Debt Service, the Cover Ratio and the Cash Waterfall

Debt service is interest plus the scheduled principal instalment, and it is what the project's cash must cover. Tapti Crossing Infrastructure Private Limited owes Rs 119.70 crore of interest and Rs 63 crore of principal, being Rs 182.70 crore, against earnings before interest, tax, depreciation and amortisation (EBITDA) of Rs 248 crore, a cover of 1.36 times in the modelled year. The cash pays that in a fixed order, and the sponsors are last.

Start with the difference between an ordinary company and a project company. An ordinary company that has a thin year can lean on something. The ordinary company has other products, other customers, a bank balance built up in better years, a parent that might help, or an asset it could sell without stopping the business. A project company has none of that. Tapti Crossing Infrastructure Private Limited, an invented single-asset toll road company, was formed to build and operate one crossing and to do nothing else. There is one stream of cash coming in, and there are obligations sitting on top of it, and if the cash is short there is no second place to look.

A single stream of cash makes the order matter more than the total. When there is only one source of cash, the interesting question stops being how much there is and becomes who gets it first. Every question about a project financing turns out to be a question about sequence: what is paid before what, what is set aside before anything is released, and who is standing at the end of the queue.

The household version of this is familiar. A household running on one salary pays the rent, then the loan instalment, then the school fees, then the electricity, and whatever survives all of that goes into savings. In a good month everything gets paid and the order is invisible, so nobody in that household could name it. In the month the salary lands late, the order becomes the only thing that matters, and the household discovers what it actually decided years ago without ever writing it down. A project financing writes it down.

Everything that follows uses one worked year of Tapti Crossing Infrastructure Private Limited. Project cost Rs 1,800 crore, funded Rs 1,260 crore of debt and Rs 540 crore of equity, a 70 to 30 structure. Annual revenue Rs 310 crore, operating cost Rs 62 crore, so EBITDA is Rs 248 crore at an 80.0 per cent margin. An 80.0 per cent margin is ordinary for a road and would be extraordinary in almost any other business. The interest rate of 9.5 per cent is this project's own contracted rate, invented for teaching, and it is never a statement about what infrastructure debt costs in India or anywhere else.

What exactly is debt service, and what does a profit statement leave out?

Debt serviceThe total cash a borrower has to hand over to its lenders in a period: the interest that has accrued plus any principal the schedule says must be repaid in that same period. is two things added together, and the trouble with it is that only one of the two is visible in the place most readers look first.

Build it rather than accept it. Interest is a rate applied to a balance. Tapti Crossing Infrastructure Private Limited has Rs 1,260 crore of debt outstanding at the start of the modelled year, and at the project's own contracted rate of 9.5 per cent, the interest for that year is Rs 119.70 crore. Nothing clever happened there. The interest is one multiplication, and it checks in a second: 9.5 per cent of Rs 1,260 crore.

The second component is the scheduled principalThe slice of the borrowed amount that the repayment schedule requires be handed back in a given period. The instalment reduces what is owed rather than paying for the use of it. instalment. For the modelled year the instalment is Rs 63 crore, or 5.0 per cent of the Rs 1,260 crore drawn. Add the two and debt service is Rs 182.70 crore.

The two components do different work, and this is where careless reading begins. Interest is the price of using money that belongs to somebody else, and it is an expense. Interest appears in the profit statement as finance cost, it reduces profit before tax, and any measure built on profit has already taken account of it. The scheduled instalment does no such thing. The instalment is not a cost of anything. The instalment returns the borrowed money itself, so it reduces a balance on the balance sheet and appears in no cost line anywhere on the profit statement. The project is no poorer for paying it, in the accounting sense. The company is simply Rs 63 crore lighter in cash.

Rs 63 crore is 34.5 per cent of the Rs 182.70 crore obligation, so a project judged on profit measures alone is being judged with more than a third of its cash obligation left out of the picture. That is not a rounding issue or a technicality. The missing instalment is the single reason a project company can look perfectly comfortable on an interest measure and be uncomfortably tight on cash in the same year, and it is why the lenders on a project financing almost never look at interest alone.

The household version again. A household with a loan writes one cheque a month, and inside that cheque there is an interest part and a repayment part. The bank does not ask for them separately and the money leaves the account either way, so nobody in the household separates them. The mistake is only ever made by somebody looking at the household from outside, reading a statement of its expenses, and forgetting that a large part of that monthly cheque was never an expense at all.

How is the cover ratio built, and what must be written beside it?

The cover ratioCash available in a period divided by the debt service due in that same period. Above one, the period pays its own way. Below one, something else has to fill the gap. is the most quoted number in project finance and one of the most casually quoted. The cover ratio is one division: the cash available in a period over the debt service due in that period.

Run it. EBITDA for the modelled year is Rs 248 crore. Debt service for the modelled year is Rs 182.70 crore. Rs 248 crore divided by Rs 182.70 crore is 1.3574165, written as 1.36 times. A ratio that has not been built is a ratio whose numerator and denominator have not been checked, so a cover ratio is computed rather than quoted. Almost every argument about a cover ratio turns out to be an argument about what went into the numerator or the denominator.

Said plainly, what 1.36 times does mean is this. In this one modelled year, the project generated 36 per cent more cash than it needed for that year's interest and instalment. Turned around, the same fact reads from the other side: debt service consumes 73.7 per cent of EBITDA, and 26.3 per cent of it, being Rs 65.30 crore, survives the two payments.

The stretching starts here, so now say what 1.36 times does not mean. The ratio says nothing at all about any other year. A cover ratio is not a property of the project, or of the road, or of the structure. A cover ratio is a property of one year, computed from that year's cash and that year's obligations, and the moment either of those changes the ratio changes with it. A cover ratio without a period attached to it is a number that cannot be checked by anybody.

There is a second thing that has to travel with the figure here, and it is the honest part. The record carries no tax charge and no maintenance spending for the project company. A real toll road pays tax on its profits and spends money keeping the surface, the barriers and the collection equipment usable, and both of those come out of cash before the lenders can be paid from it. Neither is in this record, so Rs 248 crore is the most that could ever be available in this year and 1.36 times is the highest this ratio can go. The 1.36 times is a ceilingThe upper limit a figure could reach given what is being counted. Deductions that exist in the real world but are missing from the record can only push the figure down from here, never up., not a central estimate, and it should be written as one.

So the honest sentence has three parts in it, not one. The number, 1.36 times. The base, EBITDA over debt service. The period, the modelled year. And on this record, one qualification: before tax and before maintenance spending, neither of which is carried here. Anybody who reads that sentence can rebuild the arithmetic. Anybody who reads only the number cannot.

What is a cash waterfall, and why does the order matter more than the amounts?

A cash waterfallThe fixed order in which a project company's cash is applied. Each level is filled to its requirement before any cash reaches the level below it. is the agreed order in which the project company's cash is applied. A waterfall is drawn as a stack for a reason: cash enters at the top, fills the first level to its requirement, and only what is left over spills down to the next. Nothing at a lower level receives anything until everything above it is full.

For Tapti Crossing Infrastructure Private Limited the order runs like this. Operating cost comes first. The road has to be kept running or there is no cash at all. Then the lenders' interest. Then the scheduled instalment. Then any amount owed to the reserve. Then, and only then, whatever is left goes to the sponsors. Orders differ between agreements, so the order above belongs to this structure alone, recorded in its own financing arrangements, and stands as an illustration rather than as a description of what every project agreement contains.

The order does its work by being invisible in a good year and decisive in a bad one. With Rs 248 crore of EBITDA at the top of this stack, every level fills: interest gets its Rs 119.70 crore, the instalment gets its Rs 63 crore, and Rs 65.30 crore reaches the sponsors. Everybody was paid, so the outcome read on its own suggests the order is a formality. With Rs 200 crore instead, the picture changes without any level changing its rule: interest is still Rs 119.70 crore in full, the instalment is still Rs 63 crore in full, and the sponsors receive Rs 17.30 crore instead of Rs 65.30 crore. The sponsors stand last, so the whole shortfall of Rs 48 crore landed on them.

One property is worth carrying away. A shortfall in a waterfall is not shared. A shortfall is allocated, entirely, to the lowest level that the cash fails to reach, and then to the next one up if the shortfall is bigger than that level. Standing last does not mean receiving a bit less when things are tight. Standing last means receiving nothing until everybody above is whole.

Each level is called a tierOne level of the waterfall, with a stated requirement and a stated position in the order. A tier is filled to its requirement or filled as far as the cash goes, never partly skipped in favour of the next one., and the tiers are not a list of things that get paid. The tiers are a ranking. Flattening a waterfall into a list of deductions throws away the only information it carried.

Come back to the household. The household has rent, an instalment, school fees and savings, and in the ordinary month all four happen. Ask that household which of the four it would drop if the salary arrived at half its usual size, and the answer is instant and unhesitating. The ranking was settled long ago and was never written on anything. A project financing simply writes down the answer in advance. Nobody then has to work out the ranking in the month it stops being hypothetical.

Where does the reserve sit in the order, and who paid for it?

A debt service reserveA cash balance the project company must hold, sized as a stated number of periods of debt service, kept in an account the lenders control and available only to pay them. is a pot of cash the project company must keep on hand, sized as a number of periods of debt service. For Tapti Crossing Infrastructure Private Limited the requirement is two quarters of debt service. Annual debt service is Rs 182.70 crore, so one quarter is Rs 45.675 crore and two quarters is Rs 91.35 crore, exactly half of one year's obligation.

Ask what that pot is actually for. Calling it a safety cushion and stopping there is easy. A reserve converts a timing problem into a funded buffer, and it is cash the sponsors cannot take out. Those are two separate statements and both matter.

Take the timing part first. Toll revenue does not arrive smoothly. Traffic falls in a bad monsoon, a diversion closes a feeder road for eight weeks, a festival month runs hot and the next one runs cold. Debt service, meanwhile, falls due on fixed dates and does not care about any of that. Without a reserve, a project that will comfortably cover the year can still miss a payment date inside the year, and a missed payment date is a different kind of event from a thin year. The reserve stands between those two things.

The second part is the one that changes behaviour. The money in that account is not the project's spare cash. The reserve sits in an account the lenders control, it can be applied only to paying them, and it is outside the reach of a distribution to the sponsors. From the sponsors' side, a reserve of Rs 91.35 crore is Rs 91.35 crore of their money parked where they cannot use it for the life of the borrowing. The parked money is a real cost to them and it is the price the structure charges for the comfort.

Now the honest gap in the record, and it is not a small one. The record says the reserve is two quarters of debt service. The record does not say whether that Rs 91.35 crore was funded at the start, out of the Rs 1,800 crore raised for the project, or whether it is being built up out of operating cash in the years the road is running. The two readings put the cost on completely different parties, and the arithmetic of the modelled year comes out completely differently under each.

Work both readings through. The contrast is the lesson. If the reserve was funded from the original drawdown, it is already sitting full in the lenders' account and its top-up requirement in the modelled year is nil. Tier three takes nothing, and the whole Rs 65.30 crore reaches the sponsors. On the Rs 540 crore of equity they put in, that is a 12.09 per cent cash return in this year. Note carefully that the 12.09 per cent covers a single year and is not a return over the life of anything. The record carries no concession period and no schedule beyond the year modelled.

If instead the reserve is being funded out of operating cash, tier three is asking for Rs 91.35 crore and there is Rs 65.30 crore standing in front of it. The reserve requirement is 1.40 times the entire residual. Every rupee of the residual goes into the reserve, the reserve is still Rs 26.05 crore short at the end of the year, and the sponsors receive nothing at all. The same project, the same Rs 248 crore of EBITDA and the same 1.36 times cover produce either Rs 65.30 crore or exactly nil for the sponsors, and the only thing that changed is where the reserve was funded from.

When the record does not settle something that swings the answer this far, the discipline is to show both readings and label the gap, not to pick the convenient one and carry on. Every calculation below holds the first reading, that the reserve was funded at the start, and says so at each point where it matters.

What does one whole year look like when it is run down the order?

The modelled year of Tapti Crossing Infrastructure Private Limited runs below from the top of the waterfall to the bottom, with every figure shown so that each step can be checked against the one above it. The table reads as a sequence of stops rather than as a column of subtractions, and the difference between those two readings is where the costly error comes from.

Three readings come out of that table and each is worth stating on its own. The first is the cover: Rs 248 crore of EBITDA over Rs 182.70 crore of debt service is 1.36 times for the modelled year, before tax and before maintenance spending, neither of which this record carries. The second is the residual: Rs 65.30 crore for the sponsors on Rs 540 crore of equity, a 12.09 per cent cash return in this year alone, and nothing more than that. The third is the alternative reading of tier three, under which that Rs 65.30 crore goes into the reserve instead, the reserve remains Rs 26.05 crore short, and the sponsors receive nothing.

Notice that the table has a column for cash still in hand, and that this column, not the amounts column, is what makes it a waterfall. At every step, payment can only come out of what the previous step left. With the amounts column covered, the running balance still tells the whole story. With the running balance covered, the amounts column becomes an ordinary list of costs, exactly the mistake this structure exists to prevent.

What is Debt Outstanding, and what does the balance decide?

Debt outstandingThe amount still owed to the lenders at a point in time. The balance falls as scheduled principal is repaid, and it is the base on which the next period of interest is calculated. is simply the amount still owed at a point in time, and it moves for one reason in this guide: the scheduled instalment reduces it.

Tapti Crossing Infrastructure Private Limited starts the modelled year owing Rs 1,260 crore. The company pays the Rs 63 crore instalment. The company ends the year owing Rs 1,197 crore, 95.0 per cent of where it started. The instalment is the whole movement, and no other line in the waterfall touches the balance. Interest is the price of the balance rather than a repayment of it, so interest does not reduce the balance. The reserve is cash held to one side and not cash handed over, so the reserve does not reduce the balance either.

Now the part that matters. Interest is a rate applied to a balance and nothing more complicated than that, so the balance sets the next period's interest. At the same contracted 9.5 per cent, the following year's interest on Rs 1,197 crore would be Rs 113.72 crore, or Rs 5.99 crore less than this year. The fall of 5.0 per cent in the interest bill is exactly the 5.0 per cent fall in the balance, and it will happen for as long as the balance keeps falling and the rate does not move.

The balance does not determine debt service. Debt service also needs the following year's scheduled instalment, and that instalment is not in this record at all. The instalment is set by a repayment schedule agreed between the project company and the project lenders, and a schedule is a document, not an arithmetic consequence of the balance. The schedule might repay Rs 63 crore again. The schedule might repay more, rising as traffic is expected to build. The schedule might repay less in early years and a great deal at the end. Every one of those is a normal shape and this record does not say which one applies.

Why can the following year not be covered at all from this record?

Readers try this anyway, so it is worth stating flatly. Covering a year needs two things: that year's available cash, and that year's debt service. The record carries neither of them for any year after the one modelled.

Start with the cash. The Rs 310 crore of revenue is a single stated figure for the modelled year. There is no traffic forecast standing behind it, no growth assumption, no tariff path and no seasonality. So there is no honest way to produce next year's revenue from this record, and therefore no honest way to produce next year's EBITDA. A figure invented to fill that gap would look exactly like a figure that was computed. Nobody reading it could tell the difference, so it must not be produced at all.

Then the obligation. A rate on a balance is arithmetic, and both the rate and the balance are fixed, so the interest half is knowable at Rs 113.72 crore. The instalment half comes from a schedule the record does not contain, so it is not knowable.

The missing component is a third of the obligation in the one year the record shows, so knowing one of two components is not knowing a ratio, and it is not even close. A reader who has Rs 113.72 crore and wants a cover ratio has to invent an instalment to get one, and the ratio then measures the invented instalment far more than it measures the project.

One habit carries all of this. The absence is named rather than estimated around. The record is missing four specific things: a concession period, a debt tenor, a traffic forecast and a year by year repayment schedule. Every calculation lives inside the one year those absences leave standing, and says so.

What is Debt Capacity, and what has to be fixed before it can be computed?

Debt capacityThe largest amount of debt a given stream of cash could carry while still meeting a stated cover requirement. Capacity is an output of assumptions, never a property of the asset on its own. is the question a sponsor and a lender both ask early: how much debt can this cash stream actually carry? Capacity sounds like a property of the project. It is not. Capacity is the answer to an arithmetic problem, and the problem has two inputs that have to be fixed before any answer exists.

The first input is the cover required. A lower requirement lets more of the cash be committed, so a structure built to hold 1.20 times can carry more debt on the same cash than one built to hold 1.50 times. A cover requirement is the sort of thing that gets stated, so most people supply this input without being asked.

The second input is the amortisation shapeThe pattern in which borrowed money is repaid over time: how much principal falls due in each period, and how that pattern rises, falls or stays level., and this is the input people forget. A repayment shape is rarely stated in the same breath as a cover requirement, and it moves the answer just as hard. Debt service is interest plus principal, so a repayment pattern that returns principal quickly makes debt service large and capacity small, and one that returns it slowly does the opposite, on identical cash and an identical cover requirement.

The record carries no tenor and no schedule, so no general shape can be used. The record does carry this project's own shape for the modelled year: interest at the contracted 9.5 per cent and principal at 5.0 per cent of the amount drawn. Add them and debt service is 14.5 per cent of whatever is borrowed. Holding that shape, capacity falls out of one division: available cash, divided by the cover required, divided by 0.145.

Run it at four requirements. At 1.20 times the structure supports Rs 1,425.3 crore. At 1.30 times it supports Rs 1,315.6 crore. At 1.40 times it supports Rs 1,221.7 crore. At 1.50 times it supports Rs 1,140.2 crore. One check confirms the arithmetic reproduces the record rather than drifting away from it. At 1.3574165 times, the cover this project actually delivered, the same shape supports Rs 1,260 crore exactly, the amount drawn. If it had not come back to Rs 1,260 crore, something in the chain would be wrong.

One small thing is worth noticing. The same arithmetic fed the rounded 1.36 times instead of the unrounded 1.3574165 puts capacity at Rs 1,257.6 crore, Rs 2.4 crore below the drawn amount. Nothing has gone wrong; rounding a ratio to two places and then multiplying a large balance by it simply moves the answer a little. The gap is a useful reminder that a rounded ratio is a reporting convenience and not the thing itself.

How do a lender, a sponsor's finance team and an analyst each read this same year?

Three readers, the same Rs 248 crore, and three different first questions. Watching what each one reaches for is the quickest way to see why the order was written down at all.

The project lenders read the year from the top of the waterfall downwards, and their first question is about the distance between the cash and their own tier. The lenders see Rs 248 crore arriving and Rs 119.70 crore of interest sitting immediately beneath it, and they care about how far the cash would have to fall before it stopped reaching them. On these figures, cash could fall by Rs 65.30 crore, or 26.3 per cent, before the instalment starts to go unpaid, and by Rs 128.30 crore, or 51.7 per cent, before the interest itself is touched. The 51.7 per cent is the number that gets underwritten. The lenders also read the reserve as their own, and they will want to know which of the two funding readings applies long before anybody else asks.

The sponsor's own finance team reads the year from the bottom upwards, and their first question is what actually gets released. Rs 65.30 crore on Rs 540 crore of equity is a 12.09 per cent cash return in this year, and every word of that sentence is load bearing: it is cash, not profit, and it is this year, not a return over the life of anything. The record carries no concession period and no schedule beyond the modelled year. A finance team that reports it any other way has produced a number nobody can check.

An analyst sitting outside the structure has the hardest job of the three. The analyst usually receives the outcome and not the order, and the outcome looks the same whether the order was tested or not. Rs 65.30 crore to the sponsors is exactly what appears in a year where every tier was comfortably full, and exactly what appears in a year where the reserve happened to be pre-funded and would otherwise have swallowed the lot. The analyst's job is to ask which one it was.

Where a sponsor is a listed company, its own reporting of a project's cover ratio to the market sits under the disclosure requirements that the Securities and Exchange Board of India (SEBI) administers. A cover ratio quoted in a listed company's communication is a different artefact from the same ratio inside a lender's own file, and the two are prepared under different obligations.

The household comparison holds for all three readers. The bank that lent to the household reads the salary and asks how far it could fall before the instalment goes unpaid. The household reads what is left after everything and calls that its savings. A relative looking in from outside sees only that the savings went up this month, and cannot tell whether that was a comfortable month or a month in which something that should have been set aside quietly was not.

What stops the sponsors taking the residual when the cover is thin?

Two things, and they work in completely different ways. Separating them is worth the trouble.

The first is the order itself, and it is not a restriction at all. The order is arithmetic. The sponsors stand at tier four, so in a thin year there is simply less cash by the time anything reaches them, and in a thin enough year there is none. Nobody has to enforce that or test it or agree to it in the moment. The outcome follows from the tiers above filling first. At Rs 200 crore of EBITDA the sponsors receive Rs 17.30 crore instead of Rs 65.30 crore without a single person taking a decision.

The second is a written restriction, and it is a different kind of thing entirely. The financing arrangements between Tapti Crossing Infrastructure Private Limited and the project lenders record what happens when the cover falls below an agreed level: what has to be tested, on what dates, and what stops being permitted when the test is not met. A written restriction is a promise recorded in a document, not a consequence of the arithmetic, and it can bite in a year where cash was in fact sufficient to reach tier four.

The order the cash moves in is arithmetic, and the document that makes the order binding is a separate subject. How such a provision is drafted, on what dates it is tested, what happens when a test fails and how any of it is enforced all sit inside a lending agreement, and are covered separately. A reader who thinks the waterfall and the agreement are the same thing will look for the arithmetic in the document and the promise in the arithmetic, and will not find either. Naming that boundary is more useful than blurring it.