Dividend discount calculator

Because the Gordon terminal value TV = D × (1+g) / (Ke − g) has a vertical asymptote at g = Ke — as growth approaches the cost of equity the output explodes toward infinity, which is a math artefact, not real value. Most free DDM calculators happily emit that divergent number and rely on you noticing it is nonsense. This one structurally forces initial growth into a safe band — at least 1pp above terminal, and at most the midpoint of Ke and terminal minus half a point — so the math can never collapse. The initialGrowth field in the audit always shows the rate actually used: if it does not match what you typed, the clamp activated and you were running an aggressive assumption.

It is normal for DDM, not a bug. The explicit horizon is only 5 years and dividends grow slowly, so the Gordon terminal piece routinely carries 60–80% of the value even for genuinely-valuable dividend stocks. That is exactly why the reliability penalty here triggers at 70%, not at 60% like the EPS DCFs — penalising at 60% would mark every DDM run as low-reliability and tell you nothing. The 70% threshold flags only the most terminal-dominated cases, where even small changes to Ke or terminal growth swing the answer dramatically.

Dividends are a sticky, slow-moving cash flow — mature payers raise dividends in single digits per year and the path is highly predictable. A longer explicit horizon adds no real information; it just gives the terminal value an extra discount while inflating its share of the total relative to Stage I, making the model even more terminal-dominated. That is the opposite of what you want from a model whose lead warning is about terminal-value dominance. Five years is the calibrated horizon for a dividend stream.

It is strictly required for two archetypes: mature dividend payers (utilities, telecoms, mature consumer staples — where the dividend is the primary equity-return mechanism) and REITs (which distribute ~90% of taxable income as dividends by structure, so a dividend-anchored lens is uniquely apt). For every other archetype the calculator will still compute a value if the company happens to pay a dividend, but the synthesis layer does not expect DDM in the composite — read it as informational only. Notably, DDM is required for REITs precisely where the earnings-based DCFs are excluded.

The model is excluded with reason no_dividend_or_ke. There is no way to anchor a dividend discount model on zero dividends — the same exclusion fires if Ke is missing or if Ke is not greater than terminal growth, because all three break the Gordon math in the same way and the user-facing fix is the same. For a non-payer, use the Discounted Earnings, FCFF DCF, or EV/Revenue lens instead. There is no FAILED state on this model: once applicability passes, the clamp guarantees the math returns a sane number.