Terminal value

You can't forecast a company's cash flows forever. So you forecast maybe five years in detail, then you need a number that stands in for every year after that. That number is terminal value, and here's the part beginners miss: it's usually most of your answer.

In a typical DCF, terminal value is 60% to 80% of the total enterprise value. Read that again. The five years you sweated over in the model are the minority of the output. One growth assumption or one multiple, sitting way out past the forecast, does most of the work. That's why interviewers press on it.

Two ways to compute it

There are exactly two methods you need to know. Learn both, and know when each is used.

Method 1: Gordon growth (perpetuity growth)

You take the final forecast year's free cash flow, grow it at a small perpetual rate forever, and capitalize it:

$TV = \frac{FCF_n \times (1+g)}{WACC - g}$

Here $FCF_n$ is the last year of your explicit forecast, $g$ is the long-run growth rate, and $WACC$ is your discount rate. The logic: a stream of cash growing at a steady rate forever is worth the next year's cash divided by (discount rate minus growth). It's the value of a growing perpetuity.

The one judgment call is $g$. It has to be a rate the company can sustain literally forever, so it's tied to the long-run growth of the whole economy. Think long-run GDP, roughly 2% to 3%.

Method 2: Exit multiple

Instead of assuming perpetual growth, you assume you sell the business at the end of the forecast at a market multiple. Take the final-year metric (usually EBITDA) and apply a multiple pulled from comparable companies:

$TV = EBITDA_n \times \left(\frac{EV}{EBITDA}\right)_{\text{exit}}$

If final-year EBITDA is $200 and comparable companies trade at 8x EV/EBITDA, terminal value is $1,600. This method grounds your answer in what the market actually pays today, which is why bankers often lean on it in practice.

The two methods should agree, roughly. If they don't, one of your assumptions is off.

Discount it back, always

Whichever method you use, terminal value lands at the end of your forecast period. It is a future number. So it gets discounted to present value alongside everything else:

$PV_{TV} = \frac{TV}{(1+WACC)^n}$

Then you add it to the present value of the explicit-year cash flows to get enterprise value.

Here's how the pieces stack up in a clean example. Say unlevered FCF in year 5 is $100, WACC is 10%, and perpetual growth is 2%:

| Component | Calculation | Value | | --- | --- | --- | | Terminal value (Gordon) | $100 \times 1.02 / (0.10 - 0.02)$ | $1,275 | | PV of terminal value | $1{,}275 / (1.10)^5$ | ≈ $792 | | PV of years 1 to 5 FCF | (sum of discounted forecast) | ≈ $340 | | Enterprise value | 792 + 340 | ≈ $1,132 |

Notice the terminal value is about 70% of enterprise value even after discounting. That's the whole point. Your DCF lives or dies on that one block.

Cross-check the two methods against each other

Because so much value rides on it, you don't trust a single terminal value. You compute it both ways and see if they line up.

The clean trick: take the exit-multiple terminal value and back out what perpetuity growth rate it implies. If your 8x exit multiple implies a 6% perpetual growth rate, something's wrong. No company grows 6% forever. That tells you the multiple is too rich or your forecast is too low. Run it the other way too: take your Gordon growth terminal value and see what exit multiple it implies, then sanity-check that against where comps actually trade.

Stress-test it, don't point-estimate it

Since one assumption drives the majority of value, the single terminal value number is fragile. A DCF output is a range, not a point. You build a sensitivity table: flex WACC across the top, growth rate (or exit multiple) down the side, and read enterprise value off the grid. That range is your real answer. If a half-point change in $g$ swings your valuation 30%, you say so. That honesty is what separates a thoughtful model from one that hides its own fragility.