How Capital Allocation Affects Long-Term Firm Value

Capital allocation is the running decision about where each rupee of a company's capital goes. One rupee traced through Sankalp Industrial Systems Limited, invented for this guide, shows the shape of it. Raised at a blended 12.00 per cent and already earning 15.00 per cent, it carries a 3.00 point spread, and that spread on Rs 12,00,00,00,000 of capital is the Rs 36,00,00,000 of value created in the year. Value follows the spread.

The mechanism underneath that answer is one subtraction, repeated. Take what the capital earns, subtract what the capital costs, and multiply by how much capital there is. Every allocation decision a company ever makes changes one of those three terms and nothing else. A long-term value is nothing more exotic than that subtraction accumulated over many years. So allocation is judged over a decade rather than over a quarter, and the record of many decisions tells more than the loudest single one in it.

Every rupee figure below belongs to one made up manufacturer and travels with the assumption that produced it. Each figure was chosen so that every step of the arithmetic divides cleanly and can be worked through by hand. A real set of accounts never behaves so obligingly.

What is capital allocation, and how is it different from one investment decision?

Consider a household that runs a small sweet shop. In March somebody asks whether to buy a second mixing machine. Buying the second machine is one decision, and it has a beginning and an end: money goes out, output goes up, and by June it is roughly clear whether it worked. A different question sits behind that one. Over the last five years, where did every rupee this household earned actually end up? Some went into the shop. Some went to the man who lent them the deposit. Some went home as living money. Some sat in the bank doing nothing in particular. The question about where every rupee ended up is capital allocation, and it never closes.

A single investment decision is an event with a date on it. Capital allocation is the standing habit that produces those events, one after another, for as long as the business exists. Nobody sits down on a Tuesday and does the allocation. Allocation happens through hundreds of small yeses and noes, most of them never written down as decisions at all: the machine that was quietly not replaced, the customer terms that were quietly stretched, the dividend that was raised by twenty paise because last year it was raised by twenty paise.

The difference matters for a plain reason. One decision is a small share of the capital, so one decision can go badly wrong and still be recovered from. The habit is applied to everything, so the habit cannot go wrong and be recovered from. A company is, in the end, the accumulated output of its allocation habit, and the balance sheet is a photograph of where that habit has put the money so far.

Where can a rupee actually go?

There are fewer destinations than people expect. A rupee inside a company can be put back into the business, handed to the lenders, handed to the owners, or left sitting as cash. Those four are the complete list. Everything a chief financial officer does with money is one of those four or a combination of them. The list is worth drawing for one reason: only one of the four changes the amount of capital actually at work.

Put back into the business, it buys machines, builds inventory or extends credit to customers, and the capital tied up in trading goes up. Handed to the lenders, the capital does not shrink, it merely changes hands: the same plant is still there, and somebody else now has a smaller claim on it. Handed to the owners, the mirror image occurs on the other side of the sheet. Left in the bank, it still belongs to the company, but it is not doing any work.

Step one. What does the rupee cost before anybody decides anything?

Sankalp Industrial Systems Limited makes industrial valves and precision castings. At market values its shareholders have put in Rs 18,00,00,00,000 and its lenders Rs 6,00,00,00,000. The total is Rs 24,00,00,00,000, being 75.0 per cent equity against 25.0 per cent debt. Blending the price of those two gives 12.00 per cent. How that blend is built, input by input, is covered separately, and the finished figure is taken and used here.

The single most useful thing to understand about that 12.00 per cent is that it applies to every rupee inside the company, and not only to the rupees that were borrowed. There is no bank account inside Sankalp holding cheap money and another holding expensive money. Once the capital is in, it is one pool at one price. A rupee that sits in a warehouse of slow-moving spares costs 12.00 per cent a year. A rupee sitting in a customer's overdue account costs 12.00 per cent a year. Neither of those charges appears anywhere in the accounts. Both are real anyway, and every figure below depends on them being real.

The household version is uncomfortable and exact. If a household has borrowed at 12 per cent to run a shop, then leaving Rs 50,000 idle in a drawer for a year is not free. The idle Rs 50,000 costs Rs 6,000. Nobody writes a receipt for it and no bank statement shows it, but the money is gone all the same.

Step two. What does the rupee earn once it is in the ground?

Two lines make up the capital actually at work here. The trading cycle, the net working capitalWhat the day to day trading cycle ties up: money owed by customers plus goods on the shelf, less what is still owed to suppliers., ties up Rs 1,80,00,00,000. The plant, the net fixed assetsLand, buildings, plant and machinery carried at what is left of their cost once the wear charged so far has been taken off., ties up Rs 10,20,00,00,000. Together they come to Rs 12,00,00,00,000 of invested capitalThe total a business has tied up in order to trade, counting both the trading cycle and the plant. It is what the capital charge is applied to.. Against that, operating profit after tax in the base yearThe last completed year, the one every forecast year is counted forward from. Here it is called Year 0. is Rs 1,80,00,00,000. Divide the one by the other and what the capital returned comes out at exactly 15.00 per cent.

Two things are worth pausing on. First, this is a return on the capital, not a margin on the sales, and the two answer different questions. A margin asks how much of each rupee of revenue survives to the bottom. A return on capital asks how hard each rupee of capital is being made to work. Second, the figure is 15.00 per cent for the capital already in the ground. The company then puts a further Rs 1,00,00,00,000 to work in each of the five years that follow, and the forecast assumes every one of those rupees earns 18.00 per cent rather than 15.00.

The gap between 15.00 and 18.00 is the hinge of everything that follows, and nobody measured it. Somebody assumed it. The assumption gets named properly below, together with the condition it needs and the evidence anybody has produced for that condition. For now, hold on to the shape: old money at 15.00, new money at an assumed 18.00, and every rupee of both charged at 12.00.

Step three. What does the gap between the two produce?

Fifteen earned against twelve paid is a spread of exactly 3.00 points. Applied to the Rs 12,00,00,00,000 of capital at work, that is Rs 36,00,00,000 of value created in the year. Three numbers and one multiplication, and it is the entire mechanism. The name of that measure, and the second route to the same figure, are settled in a subject of their own. The arithmetic does the work first and the vocabulary can wait.

Notice the shape of it. Value created is a rectangle: the height is the spread and the width is the capital, and the rupees are the area. Because the rectangle shows immediately what happens when a company grows, it does more work than any definition. Growing means widening the rectangle. Widening a rectangle only adds area if the rectangle has some height. A rectangle of zero height, widened, gives zero, however impressive the width becomes.

There is a second way in, through the capital chargeWhat a year of using the capital costs, worked out by applying the cost of capital to the amount tied up. It appears in no set of accounts.. Apply the 12.00 per cent to the capital at work and the charge for the year is Rs 1,44,00,00,000. The year's operating profit after tax, Rs 1,80,00,00,000, has to clear that charge before a single rupee is left over, and what remains once it has is Rs 36,00,00,000 all over again. Both routes are the same subtraction wearing different clothes.

Where does every part of the Rs 36,00,00,000 come from?

A number that arrives whole is a number a reader has to take on trust. The Rs 36,00,00,000 does not need to be taken on trust at all, so here it is line by line.

Run it the other way as a check. A return of 15.00 per cent against a cost of 12.00 per cent is 3.00 points, and 3.00 per cent of Rs 12,00,00,00,000 is Rs 36,00,00,000. The two routes are algebraically the same statement, so they must agree exactly, and they do.

The most important line in that table is the one no set of accounts anywhere will print: the charge for the capital. A profit and loss account will charge Sankalp for its interest. No statement it publishes will ever charge it for the Rs 18,00,00,00,000 of shareholder money, and shareholder money is three quarters of the funding. The silence about shareholder money is why a company can report a rising profit for a decade while quietly consuming value the whole time. The same silence is why an allocation question can never be answered off the reported numbers alone.

Why does the return on invested capital climb to 15.88 per cent by Year 5?

By the end of Year 5 the capital at work is Rs 12,00,00,00,000 plus five years of Rs 1,00,00,00,000, or Rs 17,00,00,00,000. Operating profit after tax is Rs 2,70,00,00,000. Divide and the return on invested capital reads 15.88 per cent, up from 15.00.

A reader seeing that in a presentation would take it as good news about the factory. The climb is nothing of the kind. The return climbs for exactly one reason: new capital is assumed to earn 18.00 per cent while the base it is being added to earns 15.00, so the average of the two drifts upward every year that new money arrives. Nothing about the existing business changes by a single rupee across the whole forecast. No line runs faster, no machine is upgraded, no customer pays sooner.

The arithmetic of an average makes this obvious once it is drawn. The Rs 5,00,00,00,000 of new capital is 29.41 per cent of the Rs 17,00,00,00,000 total, so the blended return sits 29.41 per cent of the way along the road from 15.00 to 18.00. Three points travelled 29.41 per cent of the way is 0.88 of a point, and 15.00 plus 0.88 is 15.88. Nothing else moved the number.

What did this company actually allocate, over the five years to the base year?

Here is the record. The record is short, it is the kind of thing any company will put on one slide, and reading it is the ordinary way people form a view about whether a management team is any good with money.

The Year 0 figure of Rs 3.60 is Rs 2.60 of regular dividend plus a special dividendA payment declared once and marked as not repeating, so that nobody treats it as the floor for next year. of Rs 1.00, being Rs 52,00,00,000 and Rs 20,00,00,000. Alongside all of that, the forecast puts a further Rs 1,00,00,00,000 to work inside the business, once in each of the five years ahead, a quarter of it funded by net new borrowingWhat was drawn from lenders in the year after taking off whatever was repaid to them. A positive figure means the lenders put money in rather than taking it out. of Rs 25,00,00,000 a year. Not one year of this record moves the cash balance of Rs 1,20,00,00,000, of which Rs 80,00,00,000 is excess cashThe part of a cash balance the trading cycle does not need, which could be handed out tomorrow without stopping anything..

Look hard at that record and notice what it does not contain: it does not say what any of the spending earned, and it does not charge for the capital used. Three of the four destinations are in use here. Money went back into the business, money went out to owners twice over, and cash sat still. The lender line is running backwards: nothing was repaid and Rs 25,00,00,000 a year came in, so the lenders are a source in this record rather than a destination. Five lines of real, disclosed, checkable activity, and not one of them establishes whether a rupee of value was created.

What is the one assumption the whole result rests on?

Everything in this guide that moves, moves because of a single number: the next rupee is assumed to return 18.00 per cent, against 15.00 on every rupee already sunk. Nothing else here carries anything like that weight. The 18.00 per cent is a choice somebody made rather than a fact somebody measured. Read it precisely: it does not claim what most readers take it to claim.

The tempting reading is that the 18.00 per cent is a claim about margins, as though newer plant were somehow more profitable on each rupee of sales. The 18.00 per cent says nothing about margins. Look down the profit line and nothing varies: every rupee of revenue in this model leaves exactly the same paise behind after tax, whichever vintage of capital produced it. The difference sits somewhere else altogether, in how hard each rupee of capital is made to spin.

Read the top row twice. The two amounts in it are identical, and that is not a slip. A turn of exactly 1.00 means the capital and the revenue standing on it are the same size, by definition rather than by accident. Now read the row below: the added revenue is a fifth larger than the money that bought it, the margin applied across both rows never changes at 15.00 per cent, and the whole of the 18 against 15 falls out of that one difference. Rs 18,00,00,000 is what a year of new money adds to the profit line, and the forecast takes that same step five times over. So the claim underneath this forecast is not that the newer business earns more per sale, but that the newer money spins one and a fifth times where the older money spins once.

Stated as a claim about turns, the assumption is one somebody can actually go and interrogate. A vendor with a Rs 20,000 cart that turns over Rs 24,000 of sales in a year is running at 1.20 turns. If he buys a second cart for Rs 20,000 and it sits in a slower lane doing Rs 16,000, his new capital is spinning at 0.80 and his blended return has fallen, whatever the profit line says.

What has to hold for the 18.00 per cent, and who has shown that it does?

For the 18.00 per cent to hold, the new capacity would have to be genuinely more productive per rupee than the capacity already installed. Concretely: the next valve line would have to produce Rs 1.20 of revenue for every rupee sunk into it, where the existing plant produces Rs 1.00. The claim is not absurd. Newer machines are often faster, a new line can be laid out better than one that grew by accretion over twenty years, and a business adding capacity into demand it already has does not need to spend on winning the customer twice.

Evidence for that extra productivity is missing entirely. Nothing in this case records a productivity study, an engineering estimate, a comparison of the old line against the new, or even a management statement to that effect. The 18.00 per cent sits in the forecast because somebody put it there. Every rupee of the climb from 15.00 to 15.88 per cent, and Rs 30,00,00,000 of the Rs 66,00,00,000 of value created at Year 5, rests on it.

A reader who wants to disagree with anything here should therefore start there and not anywhere else. Attacking the 12.00 per cent cost of capital moves the answer, but that figure was built input by input from observable weights and the build can be inspected. The revenue and the capital move together, so attacking the revenue forecast moves the answer less than expected. The 18.00 per cent is the one number with nothing underneath it, and it is doing the heaviest lifting of all.

What does the same arithmetic do when the spread runs the other way?

Everything so far has assumed the spread is positive. Turn it over. If new capital earns less than the 12.00 per cent it costs, then every rupee added subtracts. Not less than it might have added: subtracts, in the plain sense that the company is worth less afterwards than before.

Take the same forecast and assume new capital earns 10.00 per cent instead of 18.00. Five years of Rs 1,00,00,00,000 still goes in, so the capital at work still ends at Rs 17,00,00,00,000. Operating profit after tax at Year 5 is Rs 2,30,00,00,000, the charge for the capital is Rs 2,04,00,00,000, and value created is Rs 26,00,00,000. Value created was Rs 36,00,00,000 at Year 0. Rs 5,00,00,00,000 of new capital has destroyed Rs 10,00,00,000 of value, and it happened while every year looked like growth.

Check it directly and the answer is the same. The new capital earns 10.00 per cent on Rs 5,00,00,00,000, or Rs 50,00,00,000. The charge on the same amount at 12.00 per cent is Rs 60,00,00,000. The difference is Rs 10,00,00,000, going out.

So the whole space has one line running across it, and the line is at the cost of capital. Above it, growth adds. Below it, growth subtracts. Exactly on it, growth is completely neutral and a company can double in size without moving its value by a rupee. The neutral case is the one nobody expects, and it is worth sitting with for a moment. Growth is not a direction of travel; it is a multiplier on whatever sign the spread already has.

What do five years at 12.00 per cent actually produce?

The neutral case is the least intuitive thing in this whole trace, so it deserves working out rather than asserting. Suppose new capital earns 12.00 per cent instead of 18.00, with nothing else changed. The comparison is constructed out of the locked inputs.

In the middle column, five years of allocation produced exactly nothing: Rs 36,00,00,000 at the end, Rs 36,00,00,000 at the start, on Rs 5,00,00,00,000 more capital. Every one of those five years had a capital plan, a board approval, a set of purchase orders and a paragraph in the annual review about disciplined growth. All of it was real. None of it moved the number the company exists to move.

And the middle column is not the bad case. The middle column is the neutral one. The right hand column is the bad case, and it is only two points further along.

How is an allocation record told apart from an allocation story?

Telling the two apart turns out to be a checkable procedure rather than a matter of judgement, and the test is short. A record states three things: what was spent, what the spending earned, and what the capital used cost. A story states the first and calls it investment.

The test is not whether the numbers are big, audited or detailed; it is whether all three terms of the subtraction are present. A list of amounts spent, however precise, cannot answer the question, because it supplies one term out of three. Two terms are not enough either: a company that states what it spent and what it earned still has not charged for the capital, and a positive return is meaningless until what it had to beat is known.

There is a second, softer signal worth knowing. A record names its assumptions and a story hides them inside a result. Sankalp's own record, once the 18.00 per cent is written down as an assumption rather than presented as an achievement, passes the test. The record is not good because the number is high. The record is good because the number is visible and can therefore be argued with.

What can a long record show, and what can it still not show?

Stretch the window and something appears that no single year contains. One positive spread can be an accident: a good year in the sector, a competitor's plant that went down, a contract that happened to land. Ten years of positive spread across different conditions is much harder to explain that way. A long record does not prove skill, but it does make luck an expensive thing to keep believing in.

The habit is really the whole argument for judging allocation over a decade. The signal in any one decision is drowned in noise. Across a hundred decisions the noise partly cancels and the habit does not, so the signal comes through.

A long record still cannot show what would have happened under a different allocation. Sankalp put Rs 1,00,00,00,000 a year into the business. The alternative was never run, so nobody will ever know what it was worth. No experiment was performed and none can be. The honest reading of even an excellent record is that the spread held, repeatedly, under the conditions that actually occurred. The spread holding is a real and useful thing to know. Knowing it is not the same as knowing the money went to the best available place.

How does a lender, an analyst or an owner actually use this?

Three people read this arithmetic and each of them reads a different line of it. The same set of figures does not mean the same thing to all three.

A lender reads the borrowing line and asks what is on the other side of it. Sankalp draws Rs 25,00,00,000 a year and repays nothing, so the debt is growing while the plant it funded is depreciating. A lender is not paid more when the spread is wide, so the upside of a good allocation does not reach them; what reaches them is the downside of a bad one. Their question is therefore narrow and practical: is the new capital producing cash that services the growing balance, or is the borrowing funding the dividend by a longer route?

An analyst reads the return line and immediately looks for what moved it. Faced with a climb from 15.00 to 15.88 per cent, the useful next question is never whether the number went up. The question is which of two things moved: the existing base, or the mix. On this company it is entirely the mix, and an analyst who splits it that way has learnt something the headline hides. The practical form of that habit is to rebuild the return on the opening capital rather than the closing capital, and see whether the answer survives.

An owner reads the whole record and lands on one question: what did the retained money buy? Rs 3,04,50,00,000 came out to owners over five years and Rs 5,00,00,00,000 is going back in over the next five. The second figure is larger than the first. An owner is being asked to reinvest more than they were paid, on the strength of an 18.00 per cent that nobody has evidenced. Being asked that is perfectly reasonable. Being asked it without being told is not.

And the household version, the same question in smaller money. If the shop earns 15 paise a year on every rupee put into it and the loan costs 12, then putting the year's surplus back in is worth doing, and the household keeps three paise a rupee for the trouble. If a slower second shop would earn 10, the same surplus is better used almost anywhere else. Nobody in that shop would call this capital allocation. The choice is the only decision they make all year that they cannot undo.