ValueWhy the last line of your DCF is most of the answer
A DCF forecasts cash flows for a handful of years, then has to account for everything after. That "everything after" is the terminal value, and it is not a rounding item: with typical inputs it represents 70-90% of the total enterprise value, as the share figure above shows with your own inputs.
The intuition is compounding in reverse. Even discounted at 9-10% a year, an infinite stream of growing cash flows adds up to more than a decade of explicit forecasts. This has an uncomfortable implication: your DCF is mostly a bet on steady-state assumptions (the perpetual growth rate and the discount rate), not on your careful year-by-year forecast. Anyone who shows you a DCF without stress-testing the terminal value is showing you the garnish, not the meal.
Method 1: Gordon growth (the perpetuity formula)
The Gordon method values the terminal period as a growing perpetuity:
TV = FCFn × (1 + g) / (r − g)
where FCFn is the final forecast year's free cash flow, g is the perpetual growth rate, and r is the discount rate. The formula is the closed form of an infinite geometric series: every future year's cash flow, grown at g and discounted at r, summed to infinity. It converges only if r > g; if growth matched or exceeded the discount rate, the business would eventually be worth more than the entire economy, which is the math telling you an input is wrong.
The strength of the method is internal consistency: it uses the same discount rate as the rest of your DCF, so the whole valuation speaks one language. Its weakness is that the answer is extremely sensitive to g, as the sensitivity table demonstrates.
Method 2: Exit multiple
The exit-multiple method sidesteps the perpetuity math and asks what the market would pay:
TV = EBITDAn × EV/EBITDA multiple
If comparable businesses trade at 10x EBITDA, the terminal value is ten times the final year's EBITDA. This is a market cross-check rather than a fundamental valuation: it imports the market's current pricing of similar assets into your model. That is both its appeal (grounded in observable transactions) and its flaw (if the whole sector is overvalued, your "independent" valuation inherits the bubble). It is also slightly circular for public-company valuation: you are valuing a stock partly by what the market pays for similar stocks.
The terminal growth trap
The single most common DCF error is an aggressive perpetual growth rate. A business cannot grow faster than the economy forever, or it eventually becomes the economy. Long-run nominal GDP growth in developed markets runs roughly 2-4%, and that is the ceiling for g in almost every case. Common traps:
- Carrying the forecast growth rate into perpetuity. Growing at 15% for five years is plausible; growing at 15% forever is fantasy. The two rates answer different questions.
- Forgetting inflation. A 2.5% perpetual growth rate is mostly inflation plus a little real growth. In real (inflation-adjusted) terms, mature businesses often grow at roughly 0%. If your DCF is in nominal terms, g should include inflation; if real, it should not.
- Ignoring fade. Competitive advantage erodes. Many practitioners fade the growth rate down over the forecast rather than dropping off a cliff into perpetuity, which is more realistic and usually lowers the valuation.
Reconciling the two methods
Compute both and compare. If the Gordon value is far above the exit-multiple value, either your perpetual growth is too generous, your discount rate too low, or the sector multiple is temporarily depressed. If the exit multiple is far above Gordon, the market may be pricing in growth your model does not, or the sector may be expensive. A large gap is not a tiebreaker to ignore; it is the model telling you where your assumptions disagree with the market, which is exactly the question worth answering before you invest.
Limitations
- Steady state may never arrive. Both methods assume the business reaches a stable, mature equilibrium. Cyclical businesses, disruptors, and declining industries may never settle, making any terminal value a rough guess.
- The discount rate is doing enormous work. A one-point change in WACC moves the terminal value by 15-25% in typical cases. If your WACC is a guess, your terminal value is a guess squared. Build it carefully with the WACC calculator.
- Terminal value inherits every upstream error. It is computed from your final-year FCF, which came from your growth assumptions, which came from your starting FCF. Errors compound forward, so the terminal value magnifies them.
Frequently asked questions
What exactly does terminal value represent?
Everything after your forecast ends: all cash flows from year n+1 to infinity, compressed into one number sitting at year n, then discounted back to today like any other future amount. It is not a separate asset; it is the tail of the same cash flow stream.
Which method do professionals use?
Most published DCFs show the Gordon growth method as primary, with an exit multiple as a sanity check, or vice versa in private equity where an actual sale at a multiple is the plan. Showing both and explaining the gap is better practice than picking one and hiding the other.
What perpetual growth rate should I use?
At or below long-run nominal GDP growth for the business's markets, typically 2-3% for developed economies. Higher is defensible only with a specific, durable reason, and anything above 4-5% deserves serious skepticism.
Why do the two methods give different answers?
They answer different questions: Gordon asks what the cash flows are worth at your required return; the exit multiple asks what the market pays for similar earnings today. When they disagree, the disagreement itself is informative, so investigate it rather than averaging it away.
Which terminal value method is better: Gordon growth or exit multiple?
Neither is strictly better. Gordon growth suits stable, mature businesses where a perpetual growth rate is defensible. The exit multiple is a useful cross-check based on what comparable companies trade for. If the two methods disagree wildly, your assumptions need another look.