A discounted cash flow states every assumption out loud, which is its great virtue and the reason it is so easy to abuse. The output moves enormously on inputs that cannot be observed.
14 min read · Level 3
The standard criticism of discounted cash flow is that it produces whatever answer the author wanted, and the standard defence is that this is a failure of the practitioner rather than the method. Both are true and both miss what is interesting. The method's real contribution is not the number at the end; it is that it cannot be performed without committing to a growth rate and a required return in writing, where somebody else can see them and disagree. That makes it the only common technique that leaves behind a record of what you believed, and a record you can be held to is worth more than an answer you cannot check. The uncomfortable consequence is that a discounted cash flow used to produce a single figure has thrown away the part that was valuable and kept the part that was not - which is roughly the opposite of how it is normally used, and why it deserves its reputation despite being the most honest method available.
A discounted cash flow values an asset as the money it will produce, with each future amount reduced to reflect the fact that money arriving later is worth less than money arriving now. The reduction factor is the discount rate.
It has three parts. A forecast of cash for some number of years, usually five or ten. A terminal value standing in for everything after that, most often calculated by assuming the final year's cash grows at a constant rate forever. And a discount rate applied to all of it.
The cash being discounted has to match what you are valuing. Free cash flow to the firm - calculated before interest - produces an enterprise value, from which net debt is subtracted to reach the owners' share. Cash flow after interest produces the owners' share directly. Mixing the two is the same family error as mixing enterprise and equity multiples, and it is just as invisible in the arithmetic.
Because it is the only common method that requires you to state what you actually believe. A multiple lets you avoid saying how fast you think the business grows or how risky you think it is; a discounted cash flow makes you type both numbers in. Even when the output is unreliable, the act of writing the assumptions down is worth the exercise on its own.
And because it is the only method that can price something with no comparables at all - an unusual business, a single asset, a project. Where relative valuation has nothing to compare against, this still works, with the honest caveat that the answer is only as good as the forecast underneath it.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
Harbour Press again. Its free cash flow to the firm for the coming year is 1,000,000, built as follows: operating profit of 1,600,000 taxed at 25% gives 1,200,000; add back the 400,000 depreciation charge to get 1,600,000; subtract 500,000 of capital spending and 100,000 absorbed by working capital, leaving 1,000,000.
Notice the tax is taken on operating profit, not on profit after interest. That is what makes this a figure for the whole business, which is what produces an enterprise value.
Assume it grows 8% a year for five years, then settles into a constant rate forever. Take a 10% discount rate and 2.5% perpetual growth as the base case.
Year Cash flow Discounted at 10% 1 1,000,000 909,091 2 1,080,000 892,562 3 1,166,400 876,334 4 1,259,712 860,400 5 1,360,489 844,757 ---------- Present value of years 1-5 4,383,144
The terminal value takes year 5's cash, grows it once at 2.5% to 1,394,501, and divides by the discount rate less the growth rate - 0.10 minus 0.025, which is 0.075. That gives 18,593,349, which is a value at the END of year five, so it is discounted back five years: 18,593,349 divided by 1.10 to the fifth, or 1.61051, giving 11,545,007.
Enterprise value: 4,383,144 plus 11,545,007, which is 15,928,151.
Now look at where that came from. 11,545,007 of the 15,928,151 is terminal value - 72.5% of the answer is a single formula about years six to infinity, and only 27.5% is the forecast anybody actually thought about.
So the honest thing to do is move the two inputs that formula depends on.
Raise the discount rate from 10% to 11%, leaving growth at 2.5%. The years one to five are now worth 4,267,510, the terminal value becomes 1,394,501 divided by 0.085, or 16,405,896, discounted back at 1.11 to the fifth to give 9,736,101. Enterprise value: 14,003,611 - a fall of 12.1%.
Now instead raise perpetual growth from 2.5% to 3.5%, leaving the discount rate at 10%. Year five's cash grows to 1,408,106, divided by 0.065 gives 21,663,170, discounted back to 13,451,124. Enterprise value: 17,834,268 - a rise of 12.0%.
One percentage point, in either direction, on either input, moves the answer about 12%. Now move both, one point each, in defensible directions:
growth 1.5% growth 3.5% rate 9% 16,470,000 21,143,000 rate 11% 12,894,000 15,409,000
The top corner is 21,143,254 and the bottom is 12,893,770. The top is 1.64 times the bottom - a 64% spread, produced entirely by moving two unobservable inputs by one percentage point each. Nobody made an error, and nobody was being unreasonable at any corner.
That is what the method is telling you, and the table is the output. The single figure of 15,928,151 in the middle of it is the least informative thing on the page.
The first mistake is presenting the base case as the answer. The table is the answer; the base case is one cell of it.
The second is a terminal growth rate above the long-run growth of the economy the business operates in. A business growing forever at 6% while its economy grows at 2% eventually becomes the entire economy, which is a claim nobody would make out loud and many spreadsheets make silently.
The third is mismatching the cash flow and the discount rate. Cash before interest is discounted at a rate reflecting the cost of both debt and equity, and produces an enterprise value. Cash after interest is discounted at the cost of equity alone and produces the equity value directly. Using one with the other is wrong by an amount nobody notices.
The fourth is a forecast that is really an extrapolation. Five years of 8% growth typed into a column is an assumption whether or not anybody treats it as one, and a valuation whose first five years are a straight line has done its thinking only in the terminal formula.
The fifth is adjusting the discount rate until the answer matches the price. It is the easiest input to move, it is unobservable, and moving it is indistinguishable from analysis - which is precisely why the rate should be written down and justified before the first cash flow is entered.
Before interest, producing an enterprise value, or after interest, producing the equity value. Write the choice down, because every later step depends on it and the error is invisible once the numbers start flowing.
Volume, price, margin, capital spending, working capital. A forecast built from things that could be checked against reality can be argued about; a single growth rate applied to a total cannot, which is why it is so often used.
State what return you require and why - the risk-free alternative available to you, plus what this asset's uncertainty is worth. Doing it first is the entire protection against the rate becoming the dial that produces the desired answer.
It cannot exceed the long-run growth of the economy the business sells into for very long without implying the business eventually becomes that economy. Write the figure you used and the reason, and note that anything above about 3% is a strong claim rather than a neutral default.
Divide the discounted terminal value by the total. For Harbour Press that is 72.5%. Above roughly two thirds, the valuation is mostly a formula about a future nobody can see, and the conclusion should be reported that way.
Two inputs, two values each, four corners. Do it before reaching a conclusion, because a table built afterwards tends to be centred on the answer you already have rather than on the assumptions you would actually defend.
Find the combination of discount rate and growth that produces the price being asked. Then ask whether you believe that combination. This is usually more useful than the forward calculation, because it converts an abstract disagreement about value into a concrete question about two numbers.
When the asset produces reasonably predictable cash, when there are no good comparables, and whenever you want to know what assumptions a market price is carrying.
Avoid it as a primary method on businesses whose cash flow five years out is genuinely unknowable - early-stage companies, single-product businesses facing a technology shift, anything whose survival is the open question. The arithmetic will still produce a number, and that is exactly the danger: precision applied to a forecast nobody believes reads as rigour and is not.
A shape that appears repeatedly in valuations built under a deadline is a discount rate that changed late in the process. It is the easiest input to move, it cannot be checked against anything observable, and a change to it looks exactly like a refinement of the risk assessment. Because the terminal value carries most of the answer and the rate sits in that formula's denominator, a shift of half a point produces a large movement in the conclusion for a change that can be described in a single defensible sentence.
Fix and justify the discount rate before the first cash flow is entered, and date the justification. A rate revised after the answer appeared is not a risk assessment; it is a steering wheel.
Forecasts in published models frequently show a constant growth rate across every year of the explicit period, which is a strong claim presented as a neutral default: real businesses have good years and bad ones, and a smooth line asserts that this one will not. The reason it is so common is structural rather than lazy - a constant rate is the only assumption that requires no specific knowledge about the business, so it is what remains when the forecast is built by somebody who has none.
A flat growth line across the explicit forecast is a signal about how much was known, not about how the business will behave. Where you see one, the real analysis is all in the terminal formula.
Create a free account to read the rest and apply it to your own investing.