisitions.
Pre-Tax and Post-Tax Returns of Various Strategies (1974 – 2015)
| |
Market Weight
2000 Stocks (%)
|
Largest 2,000 Stocks EW
(%)
|
Top 100 Dividend Yield
(%)
|
Top 100 EW Value Composite (%)
|
Top 100 EW Value Composite ex-Top 25% Dividend Yield (%)
|
|
No Taxes
|
10.8
|
12.8
|
13.9
|
17.4
|
17.2
|
|
Dividends at Low Tax Rate
|
10.6
|
12.6
|
12.4
|
17.0
|
17.1
|
|
Dividends at High Tax Rate
|
9.8
|
11.9
|
9.3
|
15.6
|
16.5
|
|
Dividends at Low Tax Rate with Liquidation at 15%
|
10.1
|
12.2
|
12.0
|
16.6
|
16.6
|
|
Dividends at High Tax Rate with Liquidation at 15%
|
9.4
|
11.5
|
8.9
|
15.1
|
16.0
|
|
Dividends at Low Tax Rate with Liquidation at 35%
|
9.4
|
11.5
|
11.3
|
15.8
|
15.9
|
|
Dividends at High Tax Rate with Liquidation at 35%
|
8.7
|
10.7
|
8.2
|
14.4
|
15.3
|
The impact of taxes on the high-dividend strategy is dramatic, especially for high-tax-bracket investors, turning outperformance into underperformance.
On an after-tax basis, the composite value strategy far outperformed the high-dividend strategy. For example, even in the no-taxes environment, equal-weighed value strategies both including and excluding the highest dividend payers outperformed the equal-weighted high-dividend strategy by more than 3 percentage points, 13.9% versus 17.4% and 17.2%. And when taxes are introduced, the performance gap becomes much larger. To repeat, while the impact of turnover isn’t considered, passive value strategies have relatively low turnover and investors can use ETFs or tax-managed value mutual funds, which seek to maximize after-tax returns by eliminating the realization of unnecessary short-term gains. Perhaps the most interesting finding is that a value approach that avoids the highest-dividend stocks has the highest after-tax returns for high-tax-bracket investors.
As Faber noted in his article, while many investors, including retirees, value the psychological benefits of those quarterly dividend checks, they are losing sight of the cost one pays for those benefits. He writes: “The ‘bird in the hand,’ so to speak, is costing about three or four in the bush.”
Summary
I don’t hate dividends. In fact, just as screening for them results in less diversified (and less efficient) portfolios, screening them out entirely would also result in less diversified portfolios because about 40% of U.S. stocks and 60% of international stocks pay dividends. Screening them out might also lead to less exposure to factors with premiums (such as profitability and quality, as well as value).
There are no logical, economic reasons for preferring dividends – none, just some purely psychological or behavioral ones. The knowledge of why such biases lead to less efficient investment results should help overcome the psychological hurdles.
Evidence from decades of academic research demonstrates the returns to dividend strategies (both high-dividend and growth-of-dividend) are well-explained by their exposure to common factors (such as market beta, size, value, momentum and quality). For example, a high dividend-to-price (D/P) strategy is another form of a value strategy. With that in mind, we can compare the returns of various value strategies to see how well a high-dividend strategy performed.
For the period from 1952 through 2016, the dividend-to-price premium was 2.3%, compared to a premium of 4.5% for book-to-price, 4.9% for cash flow-to-price and 6.5% for earnings-to-price. Not only was the dividend-to-price premium the lowest, it was the only one with a t-statistic that failed to show significance. The t-statistic was just 1.3, while the other t-statistics were 2.6, 3.1 and 3.6, respectively. Thus, not only was it the smallest premium, the premium it produced was statistically indifferent from zero.
Neither financial theory nor empirical evidence offers any support for using dividends in portfolio construction. Given the negatives of such strategies, specifically the loss of diversification and higher taxes, unless you have a compelling need for the psychological advantages described by Shefrin and Statman, there is little reason to visit the dividend exhibit in the factor zoo.
In addition, at least for taxable investors, dividend strategies are less efficient than total-return approaches. Why are dividend strategies still so popular? The answer is the same as for why active management strategies remain so popular. To wit, the human mind is a very stubborn thing.
Larry Swedroe is the director of research for the BAM Alliance, a community of more than 150 independent registered investment advisors throughout the country.
Postscript
For those interested in some additional reading on the subject of dividends, see these articles on how dividends impact returns, dividends and “The Magic Pants,” dividends and individual behavioral bias, the CEO catering test, dividend month premium and juicing dividends.
Read more articles by Larry Swedroe
