A company buying its own shares is not automatically good news. The same decision can create value when a resilient company buys an undervalued stock, and destroy value when a company borrows heavily to buy expensive shares or gives up good investments.

To reach the right conclusion, look past the announced program to what the company actually bought. Gross cash spending, the net reduction in shares, the source of earnings-per-share growth, the average purchase price, the financing and the best alternative use of the cash are parts of the same picture.

The short answer

A buyback is a decision about where the company's money should go. Management uses cash to buy its own shares instead of funding a new product, factory, research, acquisition, debt reduction or liquidity reserve.

The decision is strongest when four conditions hold together: the company has already funded its best investments, the purchase price is sensible relative to business value, the share count actually falls after employee awards and other issuance, and the balance sheet remains strong enough to absorb a downturn.

The warning also appears in four places: management announces a large authorization but buys little, the share count barely falls despite the cash spent, EPS rises while total profit remains weak, or the buyback takes priority over investment and financial flexibility.

A buyback is therefore not a conclusion. It is the start of the review. “Is the company buying shares?” is not enough. Ask: “At what price, with what money, at the expense of which alternative, and with what net reduction in shares?”

Why does this question matter?

The most visible number in a buyback announcement is usually a multibillion-dollar program. But a board authorization does not mean the company will spend the full amount. The SEC staff report explains that companies must disclose actual quarterly purchases, but an ordinary open-market authorization does not require them to complete the program.

Actual purchases are not enough either. A company may buy 100 million shares from the market while issuing 40 million shares to employees. The cash outflow reflects the cost of 100 million shares, but continuing shareholders gain ownership only from the net reduction of 60 million.

EPS can mislead for the same reason. Earnings per share equals total net income divided by the average number of shares across which that income is spread. The result rises when the denominator falls. That is a real per-share effect, but it does not mean the company created a new customer, product or dollar of total profit.

Apple's 2025 fiscal year puts these distinctions in one filing. In its Form 10-K, the company reported repurchasing 401.672 million shares, issuing a net 58.146 million shares after employee-related transactions, and reducing the year-end share count by a net 343.526 million. In the same year, total net income rose 19.50%, while the average diluted share count used for EPS fell 2.62%. Exact diluted EPS rose 22.71%.

This example does not say whether Apple's stock was cheap or expensive. It only shows that the headline repurchase amount, the net share count and EPS are different measures.

How does the mechanism work?

1. Separate the authorization from actual purchases

When a board announces a $50 billion program, it gives management the ability to buy. It does not prove that $50 billion has been spent or will be spent by a particular date. Management may consider the shares expensive, redirect the cash, respond to a weaker economy or pause the program.

That flexibility is often useful. It can stop management from buying at a poor price or depleting liquidity during a downturn. But it also limits the information in the announcement. The evidence comes later in quarterly and annual filings: shares actually purchased, cash spent, average price and remaining authorization.

Accelerated share repurchase agreements can create a firmer short-term commitment. The general rule still holds: do not confuse an announcement with a transaction.

2. Separate EPS from total earnings

The basic equation is simple:

Earnings per share = net income / weighted-average shares

If net income is 100 and the share count is 100, EPS is 1. If the share count falls to 90 while net income remains unchanged, EPS rises to 1.11. The company did not increase its total profit; it divided the same profit among fewer shares.

That increase is not worthless. A continuing shareholder now owns a larger claim on company earnings. But revenue, operating profit, total net income and operating cash flow still need to be reviewed separately to judge whether the business improved.

For reported EPS, the correct denominator is the average share count during the year. Filings also disclose a broader diluted denominator that includes the effect of employee awards and other instruments that could become shares. Comparing only the year-end share count with the previous year-end does not reproduce reported annual EPS.

3. Separate gross purchases from the net share reduction

The number of shares bought in the market does not show the net gain to continuing shareholders. Employee awards, option exercises, acquisition consideration and other issuance can reverse part of the effect.

Two separate questions are required:

  • How many shares did the company buy, and how much cash did it spend?
  • After all new issuance, how much did the year-end and diluted average share counts actually fall?

Stock-based compensation expense is also not the same measure as the number of new shares. One is an accounting expense in dollars; the other changes the share count. Subtracting stock-based compensation expense from repurchase cash to calculate a “net buyback” mixes definitions.

The cleanest reading uses three filing areas together: the share movement in the shareholders' equity note, the diluted average denominator in the EPS note, and the award structure in the stock-based compensation note.

The distinction was even clearer at Microsoft. According to its 2025 Annual Report, the company repurchased 31 million shares for $13 billion in fiscal 2025 and issued 31 million shares in the same year. Shares outstanding were 7.434 billion at both the beginning and end of the year. That does not mean the repurchase did nothing; without it, the share count might have risen. It does mean the cash spent cannot be treated as the net ownership gain for continuing shareholders.

4. Do not separate the purchase price from business value

A company creates value by buying its shares only when the price is sensible. Consider a simplified business with intrinsic equity value of $1,000, 100 shares and intrinsic value of $10 per share. The company spends $100.

At a purchase price of $8, it retires 12.5 shares. The remaining business is worth $900 and has 87.5 shares. Intrinsic value per remaining share rises to $10.29.

At $10, intrinsic value per remaining share is unchanged. At $12, the company retires only 8.33 shares and intrinsic value per remaining share falls to about $9.82.

The arithmetic is clear, but the crucial input is uncertain: intrinsic value cannot be observed directly. Management and investors can both be wrong. “Management is buying, so the shares are cheap” is not a defensible conclusion. Compare the average price with earnings power, cash generation, debt and a reasonable range of values.

5. Review financing and opportunity cost together

It is difficult to label a buyback as “cash-funded” or “debt-funded” from one line. Corporate dollars are fungible. A company can issue new debt and repay old debt in the same year; it can repurchase stock while selling marketable securities.

At a minimum, review these items together:

  • Cash generated by operations.
  • Cash spent on factories, equipment and similar investment.
  • Total cash paid for dividends and buybacks.
  • Net proceeds from new debt after debt repayment.
  • Opening and closing cash and readily marketable securities.

Debt is not automatically bad. Low-cost, fixed-rate debt can replace excess equity and improve the financing mix. The warning appears when interest costs grow faster than profit and cash, maturities become concentrated, or the company may need expensive refinancing during a downturn.

The central test is opportunity cost. If a new project is expected to earn well above the cost of the money, giving it up to fund a buyback may be an expensive mistake. If the project is weak, returning cash can be more disciplined than investing merely to appear ambitious.

What does the research show?

Strong but narrow evidence around the EPS threshold

The 2016 peer-reviewed paper by Heitor Almeida, Vyacheslav Fos and Mathias Kronlund, The Real Effects of Share Repurchases, uses a narrow threshold to study whether buybacks reduce investment. Companies that would just miss analysts' EPS forecasts without a buyback purchase more shares than otherwise similar companies that would just beat the forecasts. The researchers use this sharp change to compare the groups.

In the four quarters after these EPS-motivated repurchases, capital expenditure was lower by about 10% of the sample mean, R&D by 3% and employment by 5%. Cash holdings also fell. This is strong evidence that managers close to the analyst threshold can reduce real spending to fund buybacks.

The boundary matters as much as the result. The design does not represent every company buying because its stock is undervalued, because it has surplus cash or because it is changing its capital structure. In the paper's own analysis, future profitability and shareholder value did not generally deteriorate across this narrow group. Outcomes were weaker among the subgroup that cut real spending at the same time. The study does not support the claim that all buybacks destroy investment.

Programs that look flexible can become sticky

A 2025 peer-reviewed paper by Almeida, Ruidi Huang and Yuhai Xuan, Are Share Repurchases Really Flexible?, finds that companies have maintained programs more regularly over the past four decades and that repurchases have become less flexible than commonly assumed. Around the 2008 financing shock, companies with programs that remained in place reduced investment, employment and R&D more than similar companies whose programs ended before the crisis.

That is an important warning, but it is not universal. A continuing program during a financing crisis is not the same as an ordinary program that management can pause. The evidence still gives investors a useful question: has the past buyback commitment started to behave like a fixed expense?

The counter-story managers tell

The survey by Alon Brav, John Graham, Campbell Harvey and Roni Michaely of 384 financial executives found that managers view repurchases as more flexible than dividends, tend to buy when they consider the stock undervalued, and often use cash left after investment and liquidity needs. They care about EPS as well.

This measures managers' account of their motives, not independent evidence of actual behaviour. The sample is also old. Still, efficient distribution by a mature, profitable company after it has funded investment is a credible competing explanation.

A 2026 NBER working paper by Itzhak Ben-David and Alex Chinco provides another mechanism. If a manager's objective is to maximize EPS, a buyback can look more attractive than a dividend because it reduces the share count. This is a model. It does not prove that every manager maximizes EPS or that maximizing EPS creates business value.

Apple 2025: From the headline amount to the investor's net result

The Apple example shows how to read the filing; it is not a valuation judgment. Every input comes from the company's 2025 Form 10-K. I calculated the percentages from the unrounded values reported in the filing.

How much did the share count actually fall?

  • Opening shares outstanding: 15.116786 billion.
  • Gross shares repurchased: 401.672 million.
  • Net shares issued after shares withheld for employee taxes: 58.146 million.
  • Closing shares outstanding: 14.773260 billion.
  • Net reduction: 343.526 million, or 2.27% of the opening count.
  • The net reduction equalled 85.52% of the gross shares bought. Net issuance offset the remaining 14.48%.

This distinction prevents two errors. It does not treat the gross 401.672 million shares as the direct ownership gain for continuing holders. It also does not subtract the $12.863 billion stock-based compensation expense from the $89.3 billion purchase amount to create an undefined “net buyback.”

How much of EPS growth came from the lower share count?

Apple's total net income rose from $93.736 billion in 2024 to $112.010 billion in 2025, an increase of 19.50%.

The diluted average share count used to calculate EPS fell from 15.408095 billion to 15.004697 billion, a decline of 2.62%.

Using unrounded filing values, diluted EPS rose from $6.0836 to $7.4650, an increase of 22.71%. The filing reports the rounded values as $6.08 and $7.46.

To isolate the contributions, I first divided 2025 net income by the 2024 share count. The result was $7.2696. I then used the lower 2025 share count, lifting the result to $7.4650.

In this calculation, the lower share count accounts for about $0.1954 of the total $1.3814 increase, or 14.15%. Most of the increase comes from higher net income. A different decomposition order would assign a slightly different share to the two effects, which is why the method must be stated.

How was the repurchase financed?

Apple generated $111.482 billion of operating cash in 2025 and paid $12.715 billion for property, plant and equipment. The simple post-investment cash proxy from these two lines is $98.767 billion.

The cash-flow statement reports $90.711 billion of repurchase cash, 91.84% of that proxy. Adding dividends brings total cash returned to shareholders to $106.132 billion, or 107.46% of the proxy. The difference must be read with the company's other cash and investment movements.

Apple also raised $4.481 billion of new term debt in 2025. That one line does not prove the buyback was debt-funded. The company repaid $10.932 billion of term debt and reduced commercial paper by a net $2.032 billion. Together, those three lines show net debt repayment of $8.483 billion.

The example establishes three points: repurchase cash differs from the net reduction in shares, EPS growth can be split between profit and denominator effects, and one new-debt line cannot settle the financing question.

It does not establish whether Apple's average 2025 purchase price was below intrinsic value, which other project could have been funded, or what the program will earn in the future. Those questions are necessary before calling the program good or bad.

The strongest counterargument

The strongest criticism of buybacks is that companies sacrifice investment and workers to meet short-term EPS targets. Narrow but strong research shows that this can happen. The same pattern also has another explanation: investment falls because growth opportunities weaken, while surplus cash goes to buybacks. Seeing both movements together does not prove one caused the other.

The SEC staff report states this problem directly. Distribution, financing and investment decisions occur together, while the growth outlook affects all three. Aggregate data therefore make it difficult to conclude that buybacks caused lower investment.

The report also explains why more investment inside one company is not automatically better for the economy. If cash that would have funded a weak project is returned to shareholders and redeployed to a higher-return company that needs capital, overall allocation can improve.

Compensation evidence also weakens the simple EPS story. The SEC reviewed the 50 largest repurchasers in 2018 and 2019. In 82% of them, either there was no EPS-linked compensation target or the board disclosed that it considered the buyback effect when setting or measuring the target. This does not eliminate every EPS motive. It does mean the evidence does not support explaining most buybacks as transactions designed only to lift executive compensation.

In the SEC's aggregate review, the debt-financed share of repurchases remained around 40%, while repurchasing companies tended to be more profitable, hold more cash and have lower leverage than non-repurchasers. An aggregate ratio cannot clear a specific company. It does show that “a buyback must have weakened the balance sheet” is not a defensible general conclusion.

This does not mean the criticism of buybacks is wrong. It means the criticism must be built from the company's filings, purchase price and forgone investment.

Seven questions for reading a buyback

  • Authorization or actual purchase? Check shares actually bought, cash spent, average price and remaining authority, not merely the announced ceiling.
  • Gross purchase or net reduction? Separate market purchases from shares issued to employees, for acquisitions and for other uses. Review both the point-in-time and diluted average share counts.
  • Total profit growth or only EPS growth? If revenue, operating profit, net income and operating cash are not rising, calculate how much of EPS growth came from the smaller denominator.
  • Was the purchase price sensible? Compare the average price with a range of business values built from different growth and margin assumptions, not a single market multiple.
  • How was it financed? Compare operating cash, investment spending, total distributions, net debt movement and liquidity together. Do not use one bond issue or one cash balance as the verdict.
  • What alternative was forgone? Compare the expected benefit of research, capacity, an acquisition, debt reduction or liquidity with the expected return from buying the shares.
  • Are incentives adjusted for the mechanical effect? If executive targets use EPS, check the proxy statement for whether the board removes or considers the buyback effect.

In a strong buyback, the answers point in the same direction: actual purchases occur, the net share count falls materially, business profit grows, the price is sensible, good investments are funded and the balance sheet is resilient.

In a weak buyback, the headline is large and the net effect is small. EPS rises through the denominator while operations slow, the company buys at a high price, or investment and liquidity are sacrificed.

What would change the conclusion?

Evidence that would strengthen the case for a company:

  • A sustained and meaningful decline in the net share count over several years.
  • Growth in total earnings and operating cash before the buyback effect.
  • An average realized purchase price below a defensible valuation range.
  • Continued research and capacity investment alongside preserved debt capacity.
  • Clear board treatment of the EPS and compensation effects.

Evidence that would weaken the case:

  • Repeatedly larger authorizations but limited actual purchases or net share reduction.
  • Purchases concentrated at high valuations and stopped after the price falls.
  • Rising EPS alongside weaker total earnings, operating cash or investment.
  • Higher interest and maturity risk, or the need for expensive financing in a downturn.
  • Credible evidence that the program displaced positive-value projects.

What should readers monitor next?

  • Actual shares bought, cash spent and average price in quarterly and annual filings.
  • Year-end shares outstanding and the diluted average denominator used for EPS.
  • Stock-based compensation expense, net shares issued and unrecognized award expense.
  • Revenue, operating profit, net income and operating cash. Read EPS after these total measures.
  • Capital spending, R&D, acquisitions and any expected project returns disclosed by management.
  • New debt, debt repayment, interest expense, maturity schedule, cash and marketable securities.
  • Whether the proxy statement uses EPS in executive targets and how the board treats buyback effects.

Tracking these measures over several years is safer than relying on one period. An acquisition, large tax payment or accelerated buyback can temporarily distort an annual number. Repeating the same calculation with stable definitions helps prevent a selected year from becoming the entire story.

Conclusion

A stock buyback is neither an automatic vote of confidence nor an automatic growth problem. Value comes from price and opportunity cost, not from the name of the program.

If the company has funded good investments, used surplus cash to buy shares at a sensible price, reduced the net share count and kept the balance sheet strong, the buyback can be disciplined.

If the company buys expensive shares, merely offsets employee issuance, makes EPS look stronger than the business or gives up investment and financial flexibility, the same transaction becomes a warning.

The most useful number is therefore not the multibillion-dollar authorization. The net share reduction, the source of EPS growth, average price, financing and forgone alternative provide the real answer together.

Methodology and limitations

This Guide uses a data cutoff of 26 July 2026 and draws on the SEC's 2020 staff report, peer-reviewed academic research, one 2026 NBER working paper, Apple's 2025 Form 10-K and Microsoft's 2025 Annual Report. I use Apple as the main calculation example; the Microsoft figures are direct filing data used to show why gross purchases and the net share-count change must be read separately. Most sources focus on US-listed non-financial companies. Their conclusions do not automatically extend to banks, regulated companies or other jurisdictions.

I calculated the Apple percentages using the filing's unrounded net income and share counts. In the EPS calculation, I first applied 2025 net income to the 2024 denominator, then moved to the 2025 denominator. A different order changes the contribution assigned to each effect. The post-investment cash proxy subtracts property, plant and equipment payments from operating cash. It excludes acquisitions and other definitions of investment and is not a separately reported company metric.

Point-in-time shares measure the net ownership change, while diluted average shares are the denominator for annual reported EPS. The two were not used interchangeably. Stock-based compensation expense was also kept separate from the number of new shares.

The purchase-price example is a simple sensitivity that assumes no taxes, transaction costs or other policy changes. It does not estimate the intrinsic value of Apple or any other company.

The academic studies use different periods, samples and methods. The strong causal result around the EPS threshold was not generalized to all buybacks. The SEC aggregate evidence was not treated as company-level causality. Project-level expected returns and the exact counterfactual use of buyback cash are often not observable in public data. That gap prevents a definitive company-specific verdict.

Sources and further reading

Not investment advice; for research and educational purposes.