Every year a business generates operating cash and faces the same closed menu. There are five things you can do with a rupee of capital, and only five:
- Reinvest organically: capex, R&D, working capital, new capacity, new markets.
- Acquire: buy someone else's assets or business.
- Pay dividends: hand cash to all owners pro rata.
- Buy back shares: hand cash to the owners who choose to leave.
- Pay down debt (or build cash, the same lever with the sign flipped).
Issuing equity or debt is the menu run backward. Raising capital is a bet that you hold a use worth more than its cost.
Treat the master rule as a comparison, run zero-based every period. There's no fixed pecking order. Mauboussin frames it like this: rank every rupee's destinations by value created per rupee (after tax and adjusted for risk), then fund from the top until the marginal use no longer beats the marginal source. In practice the ranking has a natural shape. Organic projects inside the moat usually top it. At a fair stock price a buyback returns the cost of equity, which makes it the benchmark everything else has to beat. But every "always buy back" or "never pay dividends" slogan dies on contact with prices. Each use gets its decision framework and a worked number below.
Advertisement
4.7.1 Organic reinvestment: the incremental-ROIC hurdle
Fund every project whose expected ROIIC clears WACC with a margin of safety, and starve everything else, however strategic it sounds. Three habits turn this from a pious wish into practice. First, go zero-based, because last year's capex isn't this year's entitlement. Second, fund strategies, not projects, because sponsors game single-project DCFs. Call it the "everything IRRs at 18%" disease, and audit it by comparing the division's delivered ROIIC with its promises. Third, keep zero tolerance for bad growth. Killing a sub-WACC expansion is value creation that never appears on any statement.
Take a pipes-and-adhesives maker (synthetic). It can build a ₹600 crore plant that reaches steady-state incremental NOPAT of ₹90 crore, an ROIIC of 15%, against a 12% WACC. That stream is worth 90/0.12 = ₹750 crore, so the NPV is +₹150 crore, or ₹1.25 of value per rupee invested. It's the perpetuity shortcut, value per rupee = ROIIC/WACC. Cut the steady-state NOPAT to ₹60 crore (ROIIC 10%) and the same plant proposal is worth 60/0.12 = 500, an NPV of −₹100 crore. You've got a ribbon-cutting that incinerates a sixth of its budget. Notice what the shortcut buys you in meetings. Any capex plan plus a steady-state profit claim converts to value per rupee in ten seconds. Ramp-ups, working-capital drag, and execution risk all push realised ROIIC below the deck's number. Hence the margin of safety, and hence the audit of delivered ROIIC in the dossier.
Why does organic usually rank first when it's available? It compounds inside your own moat, where you know the business and information risk is lowest. And it often widens the moat (capacity, distribution, R&D versus rivals). High-ROIIC organic opportunities are the compounding machine the matrix's top-right cells describe. DMart's redeployment of ~90% of NOPAT at 17 to 18% ROIIC, which you met in the returns work, is one. Its limit is runway. Opportunity isn't infinite, and forcing ₹100 of capex into ₹40 of opportunity produces Trap 1 from the ROIIC section.
4.7.2 M&A: synergies versus the premium, against brutal base rates
An acquisition creates value for the acquirer's shareholders only if:
PV(synergies the deal makes possible) > Premium paid over standalone valueWhatever the target is worth on its own belongs to the target's shareholders, and in a competitive deal market you pay at least all of it. So the whole game is the gap between two uncertain numbers. One is the synergies: cost savings, revenue lifts, tax/capital efficiencies, all net of integration costs. Against them stands the premium, typically 20 to 40% over the undisturbed price in public deals. Everything else must eventually cash out through that inequality, or it's perfume: "strategic fit," "platform for growth," "TAM expansion."
Start with the base rates, the outside view you carry into every deal announcement. Across five decades of academic and consultant studies, roughly half to two-thirds of acquisitions fail to create value for the acquirer. Most or all of the aggregate gain goes to the target's shareholders through the premium. So the deal in total often creates value, and that value simply gets paid away at signing. Two more findings repeat across studies. Cost synergies are mostly deliverable but routinely offset by underestimated integration costs. Revenue synergies come in at well under half of deck projections, so haircut them ~50% before you even read the rationale. The profile that beats the base rate is the small-to-mid, cash-funded, bolt-on acquisition in a business the acquirer already runs, bought off-cycle. The profile that fails is the large, competitive-auction, stock-celebrated "transformational" deal at the top of the cycle.
These numbers are ranges across contested literatures: McKinsey's deal research, Mauboussin's 2014/2016 capital-allocation reports, Bruner's Deals from Hell. So train yourself to say "the base rate is against this" instead of quoting one falsely precise figure.
Worked example: an acquisition at a 30% premium
Try a worked deal. An acquirer with an equity value of ₹15,000 crore buys Target, whose standalone DCF value = market cap = ₹4,000 crore, for a 30% premium. So the price is ₹5,200 crore and the premium ₹1,200 crore. Management's deck promises ₹80 crore/yr of after-tax cost synergies from year 2 and ₹60 crore/yr of revenue synergies, against ₹150 crore PV of integration one-offs.
Underwrite it yourself at a WACC of 12%. Cost synergies are worth 80/0.12 = 667, or ≈ 595 once you discount them a year. Revenue synergies take the base-rate haircut of 50%, so 30/0.12 = 250, ≈ 223 discounted. Take off the 150 of one-offs and PV(synergies) ≈ ₹668 crore against a ₹1,200 crore premium: NPV ≈ −₹532 crore, −3.5% of the acquirer's equity. And that's before any execution stumble.
Then run it backward. Break-even requires 1,200 × 0.12 = ₹144 crore/yr of after-tax synergies, forever, which is ≈ ₹192 crore pre-tax at a 25% rate. If Target's entire cost base is ₹2,700 crore, management is implicitly promising a permanent 7% cost takeout. Has anyone in this industry ever delivered that? Ask that one question at announcement and you've done most of the M&A analysis.
Case study: Tata Steel's bid for Corus
India's museum piece is Tata Steel's bid for Corus in 2007. A nine-round auction against CSN pushed the bid from 455p to 608p per share, reaching ~$13bn enterprise value, roughly a 50% premium to the undisturbed price, at peak-cycle steel EBITDA. Treat those figures as an illustrative recollection of scale, and pull the contemporaneous filings and coverage. Don't learn this one with hindsight. Learn the mechanism: the premium was certain cash paid at a cyclical top, and the synergies were back-loaded hopes in a commodity business with no pricing power. The confession arrived over the following decade in serial impairments and an eventual exit from much of the UK footprint. Every element was visible at announcement through the inequality above: auction dynamics, peak multiple on peak earnings, cash premium vs soft synergies, cyclicality.
India has the positive base-rate profile too. Serial small bolt-ons rarely make headlines and rarely make impairments. Any IT-services firm's tuck-ins, measured in hundreds of crores, are a page from that playbook. Goodwill impairment is M&A's confession booth, a thread that runs back to your accounting work. A decade's impairment history is the cheapest audit of a management's dealmaking you'll ever run.
4.7.3 Dividends: the honest residual, and what it signals
Dividends are the default use: what remains after every positive-spread reinvestment and any below-value buyback has been funded. In formula form, sustainable payout ≈ 1 − g/ROIC, which is the value-driver formula's own reinvestment logic turned inside out. So a dividend policy is a statement about runway. A 90% payout says "we can only deploy a tenth of what we earn at our returns". At a high ROIC that's an honourable admission, and no kind of failure. Hindustan Unilever's ~90%+ payout on an enormous ROIC is the textbook case. It's asset-light and runs negative working capital, as you saw in the returns work. Growth of 8 to 10% at an ROIC north of 50% needs an RR of well under 20%. Retaining more would only pile up cash at money-market returns and drag the blended return down.
The framework runs two tests. First, is the payout consistent with the claimed growth? A company promising 15% growth at 20% ROIC needs RR = 75%, so a simultaneous 80% payout means the growth claim and the cheque can't both be true. Someone is borrowing to bridge the lie. Second, is the dividend funded by FCF or by the balance sheet? A debt-funded dividend is a leveraged recap wearing a dividend's clothes: conservation of value, plus rising distress risk.
What do dividends signal, and what changes in India?
Signalling. Lintner's findings are sixty years old and still the best empirical description. Managers set a target payout, move toward it in cautious partial steps, and cut only under duress. So the information in dividends is asymmetric. Initiations and steady growth signal confidence in persistent cash flow, because boards know a raise is quasi-permanent. A cut screams that internal forecasts broke, and markets typically treat cuts as the loudest bearish disclosure a board can make. There's a corollary for you as an analyst. A dividend held flat while FCF collapses is a board burning the balance sheet to postpone the signal, whatever the "shareholder-friendly" label says. Check dividends ÷ FCF, not dividends ÷ EPS.
India context: the tax and the promoter (as of mid-2026, check rates). Since April 2020, when DDT was abolished, dividends have been taxed in shareholders' hands at slab rates, up to the 30%+ bracket for promoters and HNIs. That made buybacks the tax-efficient return route until the October 2024 change taxed buyback proceeds as deemed dividends too. It roughly levelled the field and visibly tilted payout mixes back toward dividends.
Then there's the promoter. In promoter India the dividend cheque's largest recipient is the promoter, and that's mostly benign. TCS's dividends fund Tata Sons and, upstream, the Trusts. But audit the cases where group cash needs set the policy, meaning a stressed holdco extracting maximal dividends from an operating company. Leveraged-promoter groups have run payouts above 100% of FCF, funded by debt, in plain sight of minority holders. Vedanta-group episodes are the standing case study, so read the parent's debt schedule next to the subsidiary's payout. Treat a dividend sized to the owner's margin call as governance material, arriving ahead of the incentives section. It has stopped being a capital-allocation decision. PSU India is its own case. Government fiscal needs drive payout mandates, so the dividend tells you about the majority owner's budget and nothing about the company's runway.
4.7.4 Buybacks: value-creating only below intrinsic value, worked both directions
The framework is one comparison, and it decides everything:
Buy back stock ⟺ Price paid < intrinsic value per share (conservatively estimated)
AND no higher-value use for the cash existsThat's Buffett's two-condition rule (from the 1984 letter and the 2011 to 12 letters) and Mauboussin's "golden rule" in one line. Conservation of value explains why. A buyback creates nothing in aggregate. It transfers value between sellers and stayers, and the direction of the transfer is set entirely by price against value. A management that buys back "because we generated cash" or "to offset dilution" or "to hit the EPS bridge" is spending money, and allocation has nothing to do with it. Work the maths carefully in both directions once and you'll own it for good.
Setup (synthetic). Picture a company with 100 crore shares and an intrinsic equity value of ₹15,000 crore, or ₹150/share. Of that, ₹1,200 crore is excess cash earning 3% after tax, which is ₹36 crore of the company's ₹900 crore earnings (EPS ₹9.00). Management deploys the ₹1,200 crore on a tender buyback.
Direction 1: the stock trades at ₹100 (33% below value). The tender retires 1,200/100 = 12 crore shares. Continuing owners now hold a business worth 15,000 − 1,200 = ₹13,800 crore across 88 crore shares, which is ₹156.82/share, +4.5%. That ₹6.82 gain per continuing share is the sellers' loss, rupee for rupee. Those 12 crore shares sold ₹150 of value for ₹100, so ₹600 crore was transferred = 88 crore × ₹6.82 ✓. Nothing was created. All the stayers did was win the transfer. This is Henry Singleton's entire Teledyne playbook, which you'll meet again with the masters, and it's what Buffett called "the surest way to use funds intelligently": buying your own known business below its value is the rare acquisition with no premium, no integration risk, and perfect due diligence.
Direction 2: the stock trades at ₹200 (33% above value). The tender retires 1,200/200 = 6 crore shares. Continuing owners get 13,800/94 = ₹146.81/share, −2.1%. It paid ₹200 for ₹150 of value 6 crore times over, so the ₹300 crore gift went to the exiting shareholders. Same cash, same announcement language, opposite result.
Now run the EPS illusion on the same numbers. Once the company gives up the 3% cash yield, earnings are 900 − 36 = ₹864 crore. In direction 1, EPS = 864/88 = ₹9.82 (+9.1%). In direction 2, EPS = 864/94 = ₹9.19 (+2.1%). EPS rose in both directions, because the earnings yield (4.5 to 9%) beat the 3% cash yield in both, while per-share value rose in one and fell in the other. EPS accretion can't tell a brilliant buyback from vandalism. It's the wrong instrument, the way a thermometer can't weigh you. So when the proxies you'll read in the incentives section pay managers on EPS, which direction do you suppose they check?
How do you judge a real buyback in practice?
Three practical tests grade a real buyback, and they feed your dossier. First, did the company buy more when the stock was cheap and stop when it was expensive, or the reverse? Most companies do the reverse. US aggregate buybacks peaked in 2007 and at Q4-2021-type tops and collapsed in 2009, procyclical and systematically wrong-way. Second, look for a stated valuation rule, such as Berkshire's price-to-book ceiling, which later became "below intrinsic value, conservatively determined". Third, is the share count actually falling, or are buybacks merely mopping up ESOP issuance? Trust the row that nets repurchases against issuance.
India mechanics (as of mid-2026). Buybacks run under SEBI's cap of 25% of capital and free reserves, and tender offers have been the dominant route since the open-market route was phased out by 2025. Tender buybacks at a premium are functionally a mixed dividend-plus-buyback. TCS's five tenders, totalling ~₹83,000 crore across FY17 to FY24 and typically priced 10 to 20% over market, are the example. And the October 2024 shift to slab-rate taxation of proceeds removed the tax edge that made tenders a promoter-friendly channel.
4.7.5 Debt paydown: the certain return, plus insurance
Repaying debt "earns" the after-tax cost of the debt with certainty, a guaranteed 6 to 7% in our illustrative worlds. Against expected, risky returns of 12%+ elsewhere, that looks pedestrian. So growth-drunk boards chronically underrate paydown, and for two kinds of company it's chronically correct.
First, the levered cyclical. When net debt/EBITDA at mid-cycle earnings runs past ~2 to 3×, the next downturn turns "expected returns elsewhere" into forced asset sales at the bottom. Paydown buys survival, and survival has option value that no spread compensates for. You already know this as the solvency lens, and Phase 2's IL&FS and DHFL files show what the alternative looks like. Second, the balance sheet as a war chest. Going into a downturn unlevered is the capital cycle trade: the ability to buy assets, share, and distressed rivals while everyone else is deleveraging. So don't ask "is 7% > 12%?" Ask "what is the probability-weighted cost of financial distress, and what is the option value of capacity?" Neither shows up in a point-estimate IRR comparison.
Put numbers on the cyclical. A steelmaker carries ₹24,000 crore of net debt at 9.5% pre-tax, ≈7.1% after tax. On mid-cycle EBITDA of ₹8,000 crore that's 3.0×. On trough EBITDA of ₹4,000 crore it's 6.0×, which is covenant country. Use ₹4,000 crore of a boom year's FCF to deleverage and you get a certain return of ≈ ₹285 crore/yr after tax, plus trough leverage falls to 5.0×. Bigger still is the regime change in survival odds and borrowing spread. On the synthetic-rating table from your cost-of-capital work, coverage migrating one rating band can move the spread 100 to 200bp on the entire stack.
India's decade supplies the exhibits. Reliance sprinted to net-debt-zero in 2020 to 21 through a rights issue and stake sales after the telecom build. Tata Motors and Tata Steel deleveraged after 2020. In each case the market re-rated the equity as distress probability left the price. The reverse case, leveraging up at the cycle top to fund capacity or buybacks, is the signature of allocation malpractice. Watch for it in IBM's decade and in every commodity boom.
Worked example 4 puts the whole menu together, running one CFO's full year through all five uses in sequence. You'll repeat that drill in the dossier and for the rest of your career: rank by value per rupee, fund the moat first, refuse bad growth, return the rest at the right price, and keep the balance sheet matched to the cyclicality.