Debt Capacity: How Much Debt the Cash Flow Can Carry

Debt capacity is the amount of borrowing a company could carry before the terms of that borrowing stop working for it. Debt capacity is a range, not a figure. Sankalp Industrial Systems Limited, invented, holds Rs 6,00,00,00,000 of debt at a blended 8.00 per cent and covers its interest exactly 5.00 times. On its own invented rate schedule, Rs 9,60,00,00,000 would cost 9.00 per cent and take that cover to 2.78 times.

Underneath that answer sits a single awkward fact, and everything in this guide follows from it. The price of borrowing is a function of how much is borrowed, not a constant that can be looked up once and reused. If a company could borrow any amount at one fixed rate, capacity would be a division: the profit, divided by the number of times the interest should be covered, and the answer is out. Capacity would then be a property of the company alone, settled once and never revisited. But a lender does not price a rupee, it prices a position, and it reprices that position whenever the position changes. So how much can be borrowed depends on the rate, and the rate depends on how much is borrowed. The dependence runs in a circle, and the only honest way out of a circle is to lay the whole thing out point by point and read what happens at each one.

What is actually being asked when someone asks how much a company can borrow?

Start somewhere ordinary. A household earning Rs 1,20,000 a month walks into a bank and asks how large a home loan it can take. The bank does not answer with a number. The bank asks what the take-home pay is after tax, what other instalments are already running, how much is left after school fees and groceries, and how long the loan should run. Then it quotes a rate. A bigger loan against the same salary is a different proposition, so the quote is not the one the bank would give for half the amount. The household walks out with a band, not a figure, and the band moves depending on which of those questions the bank leans on hardest.

A company is the same shape of problem with bigger nouns. Capacity is never a property of the company alone; it is a property of the company measured against a chosen test. Change the test and the answer changes, without a single thing about the business having moved. An analyst who hands over a capacity figure without saying which test produced it has handed over less than the reader needs.

There is also a second thing the household example gets right, and it matters more than it looks. The bank in that story is pricing off the borrower's own income and the borrower's own commitments. Nobody in the room is quoting a market rate. Every rate below works the same way: it belongs to one invented company and describes nothing outside it.

So is it one number, or a range?

A range, and the width of that range is the finding rather than an embarrassment. The same ladder of arithmetic, applied to the same invented company on the same day, supports borrowings of Rs 8,40,00,00,000, Rs 9,60,00,00,000, Rs 12,00,00,00,000 and Rs 14,40,00,00,000. Nothing in the business changes between those four. The measure agreed to govern changes, and so does the forecast year the ladder was run on.

A capacity figure is an answer with its question deleted, unless the question travels with it. The failure worked through below is that sentence with a price attached.

What is this company, in the four lines that are actually needed?

Three things about Sankalp Industrial Systems Limited matter here and nothing else does. The company manufactures for industry, so it carries plant and stock. The shares are listed, so the equity side of its mix has a traded price rather than an estimate. And the company already borrows, so the ladder below starts from an occupied rung rather than from nil. Every figure under this paragraph is restated from the single case record this subject is taught on.

A ladder built on a wrong base is a very convincing wrong answer, so two checks are worth doing before anything is built on those lines. Operating profit of Rs 2,40,00,00,000 is EBITDA of Rs 2,88,00,00,000 less Rs 48,00,00,000 of depreciation and amortisation, and against revenue of Rs 12,00,00,00,000 for the same year that is a 20.0 per cent margin. Interest today is Rs 48,00,00,000, and dividing operating profit by that gives cover of exactly 5.00 times. Both agree with the record, and that agreement is the licence to go further.

The tax rate used throughout is 25.0 per cent. The rate is this invented company's own assumed effective tax rateThe share of profit a company actually parts with, taken across everything it pays, rather than any single published rate. and it is not a statement about anybody's tax system. Every after-tax interest figure below is worked with it.

Where does the 8.00 per cent come from, and why does it matter that it is blended?

The company has three borrowings, not one. The shape of the third column is the reason the rate behaves the way it does, so tabling the three is worth the space.

The blended figure is the one the rest of this guide uses, and it is exact rather than rounded. Weighting each rate by the amount it sits on and dividing by the total gives 8.00 per cent to the second decimal. After the company's assumed 25.0 per cent tax rate, the same borrowing costs 6.00 per cent.

What would a lender charge at each level of borrowing?

Here is the engine of the whole method. Attached to this invented company is an invented schedule of rates, one for each level of borrowing expressed as a share of total capital. The schedule is a set of quoted points, and it belongs to this company alone.

Read the first four quotes before reading the shape. Nil borrowing carries a notional 8.00 per cent. A tenth of total capital is quoted 7.75. A fifth is quoted 7.90. A quarter comes back to 8.00. The rate does not rise all the way along; it dips first, and only from the 10 per cent point upward does it climb without interruption. The nil point is notional anyway: with nothing borrowed there is nothing for a rate to sit on, so 8.00 per cent there describes what a first rupee would cost rather than what anything costs. The practical reading is that below roughly a fifth of total capital the schedule says nothing worth acting on, and that everything worth looking at happens above the 40 per cent rung.

One caution that costs nothing to observe and a great deal to miss. The rate applies to the whole balance, not to the increment. Move this company from its present quarter to two fifths of total capital in debt and the 9.00 per cent is charged on the full Rs 9,60,00,00,000, not on the Rs 3,60,00,00,000 of new money with the old Rs 6,00,00,00,000 left at 8.00 per cent. Any arithmetic that reprices only the new tranche will understate the interest bill at every rung, and it will understate it by more the further up the ladder the position sits.

If the rate depends on the amount, how does anyone get out of the circle?

People sometimes expect a formula here, and there is a reason there is not one. A formula would need the rate to be independent of the answer, and it is not. Enumeration replaces the formula: each level the schedule actually quotes is taken in turn, every consequence at that level is worked, and the results are laid next to each other. Reading a table at each point is not a shortcut around solving the circular question. Reading the table is what solving it looks like. The cost of doing it this way is that only the levels the schedule quotes are ever learned about, which is honest, because the schedule really is a set of points rather than a smooth curve. Interpolating between the rows would not be legitimate for exactly that reason.

Test one: how many times does operating profit cover the interest?

Interest cover is the oldest test and the easiest to compute. Operating profit, divided by the interest charge, gives how many times over the profit would pay it. Today that division lands on exactly 5.00 times. Move up the ladder and both halves of that fraction work against the borrower. How a company is financed leaves what it earns alone, so the numerator stays where it is. The denominator grows through the amount and through the rate at once.

The shaded row is the one to check the whole build against. Every figure in the shaded row reproduces the case record exactly, and that agreement is the proof that this ladder was constructed correctly rather than merely convincingly. Every other row is arithmetic worked on the invented schedule rather than a recorded figure.

Now look at what the two forces do together. Between the 25 and the 40 per cent rungs the borrowing rises by 60.00 per cent, from Rs 6,00,00,00,000 to Rs 9,60,00,00,000. The interest bill rises by 80.00 per cent, from Rs 48,00,00,000 to Rs 86,40,00,000. The extra twenty points come entirely from the rate moving on the whole balance. Between 25 and 50 per cent the borrowing doubles and the interest bill multiplies by 2.69, for the same reason with a bigger rate move behind it. The rate moving on the whole balance is why the last rungs bite so much harder than the first ones.

Test two: how many years of EBITDA would repay it, gross and net?

The second test drops the profit ratio and puts a stock against a flow instead: how many years of EBITDA would it take to repay everything outstanding, if every rupee of that flow went to repayment and nothing else moved? The question is crude on purpose, and the crudeness is what makes it comparable. Two versions of it run side by side. Gross leverage takes the whole borrowing. Net leverage takes the borrowing less the cash the company already holds, on the argument that money in the account could be applied to the debt tomorrow morning.

Nothing separates the two lines except the cash balance measured against EBITDA. Rs 1,20,00,00,000 over Rs 2,88,00,00,000 is 0.4167 times, and since neither of those two numbers moves as the mix changes, the distance is identical at every rung. The two lines are parallel by construction. The two lines will never cross and never meet. On this ladder the choice between gross and net leverage shifts the level of the answer and never the shape of it.

One small precision trap is worth meeting on purpose. Left unmet, the trap looks like a mistake. Each column below is rounded to two decimals on its own. Subtracting the printed gross figure from the printed net figure gives 0.41 at the 10, 25 and 40 per cent rungs. Every other rung gives 0.42. Nothing is wrong. The true distance is 0.4167 at all of them, and two independently rounded figures do not have to differ by the rounded form of their true difference. The gap is worked from the underlying amounts, never by subtracting two printed cells.

The nil row prints a negative net leverage of minus 0.42 times, and the minus sign is arithmetic rather than nonsense. With no borrowing at all, the cash balance stands alone in the numerator. Net leverage is a subtraction and can go through zero. Gross leverage cannot.

Test three, and it is the one that gets skipped: is the money actually there?

The first two tests both work off profit lines. Operating profit and EBITDA are real quantities, but neither of them is money sitting in an account at the end of the year. Between them and the bank balance sit the tax bill and every rupee the company had to put back into machinery, stock and receivables to stand still or grow. The third test steps over both of those and asks the blunt version of the question: after tax and after all that reinvestmentThe money put back into plant, equipment and working capital to keep a business growing, before anything is available to those who funded it., is there enough cash left to pay the lender?

For this invented company the answer to that in Year 1 is Rs 98,00,00,000. The Rs 98,00,00,000 is what the business produces after tax and after funding its own growth, before a single rupee goes to anybody who financed it. Set against it goes the interest bill after the tax saved on deducting it. At today's Rs 48,00,00,000 of interest and the company's assumed 25.0 per cent rate, that after-tax charge is Rs 36,00,00,000, so the cover is 2.72 times. Comfortable enough that nobody looks twice.

Where exactly does the cash stop carrying it?

Walk up to the 50 per cent rung and the picture changes completely. Borrowing of Rs 12,00,00,00,000 at 10.75 per cent produces interest of Rs 1,29,00,00,000, and after the tax saved on it the charge is Rs 96,75,00,000. In Year 1 the business makes Rs 98,00,00,000. The cover is 1.01 times: the whole of what the business makes in that year, once tax is paid and the growth is funded, would go to the lender, and Rs 1,25,00,000 would remain.

The middle bar in that drawing, and then the magnified strip below it, carry the whole point. The remainder is drawn honestly and is barely visible, and the near-invisibility is the entire finding. A ratio of 1.01 times reads on paper as though it clears the bar. Drawn to scale it reads as nothing left over.

Go one rung further, to 60 per cent, and the arithmetic stops being close. Interest of Rs 1,87,20,00,000 becomes Rs 1,40,40,00,000 after the tax saved, against the same Rs 98,00,00,000 of cash. The cover is 0.70 times and the shortfall is Rs 42,40,00,000. A company in that position is not paying its interest out of what the business produced; it is finding the difference somewhere else, and the usual somewhere else is more borrowing. That is the only rung on this ladder where the arithmetic itself, rather than a chosen convention, says the position does not fund itself.

Which year of cash flow should the test be run on?

Every figure in the last section came from Year 1. Year 1 was a choice, and a reader cannot see the choice unless somebody states it. The forecast has this business making Rs 98,00,00,000 in Year 1 and Rs 1,70,00,00,000 by Year 5, rising evenly in between. Run the identical 50 per cent rung against Year 5 and the cover reads 1.76 times rather than 1.01. Run the 60 per cent rung against Year 5 and it is 1.21 times rather than 0.70. A position that did not fund itself has become one that does.

A capacity test run on a growing forecast gives a bigger answer the further out it is run, and the growth doing that work is an assumption rather than an observation. The cautious reading tests Year 1, and not out of temperament: the first year has to be survived before any of the later ones happen. A company that cannot carry the interest in Year 1 does not get to reach Year 5 and find out that it could have.

The practical rule that falls out of this is small and easy to keep. Name the year alongside the ratio, always, so a reader can see which of the five produced the figure. A cover of 1.76 times and a cover of 1.01 times can describe one borrowing, at one rate, read on one afternoon.

When the three tests disagree, which one of them wins?

Put the three readings side by side and something uncomfortable happens. Interest cover is still above three times on the 35 per cent rung. Gross leverage has already reached 3.33 times EBITDA at 40 per cent, a long way from the 2.08 times the company sits at today. Year 1 cash flow cover has fallen to 1.01 times by 50 per cent and has nothing left after that. Three measures, one business, one schedule, one afternoon, and three different stopping places.

There is no arithmetic that reconciles those three, and looking for one is a mistake about what they are. The three measures disagree because they are not measuring the same quantity. One is a ratio of two profit lines, one puts a stock against a flow, and one asks whether the money is physically there after the business has been fed. A firm that fails the third and passes the first has said something specific and useful, which is that its profit looks strong and its cash does not, and that is a fact about reinvestment rather than about leverage.

The temptation this creates is obvious and it is the subject of the failure block below. When three tests give three answers, it is very easy to report the friendliest, especially if somebody in the room wants the number to be large. The friendliest reading is exactly the one that hides the most useful thing the three tests found.

Is the point where value peaks the same as the point a lender would fund?

There is a separate question running alongside everything above. How a level of borrowing feeds through the blended cost of capital into what the whole firm is modelled to be worth is worked out elsewhere in this subject. The direction of the two forces can be stated: the deduction on interest pulls the blended rate down as borrowing rises. Two things push the other way at once. Lenders reprice, and the schedule above has already shown it. A thinner residual claim is a riskier one to hold, so shareholders also want more. The blended rate therefore turns somewhere in between. For this invented company it turns where debt is 40.0 per cent of total capital, and the arithmetic behind that is worked out separately.

Two warnings are welded to that figure, and neither can be detached from it.

So no level named above is a general answer for anybody. The turn is one spot on one invented set of quotes attached to one invented business, and reading it any wider misreads a worked example.

The ladder above can be read at that same rung.

Value and fundability are two different questions with two different answers. The first asks where a model says the modelled value of the firm is highest. The second asks what position a company and its lenders would actually carry through a difficult year. Both answers stand and neither settles the other. Whether any given debt share is right, safe, suitable or appropriate for anybody is a separate judgement.

Why does the case for more borrowing run out before the tests do?

One more calculation explains the shape of everything above without borrowing a single cell from the value table. The calculation uses only the invented rate schedule and the company's own assumed 25.0 per cent tax rate, and it compares two quantities that both grow as borrowing grows.

The first is the deduction the borrowing earns. At an unchanged 8.00 per cent, that is the borrowing multiplied by 0.08 and again by 0.25, or exactly 2.00 per cent of the amount borrowed. The deduction rises in a straight line: double the borrowing and the deduction doubles with it. The second quantity is whatever the schedule adds on top of 8.00 per cent once its own deduction is taken off, or 0.75 multiplied by the borrowing and by however far the rate has moved past 8.00. Both of its factors grow together, so the extra interest accelerates.

The two quantities cross somewhere between the 45 and the 50 per cent rungs. The schedule quotes a set of points rather than a curve, so it allows no more precision than that. A crossing point interpolated between two rungs would carry a precision the quotes cannot supply.

The shape matters more than the crossing does. Read downward, the last column shows the gap rising from the 25 to the 30 per cent rung, almost flat from 30 to 35, and only then falling away. So the benefit of the next rupee of borrowing peaks well before the two quantities cross. The case for more debt weakens long before any of the three tests says stop. Every figure in that table is arithmetic on the invented schedule and on an assumed tax rate, and none of it says anything about any tax system.

Notice what produced that spread. Not a disagreement about the business, not a different forecast, not an argument about the rate schedule. Four people looking at one table, each choosing a different test, in perfect good faith. The spread of Rs 6,00,00,00,000 is manufactured entirely by the choice of measure and the choice of forecast year. Both therefore belong in the sentence that carries the figure.

What can none of these tests see?

Every measure in this guide works on a small number of lines: operating profit, EBITDA, cash, and one year of free cash flow. The short list is what makes the measures comparable across companies, and it is also what makes them partial. Five things sit outside the whole apparatus.

The cost structure above the operating profit line is the first. Whether this company's costs are mostly fixed or mostly variable decides how far operating profit falls when revenue falls, and nothing in any of these ratios reveals it. Two firms with identical interest cover of 5.00 times can behave completely differently in a bad year for that reason alone. Cost structure is covered separately.

The maturity profile is the second. Every test here asks how much is owed; none asks when. A borrowing of Rs 6,00,00,00,000 spread evenly across ten years and the same amount falling due in a single instalment produce identical leverage ratios and completely different lives. Repaying or rolling a large instalment when it arrives is covered separately.

Any covenantA written undertaking inside a loan document. It fixes a level the borrower has to stay the right side of, and hands the lender a remedy the moment that stops being true. is the third, and here the honest statement is simply that this record discloses none, so none is modelled in the ladder or in the panel above. In practice a written limit can stop a company well before any of these three tests would.

The seasonality of the operating cycleHow long money stays tied up in a business: out of the account when stock is bought, back in only once the customer has settled. is the fourth. Annual figures average away a year in which the cash balance dips for four months, and interest does not wait for the good months.

The fifth is the largest and it is an assumption of the illustration rather than a gap in the tests. Every rung of this ladder holds total capital at Rs 24,00,00,00,000 and the share price at Rs 90.00 while the mix moves, a device for isolating one variable rather than a description of what would happen if a company actually did this. Real movement along a ladder like this one changes the equity value as it goes, and how a company would make that move is covered separately in this sequence.