How to Build a Capital-Allocation Map in Eight Steps

A capital allocation map sets out every use competing for one year of capital, each sized in rupees, each carrying a return with the basis of that return named. Build it in eight steps: fix the period, count what is available, list every claim, size them, price the returns, force them onto one basis, mark the return the capital already earns, and record what each use forecloses.

What is the thing being built here?

Picture a household that has finished a year with some money left over. The roof needs attention. The scooter is old enough that it costs more each year than the last. There is a gold loan running at a rate nobody enjoys quoting. And an aunt has suggested a small shop in the next lane. Four uses, one pile of money, and the pile is smaller than the four uses added together.

The household actually needs a single sheet of paper with four rows on it. Each row says how much that use costs. Each row says how many years of saving that is. Each row says what it gives back, and it says whether the giving back is a rate someone contracted to pay or a hope somebody has. And each row says what stops being possible if that row wins. Nobody in that household would call the sheet a framework. The sheet is just the list, written down once so that it can be argued with.

A company does the same job on a larger scale and gets it wrong more often. The four rows arrive from four different people in four different formats, and each one arrives sounding reasonable. The sheet built here is that list, made honest. The sheet holds every use competing for one period's capital, sized, priced, and marked with what it costs in options rather than only in rupees. The whole point of writing it on one sheet is that a use which looks affordable on its own stops looking affordable the moment the other claims on the same rupees are sitting next to it.

The worked example throughout is Harivansh Packaging Limited, an invented listed maker of rigid and flexible packaging. Its figures are locked, so every line below can be checked against every other line. Harivansh Packaging has four proposals in front of it: buying Sundarban Polymers Private Limited, funding equity into the Tapti Crossing Infrastructure Private Limited toll road, repaying borrowings, and returning cash to holders. Each of those four routes, and the mechanism behind it, is covered separately. The map only lays the four side by side on one sheet.

Step one: which period, which entity, which basis?

Step one produces no number at all. Skipping it for that reason is the cheapest error on the whole list to avoid. Three things get written at the top of the sheet before a single rupee goes underneath: the period the map covers, the entity it covers, and whether the figures are standalone or consolidated.

The period is one financial year. Not a plan horizon, not a rolling eighteen months, not whatever the longest proposal happens to need. One year. The availability figure in step two is a one-year flow, and everything on the sheet is going to be compared against it.

The entity matters because a group is not a company. If the map covers Harivansh Packaging Limited on its own, then the borrowings on the sheet are its borrowings and the cash on the sheet is its cash. If it covers the group, a different set of figures applies, and a subsidiary's cash may not actually be available to the parent that is doing the allocating.

The basis matters most and is invisible once the sheet is typed. A map that pairs a consolidated figure for earnings before interest, tax, depreciation and amortisation (EBITDAEarnings before interest, tax, depreciation and amortisation. A profit measure taken before financing and before the charge for using up long-lived assets. How it is built was settled in the statements material and is used here as a given.) with a standalone borrowing figure is wrong before anybody starts thinking, and no later care in the arithmetic can repair it. Every figure on the map below is Harivansh Packaging on a standalone basis, and where a consolidated figure appears anywhere in this guide it is labelled in the same sentence as the number. Labelling the basis beside the number is not fussiness. Labelling is the only defence against a ratio silently struck on one base and read against a limit struck on another.

The last part of step one is a date. A share price is a reading taken at an instant, not a property of the business, so any market figure that reaches the map carries an as-of date. The share price of Rs 300/- used later in this guide is illustrative and travels with its own as-of date confirmed at publication.

Step two: how much capital is actually there?

Available capital is three separate quantities stacked on top of each other, and the commonest mistake on a map is presenting it as one. The three are cash generated in the period, cash already held, and borrowing capacity inside a stated ceiling. The three behave differently, they carry different levels of certainty, and only two of them are facts about the business.

Start with generation. Harivansh Packaging Limited earns EBITDA of Rs 477 crore. Finance cost of Rs 60 crore leaves the business, and tax of Rs 75 crore leaves the business. Rs 477 crore less Rs 60 crore less Rs 75 crore leaves Rs 342 crore. The Rs 342 crore is the flow the map treats as generated in the period.

Next comes the Rs 140 crore of cash already held. The cash balance needs no derivation and is the most certain number on the sheet, but notice that a balance is not a flow. Spending it is a one-time act. A household with Rs 1 lakh in a fixed deposit and Rs 30,000 a year of savings knows the difference instinctively and companies forget it constantly.

The third quantity is where the map earns its keep. Borrowing capacity is not a fact about the business. Borrowing capacity is the gap between where net debtBorrowings less cash. It answers what the company would still owe if it used every rupee it holds to pay lenders down today. Settled in the statements material and applied here. stands today and the ceiling the company has said it will not cross. Harivansh Packaging carries net debt of Rs 600 crore. Harivansh Packaging has not said publicly what its ceiling is, so the map shows three of them.

Read the last row of that table twice. The revenue is the same in all three, the plant is the same, the customers are the same and the EBITDA of Rs 477 crore is the same. Only a sentence in a policy moved. Headroom is an output of a stated ceiling rather than a property of the business, so a map that prints one headroom figure has hidden a decision inside a number and presented it as a measurement.

Two arithmetic checks on that drawing, both worth doing yourself. At a 1.5 times ceiling, available is Rs 342 crore plus Rs 140 crore plus Rs 115.50 crore, or Rs 597.50 crore. At 2.5 times the same addition runs Rs 342 crore plus Rs 140 crore plus Rs 592.50 crore, or Rs 1,074.50 crore. The company that can deploy Rs 1,074.50 crore and the company that can deploy Rs 597.50 crore are the same company on the same day.

Step three: which claims go on, including the ones nobody proposed?

Step three lists every claim on the capital counted in step two. Every claim, not every proposal. The distinction is the whole content of the step, and it is the difference between a map that tells the truth about what is free and one that flatters the proposals.

Three sorts of claim arrive without anybody proposing them. Maintenance spending keeps the existing capacity doing what it already does. Contracted commitments were agreed in an earlier period and land in this one. And scheduled repayments, where a lender is owed a specific amount on a specific date and no meeting is going to change that.

The household version is immediate. The roof is not a proposal. Nobody sits down and pitches the roof. The roof leaks, the money for it leaves whether or not anyone puts it on a list, and a household that budgets only the scooter and the shop has already spent money it does not have. A map listing only the things somebody argued for has overstated what is free by exactly the amount of the things nobody argued for.

For Harivansh Packaging the unproposed claim on the sheet is maintenance spending. Here the record forces an honest awkwardness into the open. There is no published capital expenditure figure for this business at all, so maintenance is carried at depreciation and amortisationThe annual charge that spreads the cost of long-lived assets across the years they are used. Settled in the statements material; here it stands in for maintenance spending only because no capital expenditure figure exists. of Rs 138 crore, and the cell says in words that this is a proxy rather than a measurement. Depreciation is an accounting charge on assets bought in the past. Keeping the plant running next year costs a different quantity that happens to be of a similar order, and the map records the difference rather than smoothing over it.

Step four: how big is each claim, in rupees and in years?

Every claim gets two sizes. The rupee figure is the one everybody writes down. The years figure is the one that makes the map do work no list of rupees can do.

Years of generation is the rupee size divided by the Rs 342 crore the business produces in a period. The years figure answers a question the rupee figure cannot: how many budgets does this claim reach into? A use that consumes three years of generation is not a decision about this year. A three-year claim is a decision about this year and the two after it, taken by people who will have left before the last of it is paid for.

Three notes on how those rows were sized. Sizing is where quiet inventions creep in. The acquisition row is split deliberately: Rs 1,140 crore is what actually goes to the sellers of Sundarban Polymers Private Limited, and Rs 180 crore is that business's own net debt. The debt comes across because the purchase is of the whole company. Together they are the Rs 1,320 crore of enterprise valueThe value of a business to all its funders together, so it counts the debt as well as the equity. The bridge from it to what the sellers are actually paid was settled in the transaction valuation material., and quoting that Rs 1,320 crore as the price paid to sellers is the single most common error made about transaction figures. The years column runs on the Rs 1,140 crore, the cash the acquirer has to find in this period.

The two Rs 140 crore rows are sized at the cash balance because that is the only anchor the record supplies. Harivansh Packaging publishes no dividend, no payout ratio and no repurchase history, so the map cannot show what the company has actually returned to holders before. The row exists to show what returning the whole cash balance would cost against the other claims, and its size is an illustration rather than a declared intention.

And the total is worth sitting with. Rs 2,278 crore of claims against Rs 836 crore of availability means the map is not deciding between good and bad uses at all; it is deciding which two thirds of the list will not happen. A sheet that lists proposals one at a time never produces that sentence.

Step five: what does each claim return, and on what?

Step five puts a return in every row, and the rule that makes step five worth doing is that the basis of the return goes in the same cell as the number. Not in a footnote, not in a column heading, not in the analyst's memory. The same cell.

Four things about a basis need naming, and any one of them left unsaid is enough to break a comparison. Whether the return is struck before or after tax. Whether it is struck before or after depreciation. Whether the denominator is book capital or a market price. And whether the number is contractual, meaning somebody has agreed to pay it, or expected, meaning somebody has forecast it.

Work two of those cells yourself, because the arithmetic is short and the habit is the point. The return on capital employedOperating profit divided by the capital tied up in the business. Settled in the statements material and used here as the number the existing capital already delivers. sets Rs 339 crore of EBIT against the Rs 2,390 crore of capital employed standing behind it, and lands at 14.2 per cent. The repayment row is the rate this company actually pays: finance cost of Rs 60 crore over borrowings of Rs 740 crore is 8.11 per cent, and at an effective tax rateThe share of a year's profit before tax that actually left this particular company as tax, which here is Rs 75 crore against Rs 300 crore, or 25.0 per cent. A reading off one year's accounts, never a published rate. of 25.0 per cent that costs 6.08 per cent after tax. The repurchase row is an earnings yieldEarnings per share divided by the share price, so it says what proportion of the price the business earns in a year. Settled in the equities material and applied here as a cost comparison. of Rs 12.50/- over Rs 300/-, or 4.17 per cent, and that Rs 300/- is illustrative and carries its as-of date.

Now look at what happens if the basis column is deleted. The map would read 14.2, 7.42, 13.8, 6.08 and 4.17, five percentages in a neat column, and a reader would rank them without hesitating. Two of those five are struck after tax and three before it. One of the five is struck before depreciation while the other four are after it. One divides by a market price and the rest divide by book capital. The percentage sign looks identical whatever quantity sat above and below the line, so a return without its basis is not a number a map can use.

Step six: can the bases be brought together?

Step six takes the returns that step five recorded and tries to get them onto one basis, so that the numbers become commensurableCapable of being measured on one shared scale. Two quantities can each be perfectly correct and still not be commensurable, in which case comparing them produces a result that means nothing.. Some of that work is arithmetic. Some of it is impossible, and the step is designed so that the impossible cases end up visible rather than quietly resolved.

Two rows on the map are already struck after tax: the repayment at 6.08 per cent and the repurchase at 4.17 per cent. Three are struck before tax. So the conversion needed is to restate the hurdle after tax and set it beside those two honestly.

Multiplying a rounded reading and printing the result as exact is its own small dishonesty. Do the restatement from the underlying figures rather than from the rounded 14.2 per cent. EBIT is Rs 339 crore. At the 25.0 per cent effective rate that leaves Rs 254.25 crore. Set that beside the Rs 2,390 crore of capital employed and the figure is 10.6 per cent. The existing capital earns 14.2 per cent before tax and 10.6 per cent after it, and which of the two is quoted decides whether a 6.08 per cent debt repayment looks distant or merely poor.

Now the cell that cannot be converted. The project's 13.8 per cent is EBITDA of Rs 248 crore over project cost of Rs 1,800 crore, so it is struck before depreciation while every other row is struck after it. Bringing it across would need the depreciation charge on a Rs 1,800 crore asset, and the record for this project carries no such figure. The record carries a project cost, a funding split, a revenue, an operating cost and a debt service schedule for one modelled year, and no depreciation at all.

So the honest cell names the gap: not reconcilable, depreciation on the Rs 1,800 crore asset being absent from the record. The entry is longer than a number and does more work than a number would. The reason tells a reader that the row cannot be ranked against its neighbours, and tells them exactly what would have to be found before it could be. A blank with a reason in it is worth more than a plausible figure. The blank cannot be misread and the plausible figure cannot be caught.

Step seven: where does the hurdle go?

The instinct is to give the existing business a row of its own, sitting among the proposals with 14.2 per cent beside it. Step seven says otherwise. The return the capital already earns is not a competitor for the money. The return already being earned is the level every competitor is read against, so the hurdle belongs across the map as a line rather than inside it as a row.

The difference sounds cosmetic and is not. A hurdle drawn as a row invites ranking: this row beats that row, so choose this row. A hurdle drawn as a line asks a different question of every row at once: does the row clear it? The two questions have different answers here, and the map is built to make sure a reader gets the second one.

Every proposal on this map sits below the line. The drawing exists to produce that sentence, and the sentence is not a conclusion about what Harivansh Packaging should do. The money is currently earning more where it stands than any of the four things being proposed for it, on the evidence assembled so far and with one row unable to be assembled at all.

Notice also the effect of the line on the strongest-looking row. The project's bar reaches nearly to the line and is drawn dashed. A pre-depreciation figure is being read against a post-depreciation line. Depreciation on a Rs 1,800 crore asset is a subtraction rather than an addition, so bringing the bar onto the same basis would only move it further left, never further right. The row closest to clearing the hurdle is the one row on the map that has not been measured the same way as the hurdle, and a map without the line drawn would have let a reader rank it first without ever noticing.

Step eight: what does each use foreclose?

The last column holds no numbers. One sentence per row, saying what stops being possible if that row is chosen. The foreclosure column is the only defence the map has against a use that looks affordable on its own and is not.

Foreclosure is what everybody understands at home and nobody writes down at work. Spend the year's savings on the scooter and the roof waits another monsoon. The trade-off is obvious in a household because the money is visibly one pile. Inside a company the pile is abstract, four proposals arrive on four different days from four different people, and each one is presented against the whole of availability rather than against the remainder.

That last row rewards a second reading, because it is the kind of thing a foreclosure column catches and a returns column never would. Repaying Rs 140 crore of borrowings with the Rs 140 crore of cash drops borrowings to Rs 600 crore while the cash balance goes to nil. Net debt is borrowings less cash, so it stays exactly where it was at Rs 600 crore, and headroom at any stated ceiling is unchanged. The company gets a lower finance cost, roughly Rs 11.35 crore a year at the 8.11 per cent it pays. Room is the one thing the repayment does not buy.

The acquisition row needs the most careful handling on the whole map, and the reason is a basis question again. The Rs 1,000 crore of new borrowing is measured against Rs 354 crore of headroom, and lands Rs 646 crore beyond the ceiling. The headroom figure is struck on this company's own net debt over its own EBITDA of Rs 477 crore, standalone. On that same standalone base, the company's own borrowings go from Rs 740 crore to Rs 1,740 crore against unchanged EBITDA, and leverage goes from 1.26 times to 3.65 times.

There is a second honest figure and it answers a different question. Consolidated, once the Rs 180 crore that Sundarban Polymers itself owes travels with it, as it does on a purchase of the whole company, net debt reaches Rs 1,920 crore, and set beside the Rs 609 crore of EBITDA the two businesses earn between them that comes to 3.15 times. Both figures are correct. Neither is a smaller version of the other. The ceiling and the headroom on this map are standalone, so the standalone 3.65 times is the row that answers whether the proposal fits, and reading a standalone ceiling against a consolidated ratio is the same error the basis column exists to prevent, wearing different clothes.

Two things are worth sweeping the slider for. The Rs 140 crore marker crosses inside the bar just past a stated ceiling of 1.55 times, and the Rs 540 crore project equity crosses inside at 2.39 times. Both crossings are policy events. On the day before and the day after, the packaging plants are identical and the customers are identical, and the only thing that changed is a sentence about how much net debt the company is prepared to carry.

The third marker never crosses. The Rs 1,000 crore of borrowing the acquisition needs would require a stated ceiling of about 3.35 times, off the top of the control entirely. A marker quietly dropped from a chart is a claim made without evidence, so the Rs 1,000 crore marker stays drawn. A use that fits only beyond every ceiling anyone would state is not a row on a capital allocation map; it is a proposal to become a different sort of borrower, and the map should say that in words rather than leave the row looking merely large.

The finished map, with all eight steps run

Here is the artefact itself. Every column below is the output of one step, and reading across a row gives a reader everything the map is capable of telling them about that use.

Two cells on that sheet are doing more work than the rest. The project's basis cell says not reconcilable and gives the reason, and that stops the highest number on the map from being read as the best row. And the acquisition's foreclosure cell says outside the map, stopping the largest proposal from being read as merely expensive. Both cells are words rather than numbers, and both would vanish from a version of this sheet that held only the arithmetic.

When is the map finished, and what is it not?

The map is finished when every row carries four things and not before. A rupee size. A duration in years of generation. A return with its basis in the same cell, or an open cell with the reason for the gap in it. And a foreclosure sentence. A row missing any one of those is a row a reader will fill in from imagination, and imagination fills in favourably.

1. Every row sized in rupeesIncluding the ones nobody proposed, and including the ones sized from a proxy, with the proxy flagged where it is used.
   Checking: does the sheet show a claim that arrived without a meeting?
2. Every row sized in years of generationSo a claim reaching into budgets that have not been written cannot pass as a decision about this period.
   Checking: does any row exceed one year, and does the sheet say so plainly?
3. Every return carrying its basis in the same cellPre-tax or post-tax, pre-depreciation or post-depreciation, book capital or market price, contractual or expected.
   Checking: could two cells on this sheet be ranked against each other honestly?
4. Every unreconcilable cell left open with its reasonRather than filled with the nearest plausible figure. A plausible figure reads exactly like a measurement.
   Checking: is there a cell that says what would have to be found before it could be filled?
5. The hurdle drawn across the sheet rather than inside itSo that every row is read against what the capital already earns instead of against the other rows.
   Checking: is the line on the same basis as the rows it is being read against?
6. Every row carrying a foreclosure sentenceNaming what stops being possible, in words rather than as a quantity.
   Checking: does any row foreclose something the sheet does not mention anywhere else?

And now the harder half: what the finished map is not. The finished map is not a ranking. The rows are not commensurable on the evidence assembled, and one of them could not be made commensurable at all, so putting them in order would be adding an ordering the cells do not support. The map is not a decision either, and nobody who reads it is relieved of making one. Every route on the sheet is covered in full separately, with its mechanism, its paperwork and its risks, and none of that reaches a cell.

A map presented as a recommendation has quietly added something no cell in it supports, and the thing it added is the very judgement the reader was supposed to make. The map removes the excuses: nobody can now say a use was affordable without saying against which ceiling, or that a return was strong without saying on what basis, or that a commitment was one year's decision when the years column says 3.33.