The dividend discount model takes the strictest possible view of what a share is: a claim on the dividends it will pay. Forecast those payments, grow them at an assumed rate, discount them back and add them up — that sum is the value. Earnings the company retains do not appear anywhere in the arithmetic, and neither does cash returned by buying shares back.
Its best-known form is the Gordon growth model, which assumes one steady growth rate for ever and collapses the whole infinite series into a single division. The same formula sits at the heart of most terminal value calculations, which is worth noticing — a DDM is, in effect, a terminal value applied to the whole life of the share.
D is the annual dividend per share, g the assumed growth rate and k the discount rate. Note the constraint hiding in the first form: g must stay below k, or the denominator goes to zero or negative and the model returns nonsense rather than a large number.
A company pays $3.00 a share a year, the dividend is assumed to grow at 5%, and the investor requires 9%:
Now move growth to 6%: the answer becomes $3.18 ÷ 0.03 = $106.00, a third higher on one percentage point. The closer the growth assumption creeps to the discount rate, the more violently the model responds — which is why a DDM is best read as a range across several growth rates rather than as one figure.
Test the growth rate against the company’s ability to fund it. A dividend cannot outgrow the earnings behind it for long, so a growth assumption above the sustainable growth of the business is really a forecast of a rising payout ratio, which has a ceiling.
Read it beside the payout ratio and the cash flow. A high, steadily growing dividend covered by two thirds of earnings is a different proposition from the same dividend covered by 110% of them, and the model itself cannot tell them apart.
In the app, the Dividend Discount Model is one of the four methods on the Intrinsic Value tab, and it is also the one method that can state a target yield — what the declared dividend would yield at the value it computes.
It is the right tool for a mature, reliably paying business with a long dividend record and a board that treats the payment as a commitment — utilities, consumer staples, large insurers, dividend aristocrats. For that kind of company it has a virtue no other method has: it values only what actually reaches a shareholder, with no step where retained cash is assumed to be well spent.
It has nothing whatever to say about a company that pays no dividend, and very little about one that pays a token amount. It is blind to buybacks, which for many large companies now return more than dividends do, and blind to the value of cash retained and reinvested. Its sensitivity near g = k makes it easy to produce any answer you like by moving the growth rate a fraction, and a single dividend cut — a board decision, not a trend — invalidates the whole series at once.