The Dividend Discount Model (DDM) is the original and most theoretically pure equity valuation method: it says a share is worth nothing more or less than the present value of all the cash a shareholder will ever receive from it — and for a minority shareholder, that cash is the stream of future dividends. Every other discounted-cash-flow method is, in a sense, a workaround for the fact that most companies don't pay out everything they earn. The model's core tension is that its honesty about what a shareholder actually receives is paid for with brutal sensitivity to two inputs nobody can know — the long-run growth rate and the required return — and with irrelevance to the many firms that pay little or no dividend.

How it's calculated / formed

The general form discounts every expected future dividend back to today:

V₀ = Σ Dₜ / (1 + r)ᵗ, summed to infinity, where r is the required return on equity.

Because an infinite forecast is impossible, practitioners collapse it with growth assumptions:

• Gordon Growth Model (single-stage). Assuming dividends grow at a constant rate g forever: V₀ = D₁ / (r − g), where D₁ is next year's dividend (= D₀(1+g)). It requires r > g, or the result is negative or infinite. CFI's worked example: D₁ = $2, r = 10%, g = 4% gives $2 / 0.06 = $33.33 (their illustration, not a market figure). The model is named for Myron Gordon and Eli Shapiro (1956).
• Two-stage DDM. An explicit high-growth period (gₛ) for n years, then a perpetual stable rate (g_L): the explicit dividends are summed and a Gordon-model terminal value is discounted back.
• H-model. Growth starts high and declines linearly to a stable rate over 2H years: V₀ = [D₀(1+g_L) + D₀·H·(gₛ − g_L)] / (r − g_L), where H is half the high-growth period. A good fit for maturing firms.
• Three-stage. Growth, transition, and maturity — the most flexible but most assumption-heavy.

Two supporting relationships (CFA Institute curriculum): the sustainable growth rate g = b × ROE (where b is the retention ratio, 1 − payout ratio), and the justified leading P/E = (1 − b) / (r − g), which links DDM directly to multiples. The required return r is usually estimated via CAPM or a build-up method.

How it's used in practice

DDM is at its best on stable, dividend-paying businesses whose payout policy reflects earning power: regulated utilities, mature consumer staples, banks and insurers, REITs, and broad equity indices (whose aggregate dividend growth tends to track nominal GDP). Analysts most often run a two-stage or H-model rather than the pure Gordon form, because almost no real company grows at one rate forever.

In practice the single most valuable use of DDM is inversion. Rather than trusting the model to produce a precise "fair value," analysts plug in today's price and solve for the implied growth or implied required return (the CFA rearrangement r = D₁/P₀ + g). This turns the model into a sanity check: "What does the market have to believe about this company's perpetual growth to justify this price?" If the implied perpetual growth exceeds long-run GDP, the market is pricing impossible expectations. At the index level, the Gordon model underlies the widely used framework that expected long-run equity return ≈ dividend yield + dividend growth (the "Gordon equation" used by Bogle and others).

Adoption, debate & evidence

DDM is universally taught (it anchors the CFA equity-valuation curriculum) but less universally used in front-office practice than free-cash-flow DCF or relative multiples, precisely because so many firms — especially high-growth and US technology names — pay no dividend or return cash via buybacks instead. A firm with a token dividend and aggressive repurchases can look badly undervalued by a naive DDM. Damodaran and the CFA curriculum both stress that for non-payers or majority-control valuations you should use free-cash-flow-to-equity instead.

The serious empirical evidence sits one layer up, at the aggregate level. Campbell and Shiller (1988) and Fama and French (1988) showed the dividend–price ratio predicts future market returns, with the key feature that the forecasting power (regression R²) rises with the return horizon — weak at one month, but explaining a substantial fraction (commonly summarized as on the order of a quarter) of the variation in cumulative returns at the two-to-four-year horizon for NYSE portfolios over 1941–1986. (Exact R² values vary by portfolio weighting and sub-period and should be read from the original tables rather than a single headline number.) A high price-to-dividend ratio today forecasts low subsequent returns over 3–5 years. This is real and robust — but it is a statement about the market's long-horizon mean reversion, not proof that the Gordon formula prices an individual stock correctly. Crucially, this predictability is interpreted as evidence of time-varying discount rates / risk premia (Campbell–Shiller decomposition), which directly contradicts the DDM's convenient assumption of a single constant r. So the model's central simplification is empirically the part most clearly wrong.

Strengths & limitations

Strengths. Theoretically grounded — dividends are the only cash a minority shareholder actually receives, so the model is hard to game with accounting choices. It is transparent, forces explicit assumptions, and is the natural tool for stable income equities and for index-level expected-return estimates.

Limitations. (1) Extreme input sensitivity — as both CFI and the CFA curriculum emphasize, tiny changes in r or g swing the output wildly, and g approaching r sends value to infinity. (2) Constant-growth fiction — perpetual steady growth doesn't describe real firms. (3) Inapplicable to non-payers and increasingly to buyback-heavy firms, which makes it blind to a large and growing share of the market. (4) It ignores earnings retained for reinvestment except insofar as they later raise dividends.

The #1 misuse is treating the point estimate as precise truth — quoting "$33.33" as fair value when a 1-point change in g might move it 30–50%. Used as a range and an implied-expectations check, it is sound; used as a precision oracle, it misleads.

Sources

• CFA Institute, Discounted Dividend Valuation refresher reading — general DDM, Gordon, two-stage, H-model, three-stage, justified P/E, g = b×ROE, when to prefer FCFE: https://www.cfainstitute.org/insights/professional-learning/refresher-readings/2026/discounted-dividend-valuation
• Corporate Finance Institute, Gordon Growth Model — formula, worked example ($33.33), assumptions, limitations: https://corporatefinanceinstitute.com/resources/valuation/gordon-growth-model/
• Aswath Damodaran (NYU Stern), Dividend Discount Models (Ch. 13, Investment Valuation) — model variants and applicability: https://pages.stern.nyu.edu/~adamodar/pdfiles/val3ed/c13.pdf
• Campbell, J. & Shiller, R. (1988), The Dividend-Price Ratio and Expectations of Future Dividends and Discount Factors, Review of Financial Studies — D/P predicts returns; time-varying discount rates: https://academic.oup.com/rfs/article-abstract/1/3/195/1580239
• Fama, E. & French, K. (1988), Dividend Yields and Expected Stock Returns, Journal of Financial Economics 22:3–25 — D/P forecasting power (R²) rises with horizon, substantial at 2–4 years for NYSE portfolios 1941–1986; exact R² varies by weighting/sub-period: https://www.sciencedirect.com/science/article/abs/pii/0304405X88900207

Disputes flagged: The empirical return-predictability result is robust at the aggregate level but undercuts the model's constant-discount-rate assumption rather than validating per-stock point valuations. Individual-stock DDM accuracy is not well supported and depends heavily on unobservable inputs.