Capital Allocation — Dividends, Buybacks & Capex

A chief executive of a mature company will make a handful of genuinely large operating decisions in a decade. They will make a capital allocation decision every single year, and each one is cumulative.

The arithmetic is unforgiving in both directions. A company that reinvests its cash at a return meaningfully above what that capital costs grows the value of each share every year it does so, and the effect multiplies. A company that reinvests at a return below its cost of capital is destroying value while reporting growing profit — because profit rises whenever more capital is deployed at any positive return, even one that does not justify the capital. Reported growth and value creation are not the same thing, and capital allocation is exactly where they part company.

This is also why two businesses with near-identical operations can deliver very different outcomes to their owners over ten years. The operations set how much cash arrives. Capital allocation decides what that cash becomes.

There are only five destinations, plus the option of leaving the cash on the balance sheet. Every annual report you will ever read is describing some mix of these five, whether or not it uses the language.

The choice is not moral. There is no permanently superior destination — a dividend is not inherently more shareholder-friendly than capex, and a buyback is not inherently better than repaying debt. What makes a choice right is the alternative it was made against, at that company, at that point in its life.

The governing test is a comparison, not a threshold. Cash should be reinvested for as long as the return on the incremental capital exceeds the cost of that capital, and returned to shareholders once it does not. Everything else in this topic is detail hung on that one sentence.

The practical difficulty is that a company reports its return on total capital, not on the new capital it deployed last year, and the two can differ enormously. A business earning a high return on a legacy asset base can be adding new capacity at a far poorer return while the blended figure barely moves. The incremental measure looks only at what changed, which is why it is the more honest number over a multi-year window.

Cost of capital is the other half of the comparison. For most Indian businesses it is a blend of what lenders charge and what equity holders require for the risk they are taking, and it is an estimate rather than a fact. Rather than defending a precise figure, work with a sensible range and see whether the conclusion survives across it. A project earning a pre-tax return in the high twenties clears any reasonable estimate; one earning nine per cent clears almost none. The return ratios topic in this module builds ROCE properly, and the DCF topic deals with discount rates in full.

Two different questions get confused constantly. The payout ratio asks what share of profit the company chose to distribute. The yield asks what that distribution is worth relative to the price you would pay for the share today. The first is about the company; the second is about the company and the market price together.

Take Vindhya Ceramics Ltd (figures illustrative throughout; not a real company). It earns ₹200 crore of profit after tax on 10 crore shares, so earnings per share is ₹20. It pays ₹60 crore in total dividend, which is ₹6 a share and a payout ratio of 30 per cent. At a market price of ₹600 the dividend yield is 1 per cent.

The retention ratio is simply the other side of the payout: 70 per cent of profit stayed in the business. Whether that retention was a good idea is answered by the return the company earns on it, which returns us to the incremental capital test. A high payout at a business with genuinely high-return reinvestment opportunities is arguably a failure of nerve; a low payout at one with nowhere good to deploy the money is worse.

Yield is a fraction, and a fraction rises when the denominator falls. That is the whole trap.

Stay with Vindhya Ceramics. At ₹600 a share and a ₹6 dividend, the yield is 1 per cent. Suppose the market price halves to ₹300 on concerns about the company's end market. The dividend has not changed and the board has done nothing, yet the yield now reads 2 per cent. A screen sorted by dividend yield has just moved the company up the list on the strength of its share price falling.

That does not make every high yield a trap. Some businesses genuinely distribute a large share of dependable cash flow — mature utilities, some public sector undertakings, businesses with limited reinvestment needs. The distinction is whether the dividend is being funded by free cash flow that recurs, or by something else.

Three checks separate them. Is the payout ratio still comfortable, or has it climbed toward or past 100 per cent because profit fell while the dividend was held? Was the dividend covered by free cash flow in each of the last few years, or paid in years when free cash flow was negative? And was any part of it funded by fresh borrowing, which the financing section of the cash flow statement will show plainly? A dividend that fails those checks is not income. It is the company returning capital it cannot really spare, and it is usually cut eventually.

A buyback is the company purchasing its own shares and extinguishing them. The shares cease to exist, so the number outstanding falls and every remaining shareholder owns a marginally larger fraction of the same business. It is a return of cash, like a dividend, but delivered through ownership rather than a payment.

Indian buybacks run under the SEBI buy-back regulations and the Companies Act, and two routes exist. In a tender offer the company invites all shareholders on a record date to offer shares at a fixed price, usually at a premium to the market, and accepts them proportionately, with a portion reserved for small shareholders. In an open market buyback the company buys through the exchange over a period at prevailing prices, with no certainty about how much it will actually complete. SEBI has been progressively tightening the open market route in favour of the tender route, so check the regulation as it currently stands before assuming a company can choose freely.

The Companies Act sets the outer limits: a buyback is capped as a proportion of the company's own paid-up capital and free reserves in a financial year, larger buybacks require a shareholder special resolution rather than only a board resolution, and there is a ceiling on the post-buyback ratio of debt to owned funds so the exercise cannot be used to hollow out the balance sheet. Treat the specific percentages as things to verify against the current text of the Act rather than remember, because they have been amended before.

Mechanically, a buyback shrinks the denominator. Profit is unchanged on the day, and it is now divided across fewer shares, so earnings per share rises.

Vindhya Ceramics earns ₹200 crore on 10 crore shares, so EPS is ₹20. It buys back 50 lakh shares at ₹600 each, spending ₹300 crore. Shares outstanding fall to 9.5 crore, and EPS on the same ₹200 crore of profit becomes about ₹21.05 — an uplift of roughly 5 per cent, produced without selling a single extra tile.

That arithmetic is why buybacks are easy to present as shareholder-friendly, and why the presentation deserves scepticism. Nothing about the business improved. Two real costs sit behind the higher EPS: the ₹300 crore is gone, along with whatever it was earning while it sat as cash or could have earned if deployed, and the balance sheet is that much thinner. If the buyback was funded by borrowing rather than surplus cash, EPS rises further still while the company becomes more leveraged, and the leverage topic in this module explains why that is not a free improvement.

There is one more use of buybacks worth naming. Where a company issues large volumes of employee stock options, a buyback of a similar size does not reduce share count at all — it merely offsets dilution. The reported share count looks stable, cash has left the company, and shareholders have effectively paid the employee compensation in cash without it ever appearing as an expense they noticed. Compare the buyback size against the options issued in the same years before crediting management with a reduction in share count.

A buyback is the company buying a stake in itself. Everything true of any purchase is true here: the price decides whether it was a good idea.

When the shares are worth more than the market is charging, buying them transfers value to the shareholders who stay, because the company retires a claim on the business for less than that claim is worth. When the shares are worth less than the market is charging, the transfer runs the other way — the shareholders who tender get a good price, funded by the ones who remain.

Put numbers on it with Vindhya Ceramics. It spent ₹300 crore retiring 0.5 crore shares at ₹600. If a considered estimate of what those shares are worth is ₹350, the company paid ₹300 crore for something worth ₹175 crore, and roughly ₹125 crore of value has moved from continuing shareholders to exiting ones. Earnings per share still went up by 5 per cent. Both statements are true at once, which is exactly why the EPS uplift on its own proves nothing.

The pattern to be alert to is timing. Buybacks cluster when companies have the most cash and the most confidence, which tends to be when their shares are most expensive — and they are conspicuously absent when prices are depressed and the cash would go furthest. Reading a buyback announcement means asking what the company appears to be worth, not only how large the announced sum is.

The two routes are not taxed the same way, and that difference has historically influenced which one companies choose. What matters for an investor reading a filing is the principle, not a rate.

The principle is that the tax burden can sit either with the company distributing the cash or with the shareholder receiving it, and Indian law has moved that burden more than once in recent years — for dividends, and separately for buyback proceeds. Because the point of incidence and the applicable rates have both changed, and because the answer also depends on whether the shareholder is an individual, a company, or a non-resident, no figure quoted in an article stays correct for long.

So the only sound approach is procedural. Establish which financial year the transaction falls in, check the treatment applicable in that year for your own category of shareholder, and check whether tax was deducted at source before the money reached you. For anything material, that is a question for a tax adviser rather than for a market article. The point to carry away is simply that a rupee returned as dividend and a rupee returned as buyback proceeds do not necessarily reach the shareholder with the same amount of tax attached, and that a company choosing between them may be optimising for that difference as much as for anything else.

Some companies grow mainly by buying other companies. Done well, it is a legitimate and powerful strategy: an operator that can buy small businesses at sensible multiples and genuinely improve them compounds faster than one confined to organic growth. Done badly, it is the most reliable value-destruction machine in public markets, because acquisitions let a company report revenue and profit growth indefinitely regardless of whether the underlying economics justify it.

The tell is on the balance sheet. When a buyer pays more than the fair value of the identifiable assets it acquires, the excess is recorded as goodwill. Goodwill that keeps growing while return on capital employed keeps falling means the company is buying growth at prices its own operations cannot support. An impairment of that goodwill later is management's own written admission that the price paid was too high, and it is one of the few places in a filing where a past capital allocation decision is formally marked as a mistake.

The other thing to watch is the gap between what is reported and what is generated. Serial acquirers tend to lean on adjusted profit measures that exclude acquisition costs, integration costs and amortisation of acquired intangibles, quarter after quarter. Costs that recur every year are not exceptional. Track the cash instead: cumulative free cash flow against cumulative cash spent on acquisitions, over a full decade.

A large capital expenditure programme has a predictable and often misread shape. Cash goes out over two or three years while the plant is built. Nothing comes back during that period, because there is no output yet. Then the asset is commissioned, depreciation begins to hit the profit and loss account, and the plant runs at low utilisation while the market for its output is developed. Only after that does the cash actually arrive.

For an outside reader, this means free cash flow looks terrible for several years in exactly two situations that are indistinguishable from the number alone: a company building something valuable, and a company destroying capital on something it should never have built. Free cash flow is not a verdict during a capex cycle. It is a description of where the company is in the cycle.

What separates the two is evidence outside the cash flow statement. Capacity actually being added, disclosed in the management discussion and the fixed asset note. Utilisation rising after commissioning rather than sitting stubbornly low. Capital work in progress converting into fixed assets on schedule rather than ballooning year after year, which is what a delayed or troubled project looks like on a balance sheet. And, eventually, return on capital employed recovering as the new asset base starts earning — the test that finally settles it.

This is also why comparing a company mid-capex against a peer that finished its expansion three years ago on free cash flow alone is meaningless. They are at different points in the same cycle, and the one that looks worse today may be the one that owns the capacity when demand arrives.

None of this can be judged from one annual report. Capital allocation is only visible across a full cycle, because the decision and its consequence are separated by years. Ten years of filings, read as a single series, is the unit of analysis.