Dividend Growth vs. Small-Cap Value: What the Factor Regressions Show

Dividend growth and systematic small-cap value attract the same kind of investor for the same stated reason: own durable, profitable, sensibly priced businesses instead of whatever the index happens to weight most heavily. The usual conclusion is that the two strategies are cousins, and that a dividend screen is a rough, cheap way to reach the same value and quality exposures a factor fund targets deliberately.

That conclusion is testable, and almost nobody tests it. So we ran the regressions. The answer is not the one the cousin story predicts.

The regressions nobody prints

The standard advice is to look through a fund’s label to its factor exposures. The standard article then declines to do it. So here are the numbers.

Each fund’s monthly excess return is regressed on the five Fama-French factors: market, size (SMB), value (HML), profitability (RMW) and investment (CMA). Factor returns come from the Kenneth R. French Data Library; fund returns are dividend-adjusted monthly totals. Each regression uses that fund’s full history through June 2026, with heteroskedasticity-robust standard errors. Loadings are the coefficients; t-statistics are in parentheses.¹

Two results stand out, and the first one is the reason this comparison is usually framed wrong.

Dividend growth has no value exposure

VIG’s HML loading is 0.002 with a t-statistic of 0.1 over 241 months. NOBL’s is 0.035 (t = 0.6). Neither is distinguishable from zero. A dividend-growth screen does not produce a cheap portfolio, and twenty years of data say so clearly.

What those funds do load on is profitability and conservative investment, both strongly and reliably. NOBL, which requires 25 consecutive years of increases, has the highest RMW loading in the table at 0.395, higher than AVUV’s. Read the screen and this makes sense: a company that has raised its dividend for a quarter century has demonstrated persistent operating profits and a management team that does not plough every dollar back into the business. That is RMW and CMA almost by construction. It says nothing about the price you pay for it.

The index rules explain the result

The methodology puts it there on purpose. VIG tracks the S&P U.S. Dividend Growers Index, which requires ten consecutive years of dividend increases and then does something most summaries skip: it removes the highest-yielding 25% of otherwise-eligible companies before selecting the index (15% for existing constituents, a buffer that reduces turnover). It also excludes REITs outright and caps any single company at 4%.²

A screen that deliberately discards the cheapest, highest-yielding quarter of its candidates will not produce a value portfolio. The measured HML loading of 0.002 is the index doing exactly what it says on paper.

Run the same test on SCHD and the methodology predicts the answer again. The Dow Jones U.S. Dividend 100 Index requires ten consecutive years of dividend payments rather than increases, then ranks the eligible universe by indicated yield and keeps only the top half by yield before scoring anything else. Only after that does it rank survivors on free cash flow to total debt, return on equity, yield again, and five-year dividend growth, summing the four ranks equally and taking the top 100.³ Three of those four inputs are levels rather than growth, and the universe was already tilted toward high yield. SCHD’s measured HML loading of 0.20 and RMW loading of 0.34 are what that recipe produces. Calling it a dividend-growth fund misreads its own rulebook; it is a yield-and-quality fund.

NOBL sits at the other extreme. The S&P 500 Dividend Aristocrats Index requires 25 consecutive years of increases, restricts the universe to S&P 500 members, and equal-weights the survivors with a 30% sector cap.⁴ A quarter century of uninterrupted increases is the most demanding durability filter of the three, and it produces the highest profitability loading in the table alongside a value loading of 0.04.

Other researchers reach the same split. Vanguard’s own research found that high-dividend-yield strategies are explained largely by value and low volatility, while dividend-growth strategies are explained by low volatility and quality.⁵ S&P Dow Jones Indices published three-factor regressions on its high-yield dividend indices showing market loadings around 0.72 to 0.78, size loadings statistically indistinguishable from zero, and value loadings of 0.53 to 0.61.⁶ High yield is a value strategy. Dividend growth is a quality strategy. Collapsing them into “dividend investing” hides the single most important thing about each.

Splitting them also settles a disagreement in the prior literature. Fisher, using Barra attribution over 1979 to 2012, concluded dividend strategies were a value tilt in disguise. Blitz found MSCI High Dividend loaded 0.83 on low volatility and was largely redundant with a low-volatility index. Vanguard found both, depending on the strategy.⁷ The fund-level regressions show why all three can be right: they were describing different products. Yield screens land on value, growth screens land on quality, and both land below a market beta of one.

As a check on these estimates, the one prior published regression of a named dividend ETF, PWL Capital’s five-factor run on VIG from June 2006 to June 2020, reported an annualized alpha of −0.48% with a t-statistic of 0.55.⁸ Six additional years of data move that to −0.44%, which is about as close as independent estimates get.

Small-cap value buys an exposure the dividend funds cannot

Every dividend fund in the table has a size loading between −0.09 and 0.16. AVUV, DFSV and VIOV sit at 0.91. No dividend screen reaches small caps, because the companies capable of a decade of uninterrupted dividend increases are, with few exceptions, large and established.

The population statistics make the constraint concrete. As of the start of 2026, 409 of the 500 S&P 500 members paid a dividend, or 81.3%. Among US domestic common issues outside the S&P 500, the figure was 19.4%.⁹ A dividend requirement is close to a large-cap requirement, which is why no amount of dividend-fund weighting moves a portfolio’s size loading.

The value loadings differ in degree rather than kind: AVUV at 0.54 against VYM at 0.30 and SCHD at 0.20. The profitability loadings are close, with AVUV at 0.27 and SCHD at 0.34. So the funds are not opposites. They overlap on RMW and, for the yield-oriented funds, partly on HML. They diverge almost completely on size and on market beta, where the dividend funds cluster near 0.85 and the small-value funds near 1.00.

Ninety-nine years of sorting on dividend yield

Fund regressions cover twenty years at best. The longer question is whether dividend yield has ever been a reliable way to sort stocks by expected return. Kenneth French publishes value-weighted portfolios formed on dividend-to-price back to July 1927, which gives 1,188 months to work with.

The highest-yield 30% of the market returned 11.32% annualized against 10.10% for the lowest-yield 30%. That gap looks like a dividend premium until you ask whether it is distinguishable from noise. It is not. The high-minus-low spread averaged +1.05% per year with a t-statistic of 0.80 over the full 99 years. Restricted to 1963 onward it falls to +0.26% per year, t = 0.17. The decile version of the same spread is −0.13% per year post-1963.

For comparison, the value premium (HML) over the same 1926 to 2026 span averaged 4.25% per year. Ninety-nine years of sorting on book-to-market produces a signal. Ninety-nine years of sorting on dividend yield produces a number you cannot separate from zero. And once the five factors are accounted for, the high-yield portfolios show negative alphas rather than positive ones.

Dividend payers did earn better risk-adjusted returns over the 99 years: a Sharpe ratio of 0.48 for the highest-yield 30% against 0.35 for non-payers and 0.45 for the market. The regression assigns that to value and profitability rather than to the dividend. And the comparison that matters to someone choosing a fund is against the market, not against non-payers, which changes the picture enough to be worth its own section.

The peer-reviewed version of this result is Chen and Israelov, who find the historical outperformance of high-dividend stocks is completely explained by value, quality and defensive factors, and that adding a dividend filter to a factor portfolio produces lower after-tax returns than the dividend-agnostic version.¹⁰

The counterweight is worth knowing. Naranjo, Nimalendran and Ryngaert found a positive relation between dividend yield and risk-adjusted NYSE returns from 1963 to 1994 that survived multifactor controls, though it was concentrated in smaller stocks and zero-yield stocks, and the authors concluded the effect was too large to be a tax story.¹¹ It is a 1998 paper using data through 1994, which is the profile of a finding you should expect to have decayed. Arnott and Asness separately found that higher aggregate payout ratios predicted faster subsequent earnings growth, which cuts against the assumption that retained earnings are the engine of growth.¹² Neither result rescues dividend yield as a cross-sectional expected return signal, but a guide that pretends the evidence is unanimous is not worth reading.

Does a dividend tilt reduce risk?

Every dividend fund in the regression table has a market beta near 0.85, which invites the conclusion that a dividend tilt is a way to own equities with less risk. The same 99 years of data that answer the premium question answer this one, and they answer it the other way.

The highest-yield 30% of the market carried a beta of 0.941, not 0.85. Its volatility was higher than the market’s, 19.58% against 18.42%, and its worst drawdown was deeper, −87.9% against −83.7%. A century of data gives no support to the idea that tilting toward dividend payers lowers the risk of an equity allocation.

The crisis record is worse than the summary statistics suggest, because the cushioning is concentrated in a particular kind of decline. Against the market in the seven deepest drawdowns:

High yield fell further than the market in four of the seven, including the two worst equity markets on record. It cushioned in 1973-74, 2000-02 and 2022, and those three have a common feature: the decline started in expensively priced growth stocks, so anything cheap held up. That is a style bet paying off in the conditions that favor it, which is a different proposition from carrying less risk.

VIG is the one fund with a long enough record to test directly, and it looks better than the long history predicts. From June 2006 to June 2026 it returned 10.30% annually at 13.31% volatility with a −41.1% maximum drawdown, against 11.18%, 15.73% and −50.8% for a total-market fund. It also edged a beta-matched blend of the market and T-bills on both return and drawdown. Its beta was stable across subperiods, between 0.77 and 0.84.

Three things keep that from generalizing. It is one fund over one window that begins in April 2006, two years before the single crisis most favorable to a quality screen. The mechanism is the profitability screen and the index’s exclusion of the highest-yielding quarter, which is the argument this guide has been making, and the broad high-yield portfolio that lacks both of those features lost 55.4% over the same crisis. And a −41% drawdown is an equity outcome. A fund that falls 41% has not removed equity risk from a portfolio; it has repriced which stocks inside the equity sleeve you own.

This matches what the library says about defensive equity generally. Novy-Marx finds the performance of defensive strategies is explained by controlling for size, profitability and valuation, while value and profitability strategies cannot be explained by defensive equity in return.¹³ Low beta is a factor exposure that lives inside an equity allocation, and rearranging the stocks inside an allocation does not remove the risk of owning stocks. The same pattern shows up in minimum-volatility funds, which have delivered the lower volatility they advertise without turning it into a risk-adjusted advantage, covered in the HEDGEFUNDIE correlation bet.

If the goal is to carry less equity risk, the instrument that does that is a smaller equity allocation. The last row of the table shows what a deliberate risk dial looks like: holding the market and T-bills at the same 0.94 beta gave up 32 basis points of annual return against the market and took 1.1 points off volatility and 2.2 points off the worst drawdown, at an unchanged Sharpe ratio. The high-yield portfolio earned more than either, 11.32%, but it did so by carrying more volatility and a deeper drawdown, which is the signature of a return-seeking tilt rather than a risk reduction. A dividend fund also reaches its beta by concentrating in staples, utilities and energy, which is uncompensated risk arriving alongside the compensated kind.

Turning loadings into expected returns

Loadings become interesting when multiplied by premia. Over July 1963 to June 2026 the annualized factor premia were 7.21% for market, 2.24% for size, 3.58% for value, 2.87% for profitability and 2.95% for investment. Applying each fund’s measured loadings gives a model-implied expected excess return.

VIG lands at 7.19% against VTI’s 7.34%. On sixty-three years of realized premia, a dividend-growth fund has roughly the expected return of the total market, reached by giving up market beta and buying back the difference in profitability and investment. The profitability tilt is real and it is well supported; it is also, at this magnitude, close to a wash against simply owning the index.

AVUV lands at 12.21%, about 4.9 points above the market. That number should not survive contact with your financial plan. McLean and Pontiff studied 97 published return predictors and found returns roughly 26% lower out of sample before publication and 58% lower after publication, the gap between them being the cost of everyone reading the paper.¹⁴ Apply that 58% haircut to the size, value, profitability and investment premia, leaving the market premium alone, and the 4.9-point edge falls to about 2.3 points. Subtract the 0.22-point fee difference against a total-market fund and roughly 2.1 points remain on the sleeve. Hold that sleeve at 20% of equity and it contributes about 0.4 points a year to the portfolio, which is what the calculator below reports for that allocation.

That is the correctly sized version of the small-value case. It is worth having and it is not going to change your retirement date on its own. Any argument that depends on the full 4.9 points is an argument built on a backtest.

The behavioral argument does not survive the data

The most common defense of dividend growth is behavioral. Rising income is easy to hold; a small-value sleeve that trails the S&P 500 for eight years is not. Hartzmark and Solomon documented the underlying psychology, finding that investors trade as if dividends and capital gains are separate attributes, rarely reinvest dividends into the paying stock, and demand dividends more when rates are low and markets are poor, which lowers subsequent returns on payers.¹⁵ Mental accounting that produces discipline is still discipline, so the argument is not silly.

It just does not match the record. Measured against the market over 1927 to 2026:

The highest-yield decile trailed the market in 41% of rolling ten-year windows and lost 59.7% of its relative wealth from peak to trough, bottoming in February 2000. Small value trailed in 17% of ten-year windows. In the ETF era the pattern repeats: SCHD’s worst cumulative shortfall against VTI is −35.5% and NOBL’s is −40.4%.

What dividend funds genuinely offer is lower tracking error, around 5% to 8% annualized against 15% for AVUV. Volatility of the gap is not the same as the size or duration of the gap. Dividend strategies drift away from the market slowly enough that the drift is easy to tolerate month to month, and they have historically drifted just as far. If the behavioral case for a strategy is that you will still own it in year ten, the dividend record is not obviously the stronger one.

The case for dividend growth, and where it stops

Try it: blend the exposures

What most investors are really deciding is what happens to their whole equity allocation when they add a sleeve of one fund or the other. The calculator uses the measured loadings above rather than assumptions about what the labels mean.

Two things are worth trying. First, set the premium scenario to Zero and see whether you would still hold the allocation. Second, notice what happens to the size loading when you add a dividend sleeve: nothing. Size is the exposure a dividend screen cannot reach at any weight.

Can you hold both?

Because the two screens select on different things, a dividend sleeve and a small-value sleeve are less redundant than they look. A 20% VIG plus 20% AVUV allocation, with the remaining 60% in a total-market fund, lands at a 0.17 size loading, 0.12 value, 0.10 profitability and a 0.98 market beta: a modest tilt across three styles at market-level equity risk, because VIG’s 0.85 beta offsets AVUV’s 1.06.

Whether that is worth doing is a separate question. Two sleeves means two sets of fees, two tracking-error stories to sit through, and a more complicated rebalancing policy. Most investors who want a quality tilt and a small-value tilt would be better served by one fund that targets profitability and value together, which is what the systematic small-value funds already do. The interesting finding is that the combination is not self-cancelling, not that it is optimal.

Where taxes bite

Qualified dividends and long-term capital gains share the same preferential federal rate schedule, so the headline rate is usually not the difference. The difference is control. A dividend is taxed when the company declares it; a capital gain is taxed when you choose to realize it. A strategy that requires substantial distributions gives up that timing choice.

This matters far less than it is usually made to matter, and only in one place. In a Roth IRA, traditional IRA, 401(k) or HSA, the current-year dividend tax question does not exist, which leaves a dividend-growth fund as a large-cap quality screen and nothing more. In a taxable account the drag is real and compounds, which we work through with dollar figures in why I avoid SCHD. Small-value funds are not tax-free either: they distribute dividends and, with higher turnover, realized gains.

The related point most dividend content misses is that a dividend screen does not identify companies returning capital to shareholders. It identifies companies returning capital in one particular form. Boudoukh, Michaely, Richardson and Roberts showed that total payout yield and net payout yield carry more information about expected returns than dividend yield alone, precisely because the dividend-price ratio changed dramatically as firms shifted to repurchases while total payout did not.¹⁶ We cover that in shareholder yield.

What I recommend

A globally diversified, market-cap-weighted portfolio is the default, and it is not a consolation prize. Neither of these strategies is required for a plan that works, and a plan that requires either one to work is a plan with a problem.

If you want a deliberate expected-return tilt, systematic value with a profitability screen and a size loading near 0.9 is the more coherent instrument. Its return thesis maps onto the characteristics the asset-pricing literature studies, it does not condition eligibility on payout policy, and the regressions confirm it buys the exposures it advertises. Size it so that the zero-premium case is acceptable, which for most people means 10% to 30% of equity rather than half the portfolio.

I would not recommend dividend growth as a way to raise expected returns or to lower risk, because the evidence supports neither. The one reason that survives is behavioral: if a rising distribution is what keeps you from selling in a bear market, buy it in a tax-advantaged account, buy it cheaply, and be clear with yourself that you are paying for a habit rather than an exposure. If you want lower equity risk, hold less equity. If you want a quality tilt, a broad profitability-screened fund buys more of it without conditioning on payout policy.

The one position the evidence does not support is choosing dividend growth over systematic value on the theory that it is a cheaper route to the same place. The regressions say it is a route to a different place.

Who each fits

How Summitward helps

The regressions in this guide are of funds. The ones that matter are of your portfolio.

• Portfolio factor analysis measures the size, value and profitability exposures you actually hold, which is often not what the fund names suggest.
• Benchmark comparison shows up-capture, down-capture and tracking error against a broad index, so the drought risk is a number rather than a feeling.
• Retirement and Monte Carlo tools let you run the plan with the style premia set to zero, which is the test any tilt should pass before you fund it.
• Tax tools show where a distribution-heavy sleeve sits across taxable, traditional and Roth accounts.

Frequently asked questions

Is dividend growth a value strategy?

No. VIG’s value loading over 2006 to 2026 is 0.002 with a t-statistic of 0.1, and NOBL’s is 0.035. Both are indistinguishable from zero. Dividend-growth screens load on profitability and conservative investment instead. High-dividend-yield strategies are the ones with value exposure, at roughly 0.30 for VYM.

Is there a dividend premium?

Not one you can measure. Sorting US stocks on dividend yield from 1927 to 2026 produced a high-minus-low spread of about 1.05% per year with a t-statistic of 0.80, and about 0.26% per year since 1963. Over the same period the value premium was 4.25% per year. Dividend-paying portfolios did have higher Sharpe ratios, but the five-factor regressions assign that to value, profitability and a market beta below one.

Which has better expected returns, SCHD or AVUV?

On measured loadings times realized 1963-2026 premia, SCHD implies 8.68% excess return and AVUV 12.21%, against 7.34% for a total-market fund. Take that gap seriously and then cut it hard: published premiums have historically decayed by more than half after publication, and AVUV costs about 0.19 points a year more than SCHD. The remaining edge is real but modest, and it comes with materially higher tracking error.

Is dividend growth easier to hold than small-cap value?

It has lower tracking error, around 5% annualized for VIG against 15% for AVUV, so month-to-month it feels calmer. It has not had shorter droughts. The highest-yield decile trailed the market in 41% of rolling ten-year windows since 1927 with a worst relative shortfall of 59.7%; small value trailed in 17% of those windows.

Does it make sense to hold both a dividend fund and a small-value fund?

They are not redundant, since dividend funds have essentially no size exposure and dividend-growth funds have no value exposure. A blend lands at a mild quality-and-size tilt with near-market beta. Whether the extra fund is worth the added cost and complexity is a separate question, and for most people one fund that targets value and profitability together is the simpler answer.

Do these regressions prove one strategy is better?

No. They decompose returns into exposures; they do not establish that any premium will be paid in the future. None of the funds shows statistically significant alpha, which means each is doing what its construction implies and nothing more. The forward-looking question is which exposures you want and at what cost.

Key takeaways

• Dividend growth is a profitability bet. VIG’s HML loading is 0.002 over twenty years; its RMW and CMA loadings are 0.20 and 0.21 with t-statistics above 4.
• High yield and dividend growth are different strategies. VYM loads 0.30 on value and 0.15 on profitability. VIG is the reverse. Treating them as one category hides what each owns.
• Size is the exposure no dividend screen reaches. Every dividend fund tested sits between −0.09 and 0.16 on SMB; the systematic small-value funds sit at 0.91.
• Ninety-nine years of dividend-yield sorts produce no reliable spread. +1.05% per year with a t-statistic of 0.80, against 4.25% per year for the value premium over the same span.
• Size the tilt for the zero-premium case. After a post-publication haircut and fees, a 20% small-value sleeve adds roughly 0.4 points a year to a portfolio. That is worth having, and it is too small to build a retirement date on.

Related guides

• Why I avoid SCHD: the opinionated version, with the taxable-account dollar math.
• Dividends are not free money: the mechanics of why a dividend transfers value rather than creating it.
• The case for small-cap value: the premium and the droughts you have to survive to collect it.
• The profitability factor: what RMW measures and why it strengthens a value tilt.
• Is factor investing dead?: how much of the historical premium to expect, and how to size a tilt.
• Shareholder yield: why dividends alone understate what a company returns to owners.

Sources

1. Kenneth R. French Data Library, Tuck School of Business at Dartmouth. Five-factor and three-factor monthly returns and portfolios formed on D/P, built from the CRSP 202606 database; data through June 2026. Fund returns are dividend-adjusted monthly totals. Regressions are ordinary least squares with HC1 robust standard errors. One methodological note matters here: CRSP retired its legacy stock tape in January 2025, and the current one compounds daily returns with dividends reinvested on their ex-dates rather than at month-end. Schwarz, Walter and Weiss find this rewrites 9.62% of monthly returns by more than a basis point while leaving average premia and their significance largely unchanged. The effect is concentrated in exactly the population at issue here: comparing the two vintages of French’s D/P portfolios, 97.7% of months differ for the zero-dividend bucket with a mean absolute difference of 41.8 basis points, against 3.7 basis points for the low-yield tercile, because non-payers are disproportionately small, newly listed and delisting-prone. Treat the zero-dividend row as the least precise line in that table. dartmouth.edu
2. S&P Dow Jones Indices, “S&P Dividend Growers Index Series Methodology,” May 2026. Ten consecutive years of increases for the U.S. index; universe is the S&P United States BMI; removes the top 25% of eligible companies ranked by indicated annual dividend yield (15% for existing constituents); excludes GICS Equity and Mortgage REITs; float-cap weighted with a 4% single-company cap; annual March reconstitution with quarterly updates. VIG has tracked this index since September 19, 2021 and previously tracked the NASDAQ US Dividend Achievers Select Index. VIG expense ratio 0.04%, 332 holdings, 1.44% 30-day SEC yield. vanguard.com
3. S&P Dow Jones Indices, “Dow Jones Dividend Indices Methodology,” August 2026. Dow Jones U.S. Dividend 100 Index: universe is the Dow Jones U.S. Broad Stock Market Index excluding REITs; minimum ten consecutive years of dividend payments; only the top half of eligible securities by indicated annual dividend yield proceed to scoring; equal-weighted rank sum of free cash flow to total debt, return on equity, indicated yield and five-year dividend growth; top 100 selected; capped float-cap weighting with a 4.0% single-stock and 25% sector cap. SCHD net expense ratio 0.06%. sec.gov
4. S&P Dow Jones Indices, “S&P Dividend Aristocrats Indices Methodology,” July 2026. S&P 500 Dividend Aristocrats: S&P 500 membership required; at least 25 consecutive years of increased total dividend per share; minimum US$3 billion float market cap and US$5 million average daily value traded; equal weighted with a 30% single-sector cap and a minimum of 40 constituents. NOBL net expense ratio 0.35%. proshares.com
5. Schlanger & Kesidis, “An analysis of dividend-oriented equity strategies,” Vanguard Research, 2017. High-yield strategies explained largely by value and low volatility; dividend-growth strategies by low volatility and quality.
6. S&P Dow Jones Indices, “Dividend Investing and a Look Inside the S&P Dow Jones Dividend Indices,” September 2013. Exhibit 16 reports Fama-French three-factor loadings for the Dow Jones U.S. Select Dividend Index (market 0.78, SMB −0.05, HML 0.61) and the S&P High Yield Dividend Aristocrats (market 0.72, SMB 0.02, HML 0.53), January 2001 to late 2012.
7. Fisher, “Dividend Investing: A Value Tilt in Disguise?” Journal of Financial Planning, April 2013 (Barra attribution, 1979–2012, finding the yield factor’s contribution to return was negative); Blitz, “Factor Investing with Smart Beta Indices,” 2016 (MSCI High Dividend loads 0.83 on low volatility). ssrn.com
8. PWL Capital, “Five Factor Investing with ETFs.” Five-factor regression on VIG, June 2006 to June 2020: R² 94.8%, annualized alpha −0.48% (t = 0.55). pwlcapital.com
9. S&P Dow Jones Indices, indicated dividend payments release, January 7, 2026: 409 of 500 S&P 500 members paying (81.3%); S&P MidCap 400 65.0%; S&P SmallCap 600 57.0%; non-S&P 500 U.S. domestic common issues 19.4%. Long-run context from Fama & French, “Disappearing dividends,” Journal of Financial Economics 60(1), 2001, 3–43 (the proportion of non-financial, non-utility NYSE/AMEX/Nasdaq firms paying cash dividends fell from 66.5% in 1978 to 20.8% in 1999). spglobal.com
10. Chen & Israelov, “Income illusions: challenging the high yield stock narrative,” Journal of Asset Management 25(2), 2024, 190–202. springer.com
11. Naranjo, Nimalendran & Ryngaert, “Stock Returns, Dividend Yields, and Taxes,” Journal of Finance 53(6), 1998, 2029–2057.
12. Arnott & Asness, “Surprise! Higher Dividends = Higher Earnings Growth,” Financial Analysts Journal 59(1), 2003, 70–87. aqr.com
13. Novy-Marx, “Understanding Defensive Equity,” NBER working paper. Defensive performance is explained by controlling for size, profitability and relative valuations, while value and profitability strategies cannot be explained by defensive equity. nber.org
14. McLean & Pontiff, “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance 71(1), 2016, 5–32. 97 predictors; returns about 26% lower out of sample and 58% lower post-publication. ssrn.com
15. Hartzmark & Solomon, “The Dividend Disconnect,” Journal of Finance 74(5), 2019, 2153–2199.
16. Boudoukh, Michaely, Richardson & Roberts, “On the Importance of Measuring Payout Yield: Implications for Empirical Asset Pricing,” Journal of Finance 62(2), 2007, 877–915. nber.org
17. Fund expense ratios and characteristics from issuer primary sources, retrieved August 2026: VTI 0.03%, VYM 0.04% (FTSE High Dividend Yield Index, REITs excluded), VBR 0.05%, VIOV 0.10% (Vanguard fund pages and June 30, 2026 fact sheets); AVUV 0.25%, 792 holdings, actively managed (Avantis prospectus and fact sheet); DFSV 0.30%, 982 holdings (Dimensional prospectus). Note that Vanguard renamed its CRSP-benchmarked funds in July 2026 following Morningstar’s acquisition of CRSP, so VTI and VBR now track Morningstar indexes under Morningstar-branded fund names. avantisinvestors.com
18. Fama & French, “A five-factor asset pricing model,” Journal of Financial Economics 116(1), 2015, 1–22. Also Novy-Marx, “The other side of value: The gross profitability premium,” JFE 108(1), 2013, 1–28; and Asness, Frazzini, Israel, Moskowitz & Pedersen, “Size matters, if you control your junk,” JFE 129(3), 2018, 479–509 (results hold across 30 industries and 24 international equity markets).