Cash flow and share count change together
Suppose hypothetical equity value is 10,000 monetary units and 100 identical shares are outstanding. The company uses 1,000 to repurchase ten shares at 100. Assuming the shares are cancelled and other circumstances do not change, 90 shares and 9,000 of value remain. Each still represents 100. Reducing the count alone created no additional resource: cash left the company for the selling holders. The calculation does not say that a real market valuation must change exactly this way.
Now add constant hypothetical profit of 1,000 for the comparable period. Dividing by 100 gives 10 per share; dividing by 90 gives about 11.1111. The measure rises without total profit rising. Real reporting uses definitions of profit and weighted-average shares; this point-in-time calculation only demonstrates the denominator effect. Consequences also require lost cash income, financing costs and later operating changes.
A programme is not the same as execution
The announced maximum programme amount, cash actually spent and shares cancelled are three different quantities. Permission to repurchase does not mean the full amount has been bought. Even after execution, the treatment of repurchased shares may differ. Find actual transactions and capital movements, not only the headline amount. If the company also issues shares for other reasons, the final count change may be much smaller than the gross buyback volume.
Unlike primary fundraising, in which money goes to the issuer,[1](#source-cm-primary) a buyback uses company resources to settle with selling owners. This distinguishes a corporate operation from an ordinary secondary sale between investors. But the cash route alone does not establish whether the buyback price was fair. That requires assumptions about business value and alternative uses of funds. A price above or below earlier quotes is not enough to prove the economic effect.
What a changed per-share measure does not prove
A higher per-share figure can coexist with a smaller liquid reserve or more debt. Those effects are excluded from the example, so its conclusion is limited to arithmetic. Debt-funded repurchases require separate attention to interest, maturity and debt-performance risk. With own cash, distinguish accounting profit from available cash. [Profit versus cash](/guides/profit-vs-cash/) explains why they are not one amount.
To check a buyback report, build a bridge: shares before, acquired shares, cancelled shares, new issues and shares after. Show cash spent and changed obligations alongside it. Then name the exact per-share measure and period recalculated. [Market cap and multiples](/guides/market-cap-and-multiples/) separates unit value from total capital; [shares](/guides/shares/) separates ownership from a debt claim. This guide does not recommend buying companies with buybacks or predict quote reactions.
Testing the claim of value creation
The claim “EPS rose, therefore the company made more goods or received more customer cash” does not follow from one measure. In our example total profit is unchanged and the division changes because the count changes. A separate bridge is needed from revenue and costs to total profit and from total shares to the per-share measure.
Ask what happens to repurchased shares. If they are not cancelled and may later be used, documents are needed before treating the reduction as permanent. If borrowed money funded the buyback, a model without interest no longer describes the whole economics. If the purchase is not proportional among owners, individual cash flows also differ: the seller receives settlement while the holder who remains owns a changed share of the remaining capital. This does not predetermine a good or bad result. It only prevents attractive denominator growth from being treated as a costless new resource.
Buyback arithmetic does not prove later excess return
The claim that shares outperform the market after buybacks is an empirical hypothesis, not a consequence of a smaller EPS denominator. A test must predefine the universe, announcement and execution dates, result window, comparable risk and benchmark after costs. Do not select only successful programmes, omit discontinued ones or use information before its publication. Firms selected for buybacks differ in cash, valuation and maturity; later results may reflect those traits rather than a causal effect.
Separate announcement, actual purchase, cancellation, reuse and simultaneous employee issuance. Compare purchase price with a reasoned value range, lost liquidity, debt and alternative projects. A control group and risk adjustment do not automatically remove all selection differences. This guide does not collect such observations: it does not claim buyback outperformance, historical attribution or numerical premiums. It checks only the stated arithmetic and its limits, not a stock-selection recommendation.