Dividend strategies and dividend ETFs under scrutiny: irrelevance, factors, risks

In depth12 min read · updated 21 October 2026 · 14 sources

A dividend strategy focuses on stocks or ETFs with high and reliable distributions. For many investors, this is the epitome of passive income. But a dividend is not a gift: the share price falls by the same amount, and in a taxable account the distribution is usually taxed right away. This article takes a critical look at the strategy: what theory says, why investors love dividends anyway, how a homemade dividend compares with a distribution, which factor and sector bets dividend ETFs contain, and when distributions are practical nonetheless.

What a dividend strategy is

Dividend strategies select stocks by dividend yield (distributions over the last twelve months divided by the price), by the number of years of rising dividends, or by whether the dividend is covered, measured for example by the payout ratio, the share of profit that is paid out. Dividend ETFs replicate such rules as an index. The MSCI World High Dividend Yield, for example, takes stocks with an above-average dividend yield from the MSCI World, checks quality characteristics and excludes stocks whose price performance suggests that a dividend cut is likely; at the end of September 2026, it contained 367 stocks (MSCI World High Dividend Yield Factsheet 2026).

The irrelevance of dividends according to Miller and Modigliani

Miller and Modigliani (1961) showed that in a perfect market without taxes and costs, the value of a company does not depend on how much it pays out: what a company pays as a dividend is no longer part of its assets, and shareholders can create any payout they want by selling shares. A high dividend then does not mean more return, just a different split of the total return between price and distribution.

Besides dividends, companies have a second way of returning money: the share buyback. Grullon and Michaely (2002) show that US companies have gradually replaced dividends with buybacks and that buybacks are now the preferred way to start making payouts. If you only look at the dividend yield, you miss this part. A company with a low dividend and large buybacks may return more to its shareholders than one with a high dividend.

The free dividends fallacy: why investors love dividends anyway

Hartzmark and Solomon (2019) find that many private investors, and funds too, act as if dividends and capital gains were two separate things: they overlook that the dividend lowers the price, treat it as a stable income stream and rarely reinvest it in the same stock. The authors call this mistake the free dividends fallacy: the dividend appears to be free income. According to their study, demand for dividend stocks is particularly high in periods of low interest rates and weak markets, which raises the valuations of these stocks and depresses their subsequent returns.

Companies respond to this. According to the catering theory of Baker and Wurgler (2004), companies start paying dividends when investors value dividend payers at a premium, and are more likely to drop them when this demand is low. Older explanations focus on self-control: according to Shefrin and Statman (1984), the rule “only live off the income” protects capital from spontaneous spending. This is a form of mental accounting that achieves nothing mathematically but can stabilize behavior.

Homemade dividends: a worked example

The alternative to a distribution is the homemade dividend: you sell shares worth the amount you want. Illustrative figures: you own 1,000 shares at $100 each, bought at $60, and need $3,000. Option A: the company pays a dividend of $3 per share. Option B: it pays no dividend, and you sell 30 shares. Individual stocks, no trading fees.

Factor tilt: value, sectors and countries

A high dividend yield means a low price relative to the distribution. Dividend strategies therefore mainly buy cheaply valued stocks; to a large extent, they are a value strategy (see Fama-French factors and Fama and French (1993)). The dividend index factsheet shows this clearly:

Sources: MSCI World High Dividend Yield Factsheet (2026); MSCI World Factsheet (2026). Sectors according to GICS.

The dividend index has a price-to-earnings ratio a quarter lower, one sixth of the technology weight and significantly more health care, consumer staples and energy. Over ten years, it lagged far behind (cumulative +148.5% versus +250.7%), because technology dominated the decade. Since 1995, the two have been almost level. The return of a dividend strategy therefore depends above all on whether value and the dividend sectors are currently in favor. If you already hold a global index ETF, adding a dividend ETF mainly shifts weights between sectors and styles. Conover et al. (2016) also find that the advantage of dividend payers depends heavily on investment style: stocks with high dividends had the lowest risk and more than 1.5% a year more return than stocks without dividends, most clearly among growth stocks and small caps.

Dividend cuts and concentration risks

Dividends are not interest. If a company gets into difficulties, it cuts or eliminates the distribution, and the price usually falls as well. Michaely et al. (1995) show that the market reacts more strongly to dividend omissions than to initiations and that prices continue to drift in the same direction for a year afterward. The maximum drawdown from 2007 to 2009 shows how this can play out in a crisis: the MSCI World High Dividend Yield lost 62.90%, more than the MSCI World at 57.46% (MSCI World High Dividend Yield Factsheet 2026).

• Sector concentration: high dividends are paid mainly by mature sectors (energy, utilities, banks, pharma, tobacco). A strategy based on dividend yield collects them, and a shock in one sector hits many positions at once.
• Yield trap: a very high dividend yield can arise because the price has already fallen, for example because the market expects a cut. Index rules with quality filters are meant to weed out such cases; if they fail, the index buys exactly these stocks.
• Turnover: because dividend yields change constantly, dividend indices replace considerably more stocks (21.55% versus 3.18% a year). In a fund, this costs trading fees, which are reflected in the tracking difference.

Withholding tax: the higher the dividend, the more gets stuck

Countries of origin deduct withholding tax on foreign dividends. In funds, withholding tax is due at fund level and reduces the return directly: Irish-domiciled ETFs, for example, pay 15% on US dividends (Crowley 2023). The larger the part of the return that flows as dividends, the more this tax costs, so more for a dividend fund than for a broad index with a low dividend yield.

When distributions are practical after all

• Withdrawals without decisions: if you need money regularly in retirement, you get it without placing a sell order. That is convenient, but an annual sale based on a fixed withdrawal rate achieves the same predictability.
• Behavior: if distributions help you stay calm and not sell in crises, that is a real advantage, even if it is not one mathematically. The calm has its limits: dividends get cut precisely in crises.
• A clear rule against recklessness: “only spend the income” prevents the capital from shrinking. The same rule can be set up with a homemade dividend of a fixed 3%.

Advanced: dividend yield, growth and valuation

According to the model of Gordon (1959), the dividend yield is the expected return minus the expected growth. A high dividend yield can therefore mean that the company is riskier (higher r), that it will hardly grow (small g) or that it is cheaply valued. On its own, it does not indicate that the total return will be higher. Example: if stock A yields 4% with 2% growth and stock B 1% with 5%, the model expects a 6% return for both.

• Ex-dividend date and taxes: the price falls by roughly the dividend on the ex-dividend date; exactly how much depends, in studies, on shareholders' tax rates (Elton and Gruber 1970; Kalay 1982). If you buy shortly before the ex-dividend date to “capture” the dividend, you receive a taxable amount in exchange for a price drop.
• Dividend growth instead of level: indices of companies with long records of rising dividends select by a different rule from indices based on dividend yield and therefore have a different factor mix; they do not protect against cuts either.
• Interest rate sensitivity: because demand for dividends rises when interest rates are low (Hartzmark and Solomon 2019), it is likely to wane when rates rise, and dividend stocks then react similarly to bonds (our assessment).

What the research shows

The theory is clear: in perfect markets, payout policy does not matter for company value (Miller and Modigliani 1961). Empirically, taxes, costs and behavior shape the picture. Investors treat dividends as separate income, and their demand fluctuates with interest rates and market conditions (Hartzmark and Solomon 2019); companies align their payouts with this demand (Baker and Wurgler 2004) and increasingly replace dividends with buybacks (Grullon and Michaely 2002). Omitting a dividend is a strong negative signal (Michaely et al. 1995). That dividend payers did well over long data series does not contradict irrelevance: their advantage depends on style (Conover et al. 2016). Whether the dividend itself or characteristics such as valuation are behind it is tested with factor models.

Limitations and common mistakes

• Confusing dividend yield with return: a 4% dividend is 4% paid out of the total return, not interest on top.
• Ignoring the price drop: if you only count distributions, you overlook that the price falls on the ex-dividend date and that a cut hits the price on top.
• Forgetting taxes: distributions are taxable immediately, capital gains only on sale. For individual stocks, withholding tax comes on top.
• Double weighting: a dividend ETF alongside a global index ETF increases value, energy and financials and reduces technology. Check the overall allocation in the look-through.
• Extrapolating past returns: whether the strategy did better or worse in the last decade depended mainly on technology. That can change in either direction. Not investment or tax advice.

How Portfolio Lab calculates

Portfolio Lab calculates with total returns: distributions are reinvested in the price series, so distributing and accumulating versions of the same index run almost identically in the Backtest. The asset class “Equities World Dividend” is available (from 2005 an ETF on the STOXX Global Select Dividend 100, before that approximations from US data and the third with the highest dividend yield according to Fama and French). The regression under Factors shows how strongly a dividend portfolio loads on value; you can see the sector breakdown in the Look-through.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.