Dividend Discount Model (DDM)

The dividend discount model is a method of valuing a share of stock by treating it as a claim on a stream of future dividend payments and computing the present value of that stream. In its general form it makes no assumption about the pattern of the dividends: each expected payment is discounted back to today at a required rate of return and the results are added. In practice it is almost always used in a simplified constant-growth form, in which dividends are assumed to grow at a single rate forever and the infinite sum reduces to a single division: the value of a share equals next year's expected dividend divided by the required return minus the growth rate.

The constant-growth version is commonly called the Gordon growth model, after the economist Myron J. Gordon, whose 1956 paper with Eli Shapiro in Management Science on the required rate of profit and 1959 paper in the Review of Economics and Statistics on dividends, earnings and stock prices are the work that name points to. The two names are not interchangeable: the dividend discount model is the general idea, and the Gordon growth model is the one special case in which growth is constant. Neither phrase appears anywhere in the Code of Federal Regulations, and the SEC's investor glossary defines neither, which is worth knowing when comparing two published estimates: the label does not fix the inputs.

The model is an argument about what a share is, and that is its real contribution. A shareholder's only contractual claim on a going concern is whatever the company chooses to distribute. Retained earnings matter to a shareholder because they are expected to raise future distributions, not on their own account. Taking that seriously produces the model, and it produces a discipline with it: anybody quoting a price target is implicitly asserting a distribution path and a required return, whether or not they say so.

The arithmetic has three inputs and one of them is a forecast to infinity. Next year's dividend can be estimated from a declared rate. The required return is a judgment, built from a safe rate plus a premium for equity risk, and it is a subject of its own. The growth rate is the problem: the constant-growth form does not ask for a growth rate for the next five years, it asks for the rate that will hold forever. No company has ever had one, so what the input really represents is a long-run average, and long-run averages of company dividend growth are not observable in advance.

Two structural failures are worth stating plainly rather than glossing. First, a company that pays no dividend values at exactly zero under this model, because zero divided by anything is zero. That is not a defect in the arithmetic but a statement about its domain: the model applies to companies that distribute cash and says nothing useful about companies that do not, which excludes much of the market. Second, the denominator is the required return minus the growth rate, so if the assumed growth equals the required return the expression is undefined, and if growth exceeds it the result is a negative value, which is meaningless as a share price. That is not a warning to be worked around by choosing a smaller growth rate; it is the model telling the user that a firm cannot be assumed to grow faster than its cost of capital forever.

The sensitivity is the headline, and it is easy to underrate. Because the denominator is a difference between two numbers of similar size, a small move in either one is a large proportional move in the denominator and therefore in the answer. A Federal Reserve Board economist made the same point from the other direction in a 2014 FEDS Note, building a simple constant-growth present-value framework for the aggregate corporate sector and concluding that "even a fairly modest decline in discount rates of 1 percentage point can have a marked effect on valuation ratios." The consequence for this model is that its output should be read as a range produced by a range of assumptions. A single number carries false precision, and the worked example below shows why: two one-point adjustments, each individually defensible, double the estimate.

Where the model is genuinely useful. Not as a price, but as an inversion. Given a share's actual market price and its current dividend, the formula can be solved for the growth rate the market appears to be assuming, or for the return an investor would earn if that growth materialized. Framed that way it stops being a forecast and becomes a question: is the growth implied by this price plausible? That is a checkable question, and it is the version of the model a careful user keeps.

Estimate next year's dividend per share, choose a required annual return, and choose a long-run dividend growth rate below that return. Divide the dividend by the difference between the two rates. The result is an estimated value per share, which is then compared with the market price.

A hypothetical example, and the point is what happens when the inputs move. A company is expected to pay $2.20 per share in dividends next year. An investor requires an 8 percent annual return and assumes dividends grow 4 percent a year indefinitely. The estimated value is $2.20 ÷ (0.08 − 0.04) = $2.20 ÷ 0.04 = $55.00 per share.

Now raise the growth assumption by a single percentage point, to 5 percent, and change nothing else: $2.20 ÷ (0.08 − 0.05) = $2.20 ÷ 0.03 = $73.33. The estimate rose about 33 percent on a one-point change in a number nobody can observe.

Now also lower the required return by a point, to 7 percent: $2.20 ÷ (0.07 − 0.05) = $2.20 ÷ 0.02 = $110.00, twice the original estimate. Two one-point adjustments, both individually defensible, doubled the answer.

And push growth to 9 percent, above the required return: $2.20 ÷ (0.08 − 0.09) = $2.20 ÷ (−0.01) = −$220.00. A negative share value is the arithmetic refusing an impossible assumption rather than a result to be interpreted.

Pros

• Rests on a defensible idea about what a shareholder actually owns, which is a claim on distributions.
• Simple enough to compute by hand, so its assumptions cannot hide inside a model nobody inspects.
• Can be inverted: given a market price, it reveals the growth rate or the return the price implies, which is a testable question rather than a forecast.
• Forces the required return into the open, where it can be argued about.

Cons

• Returns zero for any company that pays no dividend, so it does not apply to a large part of the market.
• Requires a growth rate that holds forever, which is not a forecast anybody can make.
• Breaks down completely when the assumed growth reaches the required return, and the constant-growth form has no way to handle a company whose growth is high now and lower later.
• Extremely sensitive: one-point changes in either rate can move the estimate by a third or more, so it cannot settle a disagreement about value.
• Says nothing about buybacks, which are the other way companies return cash and which the model's dividend input does not capture.