DCF Valuation Limitations: Why DCFs Can Be Wrong
By Brian McCormick ·
A DCF can be completely correct mathematically and still give you the wrong valuation.
A DCF is ultimately a long series of assumptions about how a business will develop. Revenue growth, margins, capital spending, dilution, taxes, working capital and eventually the value of a company many years into the future all have to be estimated.
Get enough of those assumptions slightly wrong and the final fair value can be very wrong. In some ways, the most dangerous DCF is not a sloppy one. It is a clean, sophisticated looking model built on the wrong assumptions.
A DCF does not calculate the future
The basic logic behind discounted cash flow valuation is sound.
A business is worth the cash it can ultimately generate for its investors. Cash received far in the future is worth less than cash received today, so a DCF estimates those future cash flows and discounts them back into today's dollars.
Before you can discount anything, though, you have to decide what those future cash flows will be.
Suppose a company generates $1 billion of free cash flow today. You might forecast $1.1 billion next year, $1.21 billion the year after and $1.33 billion the year after that.
The spreadsheet can discount those numbers with precision, but it has no idea whether the company will actually grow cash flow by 10% each year. That growth rate came from your assumptions.
The model is not independently discovering what a company is worth. It is showing you the valuation implied by a particular set of beliefs about the company's future, and those assumptions are doing much of the work.
If you want to see this in practice, you can change these assumptions yourself in the StockUnlock DCF Calculator and see how they change the resulting valuation.
Small growth mistakes become large valuation mistakes
Growth is usually one of the first assumptions investors focus on, and for good reason. A fairly small difference compounds into something much larger when you stretch it across many years.
Imagine two forecasts for the same business.
In one case, free cash flow grows 8% annually for ten years. In another, it grows 12%.
Starting from $100 million of free cash flow, the company would be producing roughly:
- 8% growth: $216 million after ten years
- 12% growth: $311 million after ten years
Four percentage points of additional annual growth produces about 44% more cash flow by year ten.
This is one reason a DCF can give investors false confidence. Maybe you forecast 18% growth for the next five years because management has recently been growing around that rate. I care much less about whether 18% looks reasonable in a spreadsheet than what would physically have to happen inside the business to produce it.
I would want to understand how many additional customers the company needs, how much each customer needs to spend, whether the market can support that level of growth and whether pricing can hold as competitors respond. I also want to know how much investment the company has to make along the way, because answering those questions is a large part of the valuation work.
Revenue growth is not free cash flow growth
Even a good revenue forecast can produce a bad DCF if the model gets the economics underneath that wrong.
Say a company grows revenue 20%. Maybe getting there requires a large increase in sales spending, new factories, more inventory, better payment terms for customers or heavier R&D investment.
The revenue growth is real, but a lot more cash may have needed to go back into the business to make it happen.
Growth and reinvestment are economically connected. The faster a company expands, the more it often needs to put back into the business, although the amount varies enormously by business model.
A software company may be able to add customers without building much additional physical infrastructure. A manufacturer could need billions of dollars of new plants and equipment to support the same percentage increase in sales. Both companies can report 15% revenue growth while having completely different cash economics.
When I model growth, I also want to understand what has to be reinvested to create it. Otherwise the DCF can assume growth without fully accounting for what the company has to spend to achieve it.
Cyclical businesses can look cheapest near the top
Cyclical companies create a different problem because the number you are starting from may not be normal in the first place.
Take a commodity producer during a period of unusually high prices. Revenue is strong, margins are elevated and free cash flow can look spectacular. The stock may even trade at what appears to be a very cheap multiple of current cash flow.
Carry those economics forward in a DCF and you may be treating a peak year like an ordinary one.
Near the bottom of the cycle, the same modeling mistake works in reverse. Demand collapses, margins compress and free cash flow disappears. A perfectly viable business can suddenly look terrible if today's depressed financials become the foundation for a long term forecast.
This is why I would rather understand a cyclical company's economics across an entire cycle than simply take the latest year's numbers and start growing them. Normalizing earnings or cash flow is commonly used in cyclical valuation because the current period may bear little resemblance to a normal one.
The useful starting point may not be today's cash flow at all. It may be an estimate of what the business can earn across good years, bad years and everything between them. That estimate is harder to come up with, but worth the effort.
Margin assumptions can do more work than growth
Revenue growth gets a lot of attention in DCF models. Margin expansion can change the answer just as dramatically.
Imagine a business with a 10% free cash flow margin today. Your model assumes revenue eventually doubles and FCF margin gradually rises to 20%.
In that scenario, free cash flow quadruples rather than doubles. The revenue assumption may look perfectly defensible while a large portion of the upside is actually coming from what you assumed about margins.
Sometimes that is justified. A young software company may have large fixed costs that become smaller relative to revenue as it scales. A retailer might gain purchasing power. Other businesses can spread infrastructure, corporate costs or product development across a larger revenue base.
The model still needs a reason for the expansion, though. I want to understand which costs actually become more efficient, which ones rise with revenue, whether competition forces some of those savings back into pricing or spending, and whether the company needs to keep investing aggressively to produce the growth assumed.
Terminal value can dominate the entire DCF
A ten year DCF may contain detailed estimates for revenue, margins, capital expenditures and cash flow in each of the next ten years.
Then year eleven arrives, and the company does not vanish. The model therefore needs to account for all the cash generated after the explicit forecast period through its terminal value.
For many DCFs, terminal value represents a large part of the entire valuation.
There is something slightly strange about that. You can spend hours refining estimates for the next two or three years, only for a huge portion of today's calculated value to depend on a simplified assumption about what the company will look like after even your least certain explicit forecast.
Using the perpetual growth method, terminal value depends heavily on the relationship between the discount rate and perpetual growth rate. As those two numbers get closer together, terminal value rises rapidly, so even relatively small changes can matter.
Moving perpetual growth from 2% to 3%, for instance, can have a much larger effect on valuation than the one percentage point change might suggest.
Stock based compensation can make free cash flow look better than the economics
Stock based compensation commonly appears as a noncash adjustment when reconciling earnings to operating cash flow.
From an accounting perspective, that treatment makes sense, but shareholders bear an economic cost.
New shares dilute existing owners. A company that wants to offset the dilution can repurchase stock instead, but now real cash has to leave the business. Often the eventual cost shows up through some combination of both.
Suppose a company reports $500 million of free cash flow while issuing substantial equity compensation every year. Capitalizing the entire $500 million as though today's shareholders will receive all of that future value, while assuming the effect of SBC somehow disappears, is hard to justify.
The company may generate more cash over time and still leave each current share with a smaller claim on it. Alternatively, some of that cash may have to be spent repurchasing shares just to keep the share count from climbing.
There are different ways to handle SBC in valuation. You can reflect the economic cost of expected future compensation in your forecast, or model future awards, dilution and any cash needed to offset it.
What I would not do is add SBC back to free cash flow forever and then assume it has no future effect on shareholders.
Capital expenditures can be temporarily misleading
Free cash flow has its own measurement issues.
The SEC notes that free cash flow does not have a uniform definition. A common calculation starts with operating cash flow and subtracts capital expenditures.
The logic behind subtracting capex is straightforward. A business often has to keep putting money into long lived assets to maintain operations and support future growth, but annual spending can be lumpy.
One company might be in the middle of constructing a facility that will support production for the next decade. Another may be postponing maintenance and replacement spending, making current free cash flow look unusually good even though those bills have not disappeared.
The latest year's FCF therefore isn't automatically a normalized number.
For a capital intensive company, I want some sense of how much spending maintains the existing business, how much expands capacity, whether current project timing is unusual and what the business is likely to require over a longer period.
Two companies can report identical free cash flow today while being in completely different economic positions.
Working capital can temporarily inflate cash flow
Working capital can create similar distortions without anything dramatic changing in the income statement.
Maybe a company collects customer payments faster, takes a little longer to pay suppliers or sells down inventory that had built up during the prior year. Cash flow improves, but the benefit may not be repeatable.
Customers just cannot pay infinitely faster, suppliers will not wait forever and inventory cannot be reduced below what the business needs to operate.
Changes in working capital can move operating cash flow substantially from one period to another. Looking at one unusually favorable cash conversion period and assuming it repeats indefinitely can therefore distort a DCF by capitalizing a temporary benefit far into the future.
Looking across several years usually gives me a better idea of what the business is actually capable of producing.
Acquisitions can hide the real cost of growth
Acquisitive companies create another wrinkle.
Suppose a business regularly buys smaller competitors. Those deals eventually increase revenue, earnings and operating cash flow.
The cash portion of the purchase price generally appears in investing activities rather than as an operating expense. You can see this directly in company cash flow statements, where cash paid for business combinations is reported within investing activities.
So a simple free cash flow calculation might show:
Operating cash flow
minus capital expenditures
equals free cash flow
while excluding the cash repeatedly spent buying the businesses responsible for some of the growth.
Stock funded acquisitions avoid that acquisition date cash outflow, but the cost has not vanished because existing shareholders may instead absorb it through dilution.
A company that has effectively made acquisitions part of its growth engine deserves different treatment, and I would not value that growth exactly like organic growth while assuming the assets producing it were acquired for free.
The discount rate cannot rescue bad assumptions
Because changing the discount rate can move a DCF so much, it becomes an easy place to compensate for discomfort elsewhere in the model.
Maybe growth feels a little aggressive, so you raise the discount rate. Or the stock looks expensive at 10%, so you see what happens at 8%.
Growth, margins, reinvestment and dilution describe the business you expect to exist in the future. The discount rate addresses what those future cash flows are worth today given their risk. They are related parts of a valuation, but they are not interchangeable.
If the operating assumptions are wrong, changing the discount rate until the answer feels reasonable just moves the problem around.
A single DCF hides the range of possible outcomes
One of the biggest issues with presenting a DCF as one fair value is the precision of the answer.
The model spits out $84.37 per share, even though reality is not going to unfold to two decimal places.
Maybe competition intensifies and the business ends up being worth something closer to $40. Perhaps the operating assumptions work reasonably well and $80 is in the right neighborhood. Better margins, longer lasting growth or stronger economics could support something far higher.
The future exists as a range even though the DCF often presents it as a single answer.
I would rather model several versions of the business.
A lower case should describe an actual weaker operating outcome, not simply the same business with a higher discount rate. The base case should represent what I genuinely think is most likely. An optimistic case needs to show what would have to go right to justify the higher valuation.
The best use of a DCF is not predicting one exact fair value
None of these problems make DCFs useless. A good DCF forces your assumptions into the open.
Working backward from the current stock price can sometimes be more interesting than solving for one supposedly correct fair value. Instead of asking the model what the stock is worth, ask what would have to happen for today's price to make sense.
You can look at how much growth is already priced in, what margin the business would need to reach, how much cash has to be reinvested along the way and how much of the valuation comes from terminal value.
I would also want to see what continued dilution does to value per share and how much of the apparent upside disappears if the company turns out to be merely good rather than exceptional.
What investors should remember
Discounted cash flow theory is not usually why a DCF goes wrong. The future business you modeled can simply end up looking very different from the one that actually arrives.
Growth slows, margins top out sooner than expected, a cyclical peak reverts toward normal, or acquisitions make growth more expensive than the reported free cash flow suggests.
So I would never look at a DCF fair value without also looking closely at the assumptions that produced it. Those assumptions are the most important part.
You can use the Stock Unlock DCF Calculator for free here, which also includes built in analyst estimates.