Dividend Discount Model Explained: DDM Formula, Gordon Growth, Multi-Stage, and Limitations

May 9, 2026 · guides · 11 min read

Dividend Discount Model Explained: DDM Formula, Gordon Growth, Multi-Stage, and Limitations

The dividend discount model is one of the oldest and most theoretically grounded methods for valuing stocks. Its core premise is simple: the fair value of a stock is the sum of all dividends it will ever pay, discounted back to today's dollars.

When applied rigorously, accounting for growth rates, multi-phase dividend trajectories, and the required rate of return, the dividend discount model (DDM) becomes a precise tool for estimating intrinsic value for dividend-paying businesses. It is used by equity analysts, pension fund managers, and value-oriented retail investors alike.

This guide covers the full DDM framework: the one-period model, the Gordon Growth Model, multi-stage variants, the H-model, worked numerical examples, and where the model breaks down.

What Is the Dividend Discount Model?

The dividend discount model (DDM) estimates the intrinsic value of a stock as the present value of all expected future dividends. It is built on the principle that a stock's value is defined entirely by the cash it returns to shareholders, and dividends are the most direct form of that return.

The theoretical foundation comes from John Burr Williams, who articulated it in 'The Theory of Investment Value' in 1938. Myron Gordon later formalized the constant-growth variant, which became known as the Gordon Growth Model.

The general DDM formula is:

Intrinsic Value = D1 / (r - g)

Where D1 is the dividend expected next period, r is the required rate of return, and g is the expected constant dividend growth rate.

This formula assumes dividends grow at a constant rate forever. That assumption is strong, and most of the model variants discussed below exist to relax it.

The One-Period DDM

The simplest version values a stock you plan to hold for exactly one year, collecting one dividend and then selling.

Formula: P0 = (D1 + P1) / (1 + r)

Where P0 is today's intrinsic value, D1 is the year-end dividend, P1 is the expected sale price, and r is the required rate of return.

Example: A stock pays $2.00 at year end, you expect to sell at $50.00, and your required return is 10%.

P0 = (2.00 + 50.00) / 1.10 = $47.27

If the current market price is below $47.27, the model suggests potential undervaluation. If above $47.27, it may be expensive relative to your assumptions.

The one-period model is limited because it shifts the problem: now you need to estimate P1, which requires another valuation. Perpetual models avoid this by assuming dividends continue indefinitely.

The Gordon Growth Model (Constant Growth DDM)

The Gordon Growth Model is the most widely used DDM variant. It assumes dividends grow at a constant rate indefinitely, producing a clean closed-form solution.

Gordon Growth Model Formula

P0 = D1 / (r - g)

Or equivalently, using the current dividend D0:

P0 = D0 x (1 + g) / (r - g)

Where:

Key Constraint

The Gordon Growth Model only works when r > g. If the growth rate equals or exceeds the required return, the denominator approaches zero or goes negative and the formula produces undefined or nonsensical results. In practice, no company can grow faster than the overall economy indefinitely, so g is typically assumed to converge toward a long-run nominal GDP growth rate of 2% to 4% in perpetuity.

Gordon Growth Model: Worked Example

A utility company currently pays an annual dividend of $3.20 per share. Dividends have grown at approximately 4% annually for the past decade, and you expect that to continue. Using CAPM, you estimate your required return on this stock at 9%.

Step 1: Estimate the next dividend. D1 = 3.20 x (1 + 0.04) = 3.20 x 1.04 = $3.328

Step 2: Apply the Gordon Growth Model. P0 = 3.328 / (0.09 - 0.04) P0 = 3.328 / 0.05 P0 = $66.56

If the stock is currently trading at $60.00, it may correspond to potential undervaluation relative to the model estimate. If it is trading at $75.00, it may be pricing in a higher growth assumption or a lower required return than you are using.

Sensitivity to Inputs

The Gordon Growth Model is notoriously sensitive to small changes in r and g because both appear in the denominator.

Using the same D1 = $3.328:

A one-percentage-point change in either input shifts the estimated value by 20% to 40%. This is not a flaw unique to DDM; it reflects genuine uncertainty about long-run returns and growth. But it means DDM outputs are best read as ranges, not point estimates.

Required Rate of Return: How to Estimate r

The required rate of return in DDM is the minimum return an investor demands for holding the stock given its risk level. The standard approach is to use the Capital Asset Pricing Model (CAPM):

r = Rf + Beta x (Rm - Rf)

Where:

Example: Beta = 0.75, 10-year Treasury = 4.5%, equity risk premium = 5%.

r = 4.5% + (0.75 x 5.0%) = 8.25%

For lower-beta, dividend-paying stocks such as utilities and consumer staples, required returns typically fall in the 7% to 10% range. Higher-beta stocks demand higher required returns, which makes DDM less useful for them since their dividends (if any) would be small relative to the discount rate.

Sustainable Growth Rate: How to Estimate g

The sustainable growth rate is the rate at which a company can grow its dividends without changing its capital structure. The standard formula is:

g = ROE x Retention Ratio

Where the retention ratio equals 1 minus the payout ratio.

Example:

A company earns a return on equity (ROE) of 14% and pays out 60% of earnings as dividends.

Retention ratio = 1 - 0.60 = 0.40

g = 0.14 x 0.40 = 0.056 = 5.6%

This means the company can sustainably grow dividends at approximately 5.6% per year given its current profitability and payout policy. If the company tries to grow faster, it would need to either raise external capital or cut dividends.

For the Gordon Growth Model, this sustainable growth rate is a useful anchor for g. But it is still a historical measure: if ROE or the payout ratio changes, g changes with it.

Multi-Stage DDM: Two-Stage and Three-Stage Models

Most dividend-paying companies do not grow at a constant rate forever. Young or mid-cycle dividend growers may increase payouts rapidly for a number of years before settling into a stable, slower growth rate. Multi-stage DDM models capture this by dividing the forecast into two or three distinct phases.

Two-Stage DDM

The two-stage DDM assumes:

Formula:

P0 = Sum of [ D0 x (1+g1)^t / (1+r)^t ] for t = 1 to n, plus the terminal value

Terminal Value at year n = D(n+1) / (r - g2)

Present value of terminal value = [ D(n+1) / (r - g2) ] / (1 + r)^n

Two-Stage DDM: Worked Example

A company currently pays a $1.50 annual dividend. It is expected to grow dividends at 12% per year for 5 years as it matures, then settle into a 3.5% perpetual growth rate. Your required return is 9%.

Phase 1: Calculate dividends for years 1 through 5

Year 1: 1.50 x 1.12 = $1.68 Year 2: 1.68 x 1.12 = $1.8816 Year 3: 1.8816 x 1.12 = $2.1074 Year 4: 2.1074 x 1.12 = $2.3603 Year 5: 2.3603 x 1.12 = $2.6435

Discount each back at 9%:

Year 1 PV: 1.68 / 1.09 = $1.5413 Year 2 PV: 1.8816 / 1.09^2 = 1.8816 / 1.1881 = $1.5844 Year 3 PV: 2.1074 / 1.09^3 = 2.1074 / 1.2950 = $1.6273 Year 4 PV: 2.3603 / 1.09^4 = 2.3603 / 1.4116 = $1.6721 Year 5 PV: 2.6435 / 1.09^5 = 2.6435 / 1.5386 = $1.7181

Sum of Phase 1 PVs = $8.1432

Phase 2: Terminal Value

Year 6 dividend = 2.6435 x 1.035 = $2.7360

Terminal value at end of year 5 = 2.7360 / (0.09 - 0.035) = 2.7360 / 0.055 = $49.745

PV of terminal value = 49.745 / 1.09^5 = 49.745 / 1.5386 = $32.33

Intrinsic Value Estimate:

P0 = 8.14 + 32.33 = $40.47

If the stock is currently trading at $35, it may correspond to potential undervaluation under these model assumptions.

Three-Stage DDM

The three-stage DDM adds a middle transition phase between high growth and stable growth. This is useful for companies moving from rapid expansion toward maturity, where growth decelerates gradually rather than snapping from 12% to 3.5% overnight.

Phase 1: High growth (e.g., 5 to 7 years at g1 = 15%) Phase 2: Transitional growth (e.g., 5 years declining linearly from 15% to 4%) Phase 3: Stable terminal growth (g3 = 4% forever)

The math is the same structure as two-stage: sum the discounted dividends for each year in phases one and two, then apply a terminal value formula using the stable phase growth rate.

Three-stage models are more realistic for most companies, but they introduce more assumptions and therefore more estimation error.

The H-Model: A Simpler Multi-Stage Approximation

The H-model was introduced by Russell Fuller and Chi-Cheng Hsia in 1984 as a computationally simpler approximation of the two-stage DDM. It assumes dividend growth starts at a high rate and declines linearly to a stable long-run rate over a period of 2H years, where H represents the half-life of the high-growth period.

H-Model Formula:

P0 = D0 x (1 + gL) / (r - gL) + D0 x H x (gS - gL) / (r - gL)

Where gL is the long-run stable growth rate, gS is the initial short-run growth rate, and H is half the length of the high-growth period.

Example: D0 = $2.00, gS = 12%, gL = 4%, r = 9%, H = 5 (implying a 10-year transition period).

First term: 2.00 x 1.04 / (0.09 - 0.04) = 2.08 / 0.05 = $41.60

Second term: 2.00 x 5 x (0.12 - 0.04) / (0.09 - 0.04) = 2.00 x 5 x 1.6 = $16.00

P0 = $57.60

The H-model is less precise than the full two-stage model but faster to apply as a screening tool.

DDM vs DCF: Which Should You Use?

The dividend discount model and discounted cash flow analysis both estimate intrinsic value as the present value of future cash flows, but they differ in what cash flows they count.

DDM uses dividends, which are actual cash distributions to shareholders. DCF typically uses free cash flow to equity (FCFE) or free cash flow to the firm (FCFF), which captures all earnings available to shareholders regardless of whether they are paid out or reinvested.

Use DDM when:

Use DCF instead when:

Many analysts use both models as a cross-check. If DDM and DCF produce similar estimates, confidence in the valuation range is higher. Large divergences suggest the payout policy is misaligned with underlying earnings power.

Real Company Applications

Utility Companies

Regulated utilities are the canonical DDM candidate. Companies with rate-regulated cash flows pay large, predictable dividends. Their dividend growth tends to be slow and stable (3% to 5%) and their beta is low. The Gordon Growth Model with a carefully estimated r from CAPM produces reasonable intrinsic value estimates for these businesses.

Consumer Staples

Companies in consumer staples often have 20 to 40 year histories of dividend growth. Dividend aristocrats in this space have raised dividends every year for decades. The two-stage DDM is appropriate here: a slightly elevated near-term growth rate converging to a long-run sustainable pace.

REITs

Real estate investment trusts are legally required to distribute at least 90% of taxable income as dividends, making them natural DDM candidates. REITs are often valued using DDM or dividend yield methods rather than earnings-based multiples, since net income can be distorted by depreciation.

Banks

Large commercial banks have long payout histories and relatively predictable earnings. The DDM is most reliable for banks when they are operating near steady-state earnings with a stable payout ratio.

Key Limitations of the Dividend Discount Model

1. Only Works for Dividend-Paying Stocks

DDM cannot be applied to companies that pay no dividend. Amazon, Alphabet, Berkshire Hathaway, and most technology growth companies either pay no dividend or a minimal one. For these stocks, DDM produces no useful output and analysts must use DCF, earnings-multiple analysis, or other methods.

2. Extreme Sensitivity to Growth Rate Assumptions

As shown in the worked examples above, small changes in g or r produce large changes in estimated value. If you assume 5% growth instead of 4%, the Gordon Growth Model estimate can jump by 25% or more. This makes DDM results heavily dependent on assumptions that are themselves uncertain.

3. Constant Growth Rarely Holds

The Gordon Growth Model assumes dividends grow at a constant rate forever. No company actually does this. Management changes, competitive disruption, regulatory shifts, or economic cycles can alter payout policies abruptly. Multi-stage models improve realism but introduce more inputs to estimate.

4. Ignores Share Buybacks

Many mature companies return capital to shareholders through share repurchases rather than dividends. DDM ignores buybacks entirely. A company that buys back 5% of shares annually and pays a small dividend is creating substantial value for shareholders, but DDM captures none of the buyback return. The total yield model (dividends plus buyback yield as a proxy) is sometimes used to adjust for this.

5. Dividend Policy Can Be Disconnected from Earnings

A company can sustain or grow its dividend even as underlying earnings deteriorate by borrowing or drawing down cash reserves. DDM would value such a company as stable or growing when it is actually deteriorating. Always check the payout ratio and earnings trends before relying on DDM output.

6. Terminal Value Dominates

In multi-stage models, the majority of the intrinsic value typically comes from the terminal value in the stable phase. In the two-stage example above, $32.33 of the $40.47 total (roughly 80%) came from the terminal value. Small changes in the long-run growth rate assumption therefore have an outsized effect on the final estimate.

How Equity Rank Uses Dividend-Based Valuation

Equity Rank includes DDM analysis as one of multiple valuation methods within its SAVE score framework. The platform runs DDM, DCF, earnings multiples, asset-based valuation, and other methods simultaneously, synthesizing them into a composite score rather than relying on any single model.

For dividend-paying stocks, DDM inputs are derived from trailing dividend data, historical payout ratios, and sustainable growth estimates using ROE and retention ratios. If all methods converge on a similar intrinsic value range, that convergence provides higher model confidence than any individual estimate.

Summary: When the DDM Works and When It Does Not

The dividend discount model is a theoretically sound valuation framework with clear, defined applications. It works well for mature, dividend-paying businesses with stable payout policies: utilities, consumer staples, REITs, and large-cap financial institutions.

It struggles with growth companies and any business where the link between earnings and dividends is unstable. In those contexts, DCF or earnings-multiple approaches are better suited.

Used appropriately and combined with other valuation methods, DDM gives a mathematically rigorous estimate of what a dividend stream is worth. Understanding its assumptions is as important as applying its formula.