Cost of Capital: The Calculator and What Each Input Moves

Two prices go in and one rate comes out. Type in a base rate, two premium halves, a beta, a borrowing cost, a tax rate and the two market values; the tool prices the equity, takes the tax off the borrowing, and weights the pair at market. On the locked figures for Sankalp Industrial Systems Limited, invented, it returns exactly 12.00 per cent, and it stops there.

A weighted average cost of capital (WACC) is arithmetic, not judgement, and that single fact is why a calculator settles the sum and settles nothing about the choices behind it. Once the eight figures are chosen, five short lines produce the answer and there is nothing left to interpret. Every genuinely hard question in this subject lives upstream, in how those eight figures were arrived at, and that work is set out separately. The tool computes instantly and correctly from whatever it is handed. The freed attention goes to the only question a calculator can actually answer: what happens to the rate when one of the eight moves.

The shape of the problem is the same at every size, and at a small size the whole of it fits in one head. Something smaller than a listed company therefore makes the better starting point. A couple runs a tailoring unit out of two rooms. The bank lent them Rs 6,00,000 against the machines, and the rate is written on the sanction letter. The couple put in Rs 18,00,000 themselves, money that had been sitting in a deposit earning something every year. Asked what their money costs, they quote the bank rate. The bank rate is the only number anybody wrote down. But three quarters of the money in that business is theirs, and the price of theirs is the deposit they gave up plus something for the risk of failure that a tailoring unit carries and a bank deposit does not. Nobody prints that price on anything. To answer the question honestly they have to blend a written price with an unwritten one, weighting each by how much of the money it represents. Blending a written price with an unwritten one is the entire calculator, at one hundred thousandth of the scale, and it already shows where the difficulty is going to sit.

What is typed in, and where does each of the eight figures come from?

The tool asks for eight figures and not one of them is optional. Three feed the equity side directly, one is a measure of business risk, one is a tax rate, one is a borrowing cost, and two are market values. Here they are, loaded with the assumptions this worked example uses for Sankalp Industrial Systems Limited at Year 0. Every one of these is an assumption of the example, chosen round so the arithmetic can be checked by hand, and a live market would put a different figure on several of them.

The list of eight repays a second look. Fields 1, 2 and 3 are prices of things nobody sells directly. Field 4 is a statistic somebody computed from a set of other companies. Field 5 is an accounting outcome. Only field 6 is written into a contract that somebody signed, and only fields 7 and 8 can be read off a screen. Five of the eight are estimates, and the weight of the answer sits on them. A base rate or a premium for today has to come from somebody who publishes one, and Aswath Damodaran keeps current estimates of all three at pages.stern.nyu.edu.

Three of the eight do double duty. Most readers miss that part on a first pass. The effective tax rate is used once to compute the after-tax borrowing cost and once again in the step that puts the company's own borrowing back onto the beta. The two market values are used once to set the weights and once again to form the ratio that same step needs. So a field changed for one reason quietly moves a second line as well, and the drawing below is worth a minute for exactly that.

The five computed lines: how do eight fields become one rate?

Between the fields and the answer sit five short computed lines, and the tool prints every one of them on screen rather than hiding them inside the result. Printing all five is not decoration. A rate that arrives as a single number is a rate nobody can argue with, and almost every useful conversation about a cost of capital is an argument about one of these five lines rather than about the last one.

Line 2 deserves a second look because it is the only one that is not a single operation. The debt to equity ratioWhat a business has borrowed set against what its shareholders stake is worth. Both halves are taken at market prices here rather than at whatever the books happen to record. is Rs 6,00,00,00,000 over Rs 18,00,00,00,000, or one third. One less the tax rate is 0.75. Multiplying the two gives 0.25; adding one gives 1.25; and the unlevered 1.00 scaled by that 1.25 is a levered beta of exactly 1.25. Two different quantities in this example both come out at 0.75, and they mean nothing like the same thing.

Cost of Equity: which fields build the 14.00 per cent line?

The cost of equity is the price of the money that has no contract behind it, and the calculator builds it from a base rate plus a risk premium that has been scaled by the beta. Field 1 supplies the base rate of 7.75 per cent, being what a long-dated government security in the same currency yields. Fields 2 and 3 supply the two halves of the premium: 3.50 per cent for holding shares at all rather than lending to a government, and 1.50 per cent for the jurisdiction the earnings come from. The two halves add to 5.00 per cent. The levered beta of 1.25 then scales that premium, giving 6.25 points, and 7.75 plus 6.25 is 14.00 per cent.

The two halves of the premium are separated for a reason, and it is not tidiness. The equity risk premiumLending to a government pays less than holding shares does, and the yearly gap between the two is this. How anyone lands on a figure for that gap has its own treatment. for a settled market and the country risk premiumAn extra slice added when a business earns in a place whose own government paper is not treated as the benchmark. Where that slice comes from is worked out separately. are estimated by completely different methods, they are argued about by different people, and a reader who disagrees with the total usually disagrees with only one of the two. Collapsing them into a single field would hide which half the argument is about. How each of them is actually estimated, and why two careful people land on different figures for the same market on the same day, are questions covered separately. Each simply gets its own box here.

The same applies with more force to field 4. The unlevered beta measures one thing: the size of the swing in a company's own equity return set against the swing in the market's. The beta arrives already measured and with its meaning stated, rather than being derived here. Where a beta is measured from, how a peer set is assembled, and what to do about the difference between the beta of a business and the beta of its shares are all covered separately. The question taken up at length below is what a beta is worth once there is one to work with.

Cost of Debt: why is the field labelled pre-tax, and what does the tool do next?

The field asks for the rate before tax and the tool applies the tax itself. No decision on the whole screen matters more. Sankalp Industrial Systems Limited pays a blended 8.00 per cent across its three borrowing tranches. The blended 8.00 per cent is the figure that goes in. Underneath it sits the effective tax rateTax actually borne divided by profit before tax. It differs from whatever headline rate a government legislates, and each company has its own. field at 25.0 per cent. Directly under both, in a strip that cannot be typed into, the tool shows 6.00 per cent, being 8.00 times 0.75.

Why does the tax come off at all? Interest is subtracted before the tax bill is computed, so the true price of borrowing a rupee is not the whole rupee. At a 25.0 per cent rate, twenty-five paise of every rupee of interest comes back as tax the company never has to pay. Seventy-five paise is the real price. The reduction of twenty-five paise in the rupee is the tax shieldThe tax a business does not pay because interest was subtracted before the bill was worked out. It is a saving created by borrowing rather than a payment made., and the reason it belongs in the rate rather than in the cash flow is that the cash flow this rate is built to discount is worked out on a tax charge that ignores borrowing entirely. The deduction has to be counted once. The calculator counts it here.

Weighted Average Cost of Capital: what guards sit on the two weight fields?

The weights are taken from market values, and the tool will not let those two fields be labelled anything else. Field 7 asks for the market value of the equity and shows the arithmetic it wants: a share count times a price. Field 8 asks for borrowings at market. Their total here is Rs 24,00,00,00,000, so the weights come out at exactly 75.0 and 25.0 per cent. The market-value rule is the second guard built into the interface, and it exists because the balance sheet sitting on the desk offers a different pair of numbers that look just as official.

Book values and market values answer different questions. A book figure for equity records what shareholders paid in years ago plus what has been kept back since. A book figure carries no view whatever on what the business would fetch this morning, and the weights are asking about this morning. The market capitalisationEvery share in issue, valued at the price one of them trades at. It is what the stock market says the whole equity stake is worth today. is the answer to the question actually being asked. Debt is the half where the book figure is usually close enough to be harmless, and that near-enough is exactly what makes the trap dangerous. Nearly all of the damage lands on the equity side, and it always leans the same way.

Two consequences fall out of that formula and both of them can be checked in the head. Since the weights add to one, the answer must sit between the two prices: never above 14.00 per cent and never below 6.00. And since the debt weight here is only a quarter, three quarters of everything that happens to this answer happens on the equity side, where none of the inputs can be observed. Everything below this line follows from that second consequence.

Cost of Equity vs Cost of Debt: which price is estimated and which one is contracted?

One of these two prices is written in a contract somebody signed and the other is assembled out of three estimates, and the calculator gives the assembled one three times the weight. Set them side by side and the asymmetry is uncomfortable. The 8.00 per cent is a blended couponThe rate written into a loan or a bond that fixes what the borrower must pay the lender each year. across three tranches whose terms are on paper and whose rates cannot be argued about. The 14.00 per cent is a base rate that has to be chosen, a premium that has to be estimated, a country slice that has to be estimated by a different method again, and a beta that has to be taken from somebody else's peer set and then adjusted.

Think about the caterer for a wedding. The per-plate rate is negotiated hard, written down and remembered by everybody, and it is the number the two households argue about for a week. The guest count is a guess. Move the plate rate by five rupees and the bill moves a little; move the guest count by fifty and the bill moves a great deal more. Everybody argues about the written number because it is the one that feels arguable, and the guess sitting quietly beside it does most of the damage. A cost of capital has the same shape, and the sensitivity table below puts numbers on it.

The sensitivity table: what is one unit of each input worth?

When one field moves by one unit and every other field is held exactly still, the answer moves by an amount that can be computed once and then carried around in the head. The table below sets those amounts out, and few rows anywhere in this subject earn their space so easily. Every figure in it is stated in basis pointsRate counted in hundredths of a percentage point, so a move from 12.00 to 12.25 is twenty-five of them. Lenders and traders count this way because half a point sounds vague and fifty of these does not.. The differences being argued about are usually smaller than a percentage point, and calling them fractions of a point makes them sound negligible when they are not.

The ranking is what to read, rather than the rows. A full point of premium is worth 93.75 basis points and a full point of borrowing cost is worth 18.75, so the input nobody can observe moves this answer exactly five times as hard as the input written into a loan agreement. Even a single tenth of a beta sounds like a rounding difference and is worth 37.5 basis points, twice what a whole point of borrowing cost does. The two contested inputs dominate, and they dominate the two inputs anybody can look up.

The beta: why is one tenth of it worth 37.5 basis points?

The relationship between the levered beta and this answer is a straight line whose slope can be memorised in one sitting. Hold every other field where the worked example put it and the whole calculator collapses to a single expression: the answer is 7.3125 plus 3.75 times the levered beta. Check it at the default. Three point seven five times 1.25 is 4.6875, and 7.3125 plus 4.6875 is exactly 12.00. A slope of 3.75 points of rate per whole unit of beta means one tenth of a beta is 37.5 basis points, and half a beta of disagreement is 187.5 basis points, or nearly two full points of discount rate.

Where does 3.75 come from? A tenth of beta adds a tenth of the 5.00 per cent premium to the cost of equity, being 50 basis points, and then 75.0 per cent of that reaches the blend. Nought point one times five times nought point seven five is 0.0375, or 3.75 points per unit of beta. The slope is fixed by the premium and the equity weight together, so it is not a fact about betas in general. Change the premium or change the weights and the slope changes with them. Recompute the slope for whatever company is actually under examination rather than carrying this one around.

What happens when the debt weight moves, and what is being held still?

Raise the borrowing and two things move at once. Moving the debt weight is the least intuitive experiment on the whole screen for exactly that reason. Take field 8 from Rs 6,00,00,00,000 to Rs 9,60,00,00,000 and leave the equity value where it is. The total capital becomes Rs 27,60,00,00,000, so the weights shift from 75.0 and 25.0 to 65.2 and 34.8 per cent. More weight now sits on the cheaper 6.00 per cent, and the extra weight pulls the answer down. But the debt to equity ratio has risen from one third to eight fifteenths, so relevering carries the beta up to exactly 1.40, the cost of equity climbs to 14.75 per cent, and that pushes the answer back up.

The two moves do not cancel. The answer lands at 11.71 per cent, a fall of about 29 basis points. Had the beta stayed at 1.25, the weight shift on its own would have taken the answer down to 11.22 per cent, a fall of about 78, so relevering claws back exactly five eighths of what the weights gave away. The clawback is worth feeling rather than being told: a reader who moves the weight and expects the answer to drop the way the weights dropped is going to be surprised by how little happens.

The tool says plainly, on screen, that two of the things it is holding still would not hold still anywhere outside it. A company that borrows more pays more to borrow, so the 8.00 per cent would not survive the move. And the market value of its equity would not sit at Rs 18,00,00,00,000 while its balance sheet changed shape underneath. Both of those are real and both are large. Two further questions are worked out separately and at length: what extra borrowing does to the price of borrowing, and how a different funding mix would change what the business as a whole is worth. Moving a weight answers a narrower question than either of those two, and the narrower question is the one a blended rate can settle.

Why does the tax rate barely move the answer?

One point on the tax rate moves this answer by about two basis points, and the deduction that same rate creates is worth two hundred. Both of those are true at once and holding them together is the whole lesson of this block. Take the tax rate from 25.0 to 26.0 per cent with the levered beta held. The after-tax borrowing cost falls from 6.00 to 5.92 per cent, a move of eight basis points, and only a quarter of that reaches the blend because the debt weight is 25.0 per cent. Two basis points. The tax rate is the smallest row in the sensitivity table by a distance.

Now ask a different question: what is the whole deduction worth? Without it the debt would enter the blend at the full 8.00 per cent, and the answer would be 0.75 times 14.00 plus 0.25 times 8.00, being 12.50 per cent. The deduction is worth the whole 200 basis point gap between 8.00 and 6.00 on the debt line, or 50 basis points of the blend. A small derivative and a large level are different animals, and a field can have one without the other.

So carelessness with field 5 does not usually cost two basis points. Carelessness costs whatever the gap is between the figure used and the figure that should have been used, and the two commonest ways of getting it badly wrong are large. Reaching for a headline statutory rate where the company's own effective rate belongs can be several points out. Reaching for a rate that already has some other relief baked into it can be worse. The tax rate also enters the relevering step, where a higher rate makes the levered beta a little lower, and that second effect is worked out separately.

CAPM vs WACC: which line of this calculator is which?

Lines 1 to 3 of this calculator are the capital asset pricing model and line 5 with the blend is the weighted average, and the tool prints them as two separate readouts precisely so they cannot be mistaken for each other. Sharpe's model, from his 1964 paper in the Journal of Finance, prices one thing and one thing only: the return shareholders require. Feed it a beta, a premium and a base rate and 14.00 per cent comes out. Then it stops. The weighted average takes that 14.00 per cent as one of its two ingredients, pairs it with the 6.00 per cent after-tax borrowing cost, and blends them at the market weights to reach 12.00 per cent.

The two models are not rival routes to the same destination, and neither is a shortcut to the other. One feeds the other. The error this distinction prevents is a specific and expensive one: discounting a cash flow that belongs to everybody who funded the business at a rate that describes what only the shareholders require. The cash flow of the whole firm is claimed by lenders as well, and lenders are charging 6.00 per cent after tax rather than 14.00. The wrong readout makes every value computed afterwards too low, consistently, in the same direction, for a reason nobody on the review will spot from looking at the answer.

How the Cost of Capital Affects Firm Value: where does this rate go once it is settled?

The calculator produces a rate and stops, and the rate then does at least three quite different jobs, each of which converts basis points into something a business actually decides. Twenty minutes of arguing about a beta buys something only when the argument reaches a decision, and three decisions are where it lands. How a value behaves as the funding mix changes is a separate subject with its own treatment.

The first job is discounting. The free cash flow to the firmLenders and shareholders together are paid out of this: what a business has left once it has met its tax and funded its own growth. of this company, discounted at 12.00 per cent with year-end discounting, produces an enterprise valueWhat the whole operating business is worth, counting the lenders claim and the shareholders claim together. of Rs 21,28,13,79,094. A move in the rate moves that figure the other way, and by more than most readers expect. Most of the value sits far out in time, where the rate has compounded against it many times over. How much more, and why, is worked through separately.

The second job is as a hurdle. Sankalp Industrial Systems Limited appraises the projects on its list at its own 12.00 per cent. Project 1 on that list, the third valve line, costs Rs 2,00,00,00,000 at the outset and pays back Rs 65,00,00,000 a year for five years. Discounted at 12.00 per cent it is worth Rs 34,31,00,000 more than it costs, rounded here to the nearest lakh, and its own internal rate of return works out at 18.72 per cent. Project 1 clears the hurdle comfortably. A project sitting closer to the line does not, and a rate that is 94 basis points wrong changes which side of the line some project falls on. The rules for making that decision are set out separately, and what matters at this stage is only that the rate decides it.

The third job is quieter and it is the one that makes the rate feel real. Sankalp Industrial Systems Limited has invested capital of Rs 12,00,00,00,000. At 12.00 per cent, the yearly charge that capital owes before the business has created anything at all is Rs 1,44,00,00,000. The company earns 15.00 per cent on that capital, so the spread is exactly 3.00 points and there is something left over. Move the rate by one point and the yearly charge moves by Rs 12,00,00,000, a real amount of money for a business this size. The comparison of that spread has a name and a use of its own, both treated separately.

One number in this guide turns up twice, and the two appearances are unrelated. The cost of the double deduction is 37.5 basis points, and one tenth of a levered beta is also worth 37.5 basis points. Neither of those follows from the other and neither explains the other; they arrive at the same figure by two unconnected routes and it means nothing. Each is labelled where it is used, and the coincidence does no work.

How is a rate handed over by somebody else checked?

Three questions, none of which needs the model that produced the rate, will catch three different classes of error in about two minutes. An analyst is handed rates far more often than building them, usually as a single number in a cell with nothing beside it, and the person who built it is usually not in the room.

Check one is arithmetic and cannot be argued with. A weighted average of two numbers lies between them, so an answer above the cost of equity or below the after-tax cost of debt is impossible unless something is wrong. Check one catches a sign error, and it catches weights that do not sum to one. Weights that do not sum to one happen more often than might be expected when somebody has been editing a spreadsheet under time pressure.

Check two is arithmetic anybody can do unaided. The question is what equity figure went into the weights, and the answer is found by multiplying the share count by the price independently. For this company that is 20,00,00,000 shares at Rs 90.00, giving Rs 18,00,00,00,000. If the figure in the model is materially smaller, somebody has used a book number, and the error runs in one direction: book equity is usually well below market equity, so the debt weight comes out too high, the cheap price gets too much weight, and the rate comes out too low.

Check three is where most of the length above has gone. Set the debt's contribution to the blend against what the company actually pays its lenders. The contribution should be lower, and lower by roughly the tax rate. Here the company pays 8.00 per cent and the blend uses 6.00, three quarters of it, and 25.0 per cent is the tax rate. If the debt line equals the contracted rate, the deduction has been missed. If it is far below three quarters of it, the deduction has probably been taken twice.

Who actually keeps a calculator like this, and what do they keep beside it?

Almost nobody rebuilds a cost of capital from scratch each time they need one, and the value of a calculator is not that it computes faster but that it makes the eight fields visible to whoever disagrees with the answer. Three uses are worth describing. Each one uses the tool in a different direction.

A corporate development team keeps the rate in one cell of a shared model with the eight fields sitting above it and each one labelled with where it came from. When the review meeting turns hostile, and it does, the argument becomes a specific one about field 2 or field 4 rather than a general unease about the answer. Specific arguments get resolved and general ones do not. The sensitivity table is what converts the resolution into a number: if the meeting moves the premium by half a point, the rate moves by 47 basis points and everybody can see that before anybody re-runs anything.

An investor comparing two research notes on the same company uses it as a reconciliation device. Two analysts land on 12.00 and 13.20 per cent and the notes explain nothing about why. Put both sets of eight fields side by side, run each difference through the sensitivity table, and almost always one row accounts for nearly the whole gap. Usually it is the beta, occasionally the premium, and very rarely anything else. The row that accounts for the gap shows which of the two notes actually needs reading carefully.

A treasury team runs the whole thing backwards. The board has approved a hurdle of 12.00 per cent for the coming year, so the question is no longer the rate but which combinations of the eight fields are consistent with it. Running the arithmetic backwards is a different exercise on the same lines, and the sensitivity table shows how far each field can travel before the approved rate stops being defensible.

A rate travels well only in company. A number handed over on its own is a number nobody can check and nobody can argue with, and that sounds like a strength when it is not. Travelling with the rate should be the eight fields, the five computed lines, the date, and one sentence naming which of the eight the author is least confident about. The last of those is worth more than the second decimal place.