Leverage Ratios in Private Credit: Sizing the Loan

A leverage ratio in a loan document is net debt divided by earnings before interest, tax, depreciation and amortisation. One level sizes the loan at drawdown. A second, higher level is written in as a maintenance covenant and tested every quarter. Crossing that second level is not a verdict on the borrower. It is a contractual trigger for a conversation. Both levels here belong to one invented loan agreement.

The arrangement appears in a form most readers have already met. A household walks into a bank and asks for a home loan. Nobody at the bank begins from the number the household would like to have. The bank begins from what the household earns, applies a multiple to it, and the loan is whatever that arithmetic produces. Then, later, on a schedule printed in papers nobody read, the bank asks about that income again. Two measurements, years apart, built from the same two quantities, doing two completely different jobs: the first decides how much money moves, and the second decides whether anybody has to have a conversation. A leverage ratio in a loan agreement is that same arrangement, written down precisely, for a company instead of a household, and the difference between those two measurements is the subject of this guide.

Everything below runs on one loan. Position 1 of Nilgiri Direct Lending Fund I, invented, is a senior secured term loan of Rs 50,00,00,000 carrying a cash coupon of 13.5 per cent, with net debt to earnings of 3.0 times measured at drawdown and a maintenance covenant at 3.75 times tested each quarter. Every figure in this guide belongs to that invented fund and to that one invented loan agreement. Leverage levels and covenant levels are negotiated one agreement at a time, so no single pair of them describes what is ordinary anywhere in India.

What are the two parts of a leverage ratio, and why must both be named every time?

A leverage ratioNet debt divided by earnings before interest, tax, depreciation and amortisation. is one number sitting on top of another number, and almost every argument anybody ever has about a leverage ratio turns out to be an argument about how one of those two numbers was built. So take them one at a time, and refuse to move on until both have been named.

On top sits net debtEverything owed to lenders, less cash in hand.. Net debt is everything the business has borrowed less the cash it is holding. For this borrower, at the moment the loan was advanced, that figure is Rs 60,00,00,000. Net debt is not the same thing as the loan, and the difference is worth pulling apart. The fund advanced Rs 50,00,00,000. The business already had Rs 10,00,00,000 of borrowing before this fund arrived at all, and Rs 50,00,00,000 plus Rs 10,00,00,000 is the Rs 60,00,00,000 the ratio is measured on. The top of the ratio is everything the business has borrowed, not everything this one lender lent, and a reader who quietly substitutes the loan for the net debt has already got the ratio wrong by a sixth. The loan record does not say who that pre-existing Rs 10,00,00,000 is owed to.

Underneath sits earnings before interest, tax, depreciation and amortisation. For this borrower that figure is Rs 20,00,00,000. How such a figure is assembled from a set of accounts is covered separately, and the figure is taken as given here. Only three things matter here: the figure is a measure of earnings, somebody produced it, and a different somebody could produce a different one from the same accounts. The possibility of that second figure becomes the sharpest point below.

Divide one by the other and Rs 60,00,00,000 over Rs 20,00,00,000 is exactly 3.0 times. Readers lose the thread here, so say the bottom of the ratio out loud, every single time. Three times means net debt over earnings. Three times does not mean 3.0 times the loan, it does not mean 3.0 times anything the fund committed, and it is not a share of the Rs 3,00,00,00,000 this invented fund has in commitments. A multiple quoted without both of its sides named is not yet a fact, and in a document full of multiples that one habit is the whole difference between reading and guessing.

How does a ratio size a loan on the day it is drawn?

Run the arithmetic forwards instead of backwards and the sizing job becomes obvious. Start with earnings of Rs 20,00,00,000. Apply the level the parties agreed the loan would be drawn at, here 3.0 times. Three times Rs 20,00,00,000 is Rs 60,00,00,000, and that is how much total net debt this business is going to carry once the transaction closes. Now subtract what is already there. The business owed Rs 10,00,00,000 before anybody arrived, so the amount this lender can put in is Rs 60,00,00,000 less Rs 10,00,00,000, being Rs 50,00,00,000. A multiple, a measure of earnings and a subtraction for what the business already owed are the whole sizing mechanism, and the loan is whatever comes out the other end.

The mechanism has a consequence for the borrower who wanted more. The amount is not negotiated as an amount. The loan is negotiated as a multiple, and then the multiple meets the earnings figure and produces the amount. A borrower arguing for a bigger loan is really arguing about one of two things: a higher multiple, or a bigger earnings figure. Both arguments are live in real documents, and the second one is taken up under adjusted earnings below.

Here is a counterfactual on the same loan, and it is labelled a counterfactual because it did not happen. Suppose the same borrower had come to the same lender with earnings of Rs 18,00,00,000 instead of Rs 20,00,00,000, and suppose both sides still settled on 3.0 times. Three times Rs 18,00,00,000 is Rs 54,00,00,000 of total net debt, less the same Rs 10,00,00,000 already there, and the advance would have been Rs 44,00,00,000 rather than Rs 50,00,00,000. The multiple magnifies the earnings line three times over, so Rs 2,00,00,000 of difference there moved the cheque by Rs 6,00,00,000. Nothing in that sentence says anything about whether the business is a good one.

The drawdownThe moment the loan is advanced and the sizing level is measured. measurement happens once, on the day the money moves, and then it is history. The drawdown reading sits in the file for the rest of the loan's life as a description of how the transaction was put together. The reading never gets tested again, it never trips anything, and nobody is in breach of it. Which is exactly why the maintenance level exists.

What is a maintenance level, and how is it a different number doing a different job?

A maintenance covenantA level the borrower must stay inside, tested on a fixed timetable. is a level the borrower promises to stay inside, checked on a timetable that the agreement fixes in advance. On position 1 that level is 3.75 times and the timetable is every quarter. So the borrower delivers its numbers after each quarter end, the two sides recompute net debt over earnings, and the answer either sits at or below 3.75 times or it does not.

The other kind of level is worth naming, to tell the two apart when one of them appears. An incurrence testA level tested only when the borrower wants to do something, such as borrow more. is only measured when the borrower wants to do something, most often borrow more money or pay something out. Nobody checks it on a date. An incurrence test lies dormant until the borrower asks for something, and then it decides whether the answer is yes. A maintenance level is the opposite in exactly one respect: it does not care whether the borrower wants anything. The test dateThe scheduled date on which a maintenance covenant is measured. arrives on its own, four times a year, and the measurement happens whether or not anybody involved is paying attention.

The two levels of position 1 sit side by side, and this is the pairing everything else depends on. The loan was drawn at 3.0 times. The maintenance level is 3.75 times. The distance between them is 0.75 of a turn, and that distance is not decoration: it is the room the document leaves before the borrower and the lender are obliged to speak to each other. A document that set the maintenance level at exactly the drawdown level would put the borrower on the line from day one, and a document that set it far away would rarely be tested at all. Where in between it lands is a negotiated outcome between two parties, and there is no place at which it ought to land.

A run of ratios invites the opposite reading. Neither 3.0 times nor 3.75 times is usual, required, prudent or appropriate. The two levels are the numbers written into one invented agreement between this fund and one borrower. A ratio is a measurement, and a measurement does not say what a good one is.

How far can earnings fall before the test is failed?

How far earnings can fall is the question the maintenance level actually answers, and a division answers it. Nothing about the borrowing is changing for the moment, so hold net debt still at Rs 60,00,00,000. Then ask what level of earnings puts the ratio exactly on 3.75 times. Rs 60,00,00,000 divided by 3.75 is Rs 16,00,00,000. At earnings of exactly Rs 16,00,00,000 the borrower sits precisely on the line. Anything above that and the test is met. Anything below and it is not.

So the room the covenant leaves, on this side of the ratio, is Rs 20,00,00,000 less Rs 16,00,00,000, being Rs 4,00,00,000. Rs 4,00,00,000 is the headroomHow far a measure can move before a covenant level is crossed., and headroom is a distance rather than a state. As a percentage it is Rs 4,00,00,000 over the Rs 20,00,00,000 that earnings started at, being 20.0 per cent. Name that denominator or the figure is worthless: 20.0 per cent is of earnings, not of the loan, not of net debt and not of anything the fund committed. A reader who divides the same Rs 4,00,00,000 by the Rs 60,00,00,000 of net debt gets 6.7 per cent, a true arithmetic statement about two numbers and an answer to a question nobody asked.

Now the part that a single worked figure cannot teach, and it is what the control below is for. Between earnings of Rs 20,00,00,000 and earnings of Rs 16,00,00,001, absolutely nothing happens. The ratio drifts from 3.00 times through 3.20, 3.50, 3.70, 3.74, and at every one of those readings the borrower is inside its covenant and the quarterly test is met and nobody has to do anything at all. Then earnings slip one more rupee and the reading crosses 3.75, and now something has happened. A covenant is a cliff and not a slope, which is why a borrower can look completely unchanged for six quarters and then be in breach in the seventh without anything dramatic occurring in between.

How much more could the borrower owe before the same test is failed?

Same covenant, same borrower, same quarter. Now hold earnings still at Rs 20,00,00,000 and ask the question from the other end. What level of net debt puts the ratio exactly on 3.75 times? Multiply rather than divide: 3.75 times Rs 20,00,00,000 is Rs 75,00,00,000. So net debt could rise from Rs 60,00,00,000 to Rs 75,00,00,000 before the test is failed, an extra Rs 15,00,00,000 of borrowing.

Stop and look at what just happened. Nothing in this subject is misread more often. The room in that covenant has not changed at all: one covenant, one borrower, one quarter, one 3.75 times. And yet expressed on the earnings side it is Rs 4,00,00,000 and expressed on the net debt side it is Rs 15,00,00,000. Both figures describe the identical amount of room, and they sound nothing alike because they are measured on different quantities. As percentages they diverge again: Rs 4,00,00,000 is 20.0 per cent of the Rs 20,00,00,000 of earnings, while Rs 15,00,00,000 is 25.0 per cent of the Rs 60,00,00,000 of net debt. Four numbers, one covenant, and every one of them needs the quantity it is measured against attached to it or it means nothing.

Think about which person in the room quotes which. Somebody watching how the business is trading month to month is watching earnings, and to them the covenant is a Rs 4,00,00,000 question. Somebody watching the obligations the business is signing up to is watching net debt, and to them the same covenant is a Rs 15,00,00,000 question. Both of them are right, both are describing the same clause, and if they ever compared notes without naming their quantities they would each think the other had the wrong document.

What changes when the test runs on adjusted earnings instead of reported ones?

Every calculation so far has divided by Rs 20,00,00,000, which is what this borrower reported. Real loan agreements very often do not divide by the reported figure. Such documents divide by that figure with certain amounts added back to it, and the definition of which amounts is written into the agreement, sometimes at considerable length. An add-backAn amount added to reported earnings to produce an adjusted figure. is one of those amounts, and the result of applying them is called adjusted earningsReported earnings with add-backs applied, used in place of the reported figure..

Suppose this borrower proposes Rs 2,00,00,000 of add-backs. Adjusted earnings are then Rs 22,00,00,000 against reported earnings of Rs 20,00,00,000. Redoing every piece of arithmetic with the new bottom of the ratio shows what moves.

Take the two lines that matter. Rs 60,00,00,000 over Rs 22,00,00,000 is 2.73 times against 3.00 times reported, so the same borrower with the same borrowing on the same day now reads two thirds of a turn lower. And the level at which the 3.75 times test would be failed moves too, because it is 3.75 times whatever the bottom of the ratio is: 3.75 times Rs 22,00,00,000 is Rs 82,50,00,000, against 3.75 times Rs 20,00,00,000 being Rs 75,00,00,000. Rs 2,00,00,000 of add-backs bought Rs 7,50,00,000 of extra borrowing room, and not one rupee of extra earnings arrived in the business. The number moved and the business did not.

The last sentence of that table does the decisive work. The multiple in the covenant did not change. The borrowing did not change. Nothing was paid, received, sold or built. The only thing that moved was the definition of the quantity underneath, and Rs 7,50,00,000 of borrowing capacity appeared. Whether any particular add-back is proper is covered separately. The arithmetic remains, and the arithmetic is that the bottom of the ratio is a definition rather than a fact of nature.

One more line falls out of the same substitution and it is worth having. If the test now runs on adjusted earnings, then adjusted earnings can fall to Rs 16,00,00,000 before 3.75 times is reached, because Rs 60,00,00,000 divided by 3.75 is still Rs 16,00,00,000. From Rs 22,00,00,000 that is a fall of Rs 6,00,00,000, being 27.3 per cent of the Rs 22,00,00,000 of adjusted earnings, against the 20.0 per cent of the Rs 20,00,00,000 of reported earnings computed earlier. Same borrower, same covenant, same quarter, two headroom percentages that differ by a third, and the only difference is which earnings figure is in the denominator.

What does a borrower already at 4.4 times have to do to get back inside?

Position 1 aside, one borrower in this invented loan book actually crossed a line. Position 3 is mezzanine debt of Rs 30,00,00,000, and at the Year 2 Q3 test its leverage read 4.4 times against a 3.75 times maintenance level. Position 3's borrower did not miss a payment, and the fund and the borrower reset the terms before anything else happened. The content of that reset, and how a loan gets restructured before a default, is covered separately. Only the number matters here, as arithmetic.

Ask the question every reader asks next: how far back does it have to come? There are exactly two routes, one down each side of the ratio, and neither of them needs a single fact about the business.

Route one, on unchanged net debt. The reading is 4.4 times and the level is 3.75 times. Divide one by the other: 4.4 over 3.75 is 1.173, so the borrower is sitting 17.3 per cent above its level, and earnings would have to rise by 17.3 per cent to bring the reading back to exactly 3.75 times. Route two, on unchanged earnings. Divide the other way: 3.75 over 4.4 is 0.852, so net debt would have to be 85.2 per cent of what it is, a fall of 14.8 per cent. The same distance back to the same line is a 17.3 per cent rise in one quantity or a 14.8 per cent fall in the other, and the two percentages differ because they are percentages of different things.

The difference is not a rounding curiosity. Whenever a ratio is pushed back to a target by moving the top, the percentage required is bigger than the percentage that would be required by moving the bottom, because the two are reciprocals of each other rather than the same number with a sign flipped. Anybody who hears that a borrower is 17.3 per cent above its level and concludes that cutting borrowing by 17.3 per cent would fix it has overshot. A cut of 14.8 per cent is what it takes. Neither figure says anything whatever about which route is available to that borrower, or whether either one happens.

What actually happens at the moment a test is failed?

To the money, nothing. The answer deserves to be short. Everything else on this subject gets tangled without it. A maintenance test being failed means a ratio computed from two figures came out above a number written in an agreement, on a date that agreement fixed. The coupon may have arrived on time that very quarter. The business may be trading exactly as it was three months earlier. Nobody has taken anything and nothing has been sold. A covenant breach is a promise about a ratio being broken, and a payment default is money not arriving, and they are two different events that can happen entirely independently of each other.

The invented loan book contains one of each. Position 3's borrower breached a leverage covenant and kept paying. Position 5's borrower missed a coupon and did not cure it. Neither event caused the other, neither is a version of the other, and a reader who collapses the two has lost the ability to read either one. The steps that follow a breach, and what the two parties then negotiate, are covered separately.

The failed test puts an obligation on the calendar. Somebody now has to tell somebody else, and the two sides now have to speak, and the terms of that conversation are whatever the agreement says they are. Forcing that exchange is the entire function of the number. The covenant is a device for making a conversation compulsory on a date that neither side chooses in the moment.

How does somebody actually use this on a Monday morning?

The arithmetic above only becomes useful when it is attached to a job, so watch three people use these two numbers in three different ways.

The first is whoever monitors the loan for the lender. After each quarter end a certificate arrives from the borrower stating the ratio and asserting that the covenant is met. The work is not reading that assertion. The work is recomputing it: pulling the two quantities out separately, checking that the net debt figure includes the borrowing that was already there and not only this lender's own advance, and checking which definition of earnings the certificate used. The whole discipline is refusing to accept a ratio as an input. A ratio is an output, and the interesting information is always in the two numbers that produced it.

The second is whoever is reading a report rather than a loan file, including anybody looking at how an invested position is described. A position stated as 3.0 times at drawdown is a fact about how a transaction was sized on one day that has passed. A drawdown figure is not the current reading and not a claim about now. Asking when a stated ratio was measured is the same instinct as asking when a photograph was taken.

The third is closer to home, in a version anybody can feel. A household with a home loan is inside exactly this arrangement. Income decided the size of the loan on the day it was taken, and something in the papers gives the bank the right to ask about that income again. The household that keeps a little distance between its income and the level at which questions start has bought itself room, and the household that has none is one bad quarter from a conversation it did not schedule. Headroom is not a state of health, it is a distance, and the only way to know a distance is to measure both ends of it.

What does a leverage ratio not tell anybody?

Whether the borrower can pay. The closing claim deserves to be stated flatly rather than implied. A leverage ratio is a measurement of two quantities in a fixed relationship, put into a document for two purposes: to decide how much money moves at the start, and to make a conversation compulsory on a date if the relationship moves past an agreed point. A leverage ratio is not an assessment, not a score and not a view. Judging whether a business can service what it has borrowed is a discipline with its own methods, and it is covered separately.

Everything in this guide reinforces that. The same covenant produced Rs 4,00,00,000 and Rs 15,00,00,000 of room depending on which of the two quantities was moved. A definition change worth Rs 2,00,00,000 produced Rs 7,50,00,000 of borrowing capacity without a rupee arriving. A reading of 4.4 times needed either a 17.3 per cent rise or a 14.8 per cent fall to come back to the same line. Every one of those is a fact about arithmetic and a document, and not one of them is a fact about a business. None of that is a limitation of the measure. Measuring two quantities against each other is all the measure ever was.