Note · Opinion, not rated
Non-cash is not free.
Stock compensation is the largest expense that most investors have agreed to stop counting. It is paid in one of exactly two currencies, and neither of them is nothing.
The claim
When a company pays an employee in stock rather than cash, the accounting calls it a non-cash expense. That label is technically correct and it has done more damage to investor arithmetic than almost any other phrase in modern reporting, because a very large number of people have quietly translated non-cash into not real.
Here is the claim this note defends. Stock compensation is always paid. It is paid either by you, through a share count that rises so your slice of the company shrinks, or by the company, through cash spent buying those shares back so your slice does not shrink. There is no third option and there is no free version. The only genuine question is which pocket it comes out of, and that question is worth being precise about.
Where the cost actually lands
Take a company generating $1,000 million of free cash flow with 100 million shares outstanding at $100 a share. Headline free cash flow per share is $10.00. The company grants $200 million of stock compensation in the year.
Route one: let it dilute. The company issues $200 million of stock, which at $100 a share is 2 million new shares. Share count rises to 102 million, or 2.0%. Free cash flow per share falls to $9.80. The apparent cost is 20 cents a share.
Route two: buy it back. The company spends $200 million repurchasing the same number of shares so the count stays at 100 million. Free cash flow available to you is now $800 million, and free cash flow per share is $8.00. The cost is $2.00 a share.
Notice that route one looks ten times cheaper. That gap is the trap, and understanding why it is an illusion is the whole point of this note.
Why dilution looks cheap and is not
A buyback is a single year's cash leaving in a single year. Dilution is not a charge against one year of cash flow. It is a permanent transfer of a claim on every dollar the business will ever produce. You did not lose 20 cents once. You permanently surrendered 2.0% of everything that follows.
Price that properly and the two routes converge exactly. The company is worth $10,000 million. Handing over 2.0% of it transfers $200 million of value, which is precisely the $200 million of stock compensation granted. The difference between the two routes is $0.
Dilution and buybacks are the same cost wearing different clothes. One shows up in the share count and one shows up in the cash flow statement, and they net to the identical transfer of value.
Left unaddressed, that 2.0% compounds. Five years of it and the same shareholder owns 90.6% of what they started with, against a share count that has grown from 100 million to 110.4 million, without a single acquisition or capital raise.
What this does to the multiple
In the example, headline free cash flow per share of $10.00 against a $100 share price is 10.0 times. Charge the stock compensation and it is 12.5 times. Same company, same day, same filings. A quarter of the apparent cheapness was an artifact of what you agreed not to count.
This is not academic. In our initiation on Uber, stripping stock compensation out of trailing free cash flow moved the multiple from roughly 16 times to roughly 20 times before any other adjustment. That did not change the conclusion, but it changed what the conclusion had to clear.
The tell: gross buybacks against net buybacks
Here is the most useful practical habit in this note. When a company announces a buyback, do not read the headline number. Read it against stock compensation in the same period.
A company that repurchases a billion dollars of stock while issuing eight hundred million of stock compensation has not returned a billion dollars to you. It has returned two hundred million and spent the rest running to stand still. The press release says buyback. The cash flow statement says treadmill. The share count, which is the only scoreboard that cannot be spun, will tell you which one it was.
Check the diluted share count across several years. If it is flat while large buybacks are being announced, the buybacks are funding compensation. If it is falling meaningfully, capital is genuinely being returned.
Why the addback is contested
Companies that exclude stock compensation from adjusted earnings are not committing fraud, and it is worth stating their case properly before disagreeing with it. Four arguments are usually offered.
It is non-cash, so it does not affect liquidity. True, and irrelevant to whether it is a cost. Depreciation is non-cash too, and no serious analyst treats a business as if its assets last forever. The question is not whether cash moved this quarter. It is whether something of value left the owners.
The accounting figure is an estimate. This one has real merit. The expense is grant-date fair value, typically from an option-pricing model, recognized over the vesting period. It is not the amount ultimately transferred, which depends on where the share price goes. A grant that expires worthless still ran through the income statement. So the measure is genuinely imperfect.
It is volatile and obscures operating trends. Also partly fair. Compensation expense can swing on accounting mechanics that have nothing to do with how the business performed.
Everyone excludes it, so comparisons need it excluded. This is the weakest argument and also, in practice, the most decisive one. It is an appeal to convention, not to economics.
Where we come out
The first and fourth arguments do not survive contact with the share count. The second and third are real, but they argue that the measurement is imprecise, not that the cost is zero. An imprecise estimate of a real cost is more useful than a precise exclusion of it.
The simplest test is the one attributed to Warren Buffett and Charlie Munger, and it has never been answered well: if compensation is not an expense, what is it, and if it does not belong in earnings, where does it belong?
Reversing the question helps too. If paying people in stock genuinely costs nothing, no company should ever pay anyone in cash again. The reason they do not is that everyone involved understands the shares are worth something. That understanding just goes missing between the grant and the earnings release.
The adjustment, and the mistake to avoid
Handling this correctly is straightforward, and there is exactly one common error.
Do not charge it twice. You can subtract stock compensation from free cash flow and value the current share count, or you can leave cash flow alone and model the share count rising over your forecast period. Both are defensible. Doing both charges the same cost twice and produces a value that is too low by roughly the amount you are trying to be careful about.
We prefer subtracting it from cash flow and holding the share count flat. It puts the cost in the numerator where it is visible, it avoids compounding forecast error into the denominator, and it produces a per-share figure that can be compared across companies with different buyback policies.
Two further habits are worth building. Treat the diluted share count as the floor rather than the answer, because the treasury stock method excludes awards that are currently underwater and would dilute you if the stock recovered. And when a company presents an adjusted operating figure that does not add stock compensation back, notice it and give credit. It is rare, and it usually signals a management team that is comfortable being measured honestly.
What this note is not saying
Stock compensation is not a scandal and paying people in equity is often the right decision. It aligns employees with owners, it conserves cash for young companies that need it, and it can be a genuinely efficient way to attract people who would otherwise cost more in salary.
The argument here is narrower. It is that the cost of doing so should appear somewhere in the analysis, and that a valuation which excludes it is not a conservative valuation with a rounding error. It is a different valuation of a different company: one whose employees work for free.
Sources and further reading
The accounting treatment is set out in ASC 718 in the United States and IFRS 2 internationally, both of which require share-based payment to be recognized as an expense at grant-date fair value. Company filings disclose stock compensation on the cash flow statement as a reconciling item and in the equity footnote. The diluted share count and the treasury stock method are described in ASC 260. Berkshire Hathaway's annual letters have made the case against the addback repeatedly and more memorably than we have here.