Buybacks vs Stock Compensation: How Much of That $90 Billion Actually Shrank the Share Count?
💵 Buybacks vs Stock Compensation: How Much of That $90 Billion Actually Shrank the Share Count?
Welcome back, everyone! 📊 I’m your Quant Analyst, filtering out market noise using data, statistical modeling, and systematic insights. 👩💻✨
“The company returned $90 billion to shareholders” is one of the most repeated sentences in equity research, and it is measured in the wrong units. A buyback is only worth what it does to your share of the company — and every dollar of stock handed to employees pushes in the opposite direction. So I pulled the annual filings for six megacaps and measured both. To cut straight to the chase: the amount of buyback needed to retire one percent of the share count varies by more than fourteen times across companies people file in the same bucket.
▲ The headline number is always the cash going out. The number that matters is what came back the other way
📌 First, the actual numbers
All figures come from the companies’ own 10-K filings via SEC XBRL — the PaymentsForRepurchaseOfCommonStock, ShareBasedCompensation and WeightedAverageNumberOfDilutedSharesOutstanding tags. Latest completed fiscal year: Apple repurchased $90.7bn against $12.9bn of stock compensation. Alphabet $45.7bn vs $25.0bn. Nvidia $40.1bn vs $6.4bn. Meta $26.2bn vs $20.4bn. Microsoft $22.3bn vs $12.4bn. Amazon repurchased nothing while granting $19.5bn.
Source: SEC EDGAR company facts (XBRL). Ratios, share-count changes and the cost-per-point figures are my own calculations.
1. What fraction of the buyback is just mopping up pay?
Start with the simplest comparison: for every dollar spent buying stock back, how many dollars of stock went out the door as compensation in the same year?
Stock Comp as a Share of Buyback
*Latest completed fiscal year for each company, from their 10-K filings. Fiscal years differ — Microsoft ends in June, Nvidia in January, Apple in September — so these are not identical calendar periods.
The spread is immediate. Apple and Nvidia are returning capital with only 14% and 16% of it flowing back out as pay. At Alphabet and Microsoft it is over half. At Meta, 78 cents of every buyback dollar is matched by stock issued to employees. And Amazon does not buy back at all while granting $19.5 billion — a deliberate policy, but one where the dilution has no offset whatsoever.
2. The test that actually settles it: did the share count fall?
Ratios are suggestive; the share count is the verdict. I took each company’s diluted share count from its 10-K, measured the change over a three-year window, and divided the buyback spending across that window by the percentage points actually retired.
| Company | Window | Diluted shares | Buyback spent | Cost per 1pp retired |
|---|---|---|---|---|
| Meta | FY22→FY25 | −4.74% | $76.1bn | $16bn |
| Alphabet | FY24→FY25 | −1.74% | $45.7bn | $26bn |
| Apple | FY22→FY25 | −8.09% | $263.2bn | $33bn |
| Nvidia | FY24→FY26 | −1.71% | $73.8bn | $43bn |
| Microsoft | FY23→FY26 | −0.25% | $57.9bn | $228bn |
| Amazon | FY22→FY25 | +6.26% | $0 | — (count grew) |
What One Percentage Point of Share Reduction Costs
*Cumulative buyback across the window divided by percentage points of diluted share count retired. Amazon is absent because its share count rose. Nvidia’s base year is adjusted for its 10-for-1 split.
Microsoft spent $57.9 billion over three years and its diluted share count fell by one quarter of one percent. Meta got roughly fourteen times more share reduction per dollar. Both are excellent businesses and both “returned capital to shareholders” in the press-release sense. Only one of them meaningfully increased what each remaining share owns.
Apple is the instructive middle case: the largest absolute programme on the list, $263 billion, and it did move the needle — more than eight percent of the company retired in three years. Scale plus a low compensation ratio is what buys that.
3. Now let me argue against my own table
The most important caveat, stated plainly
A buyback that only holds the share count flat is not a wasted buyback. Without Microsoft’s $57.9 billion, the count would have risen materially, and every existing holder would own less of the company than they do. The spending bought something real — it just bought the absence of dilution rather than the presence of concentration. That is a completely different product from the one the phrase “returned $22 billion to shareholders” puts in your head, and the whole point of this post is that the filings let you tell which one you are getting.
Three more limits worth knowing:
- Diluted share count moves for reasons other than buybacks and pay. Convertible instruments, and the fact that in-the-money option dilution grows as the share price rises, both push it around. A rising stock mechanically increases dilution — which means the companies whose shares performed best face the strongest headwind here, and that partly flatters the laggards.
- Stock compensation is already an expense in GAAP earnings. I am not claiming it is hidden. The point is narrower: the cash leaving via buybacks and the shares arriving via grants are reported in different statements, and almost nobody nets them.
- Amazon’s growing share count is a policy, not a failure. A company reinvesting every dollar into the business and paying partly in equity is making a legitimate choice. It just means “shareholder return” is not part of the case for owning it, and the case has to rest entirely on what the reinvestment earns.
4. How to check this yourself in about two minutes
You do not need a data subscription. Every US filer publishes structured financial data, and one URL gets you the series:
data.sec.gov/api/xbrl/companyconcept/CIK<10-digit CIK>/us-gaap/PaymentsForRepurchaseOfCommonStock.json
Swap the tag for ShareBasedCompensation and WeightedAverageNumberOfDilutedSharesOutstanding and you have all three series. Filter to form: "10-K" so you get audited annual figures rather than overlapping quarters. One warning from doing this: cash-flow tags are reported year to date, so a single quarter has to be differenced out of the cumulative periods — which is exactly the trap that made my first pass show Alphabet and Meta buying back nothing.
Quick FAQ
Q. So is a high stock-comp ratio a sell signal?
No, and I would resist that reading. Meta has the highest ratio on this list and also the best cost-per-point, because its buyback is large enough to overwhelm the grants. The ratio tells you how much of the gross number is doing nothing for you; the share count tells you the result. Always go to the second one.
Q. Doesn’t a buyback support the price anyway?
Only while it is running, and that is a very different claim from increasing intrinsic value per share. The durable effect of a buyback is arithmetic: fewer shares dividing the same earnings. If the count does not fall, that arithmetic never happens, whatever the price did in the meantime.
Q. What number would you actually put in a model?
Net buyback yield — (buyback minus stock compensation) divided by market cap — rather than gross buyback yield. It is one subtraction, it uses figures both published in the same 10-K, and on this sample it changes the ranking of four of the six companies.
💡 Quant Strategy & Takeaways
Gross buyback dollars are the wrong unit. Across six megacaps the cost of retiring one percent of the share count ranged from $16bn at Meta to $228bn at Microsoft, and Amazon’s count rose 6% while it granted $19.5bn of stock. Subtract stock compensation from the buyback before you call it a shareholder return — both numbers are in the same filing.
When market volatility spikes, remove emotion and focus strictly on the numbers! 🤖
Had you ever netted stock compensation against a buyback before reading this — or does the headline number usually go straight into the model? Let me know in the comments! 📈✨
Disclaimer: This article analyses public SEC filings for educational purposes and is not financial or investment advice. All ratios and share-count calculations are my own and use fiscal years that differ between companies. Diluted share counts are affected by factors beyond buybacks and compensation. Always do your own research or consult a licensed financial advisor before investing.
Comments
Post a Comment