Project Finance: Lending Against One Asset's Cash Flow

Project finance lends against one asset's own cash rather than against a company's balance sheet. Tapti Crossing Infrastructure Private Limited borrowed Rs 1,260 crore and raised Rs 540 crore of equity to build one toll crossing, and the project lenders can reach nothing beyond that crossing. Each year the crossing's cash must cover that year's interest and scheduled principal before anything else happens.

Start with the ordinary case. The unusual one only makes sense against it. When a trading company borrows, the lender is looking at a whole business. There are several products, several customers, a stock of assets built up over years, and a history of what the business earned when trade was good and when it was not. If one product line stops earning, the others carry the interest. The loan is written against all of it at once.

Now take away every one of those comforts. Imagine a company formed last year that has never sold anything, that has exactly one thing to sell, that cannot sell anything at all until a large construction job finishes, and whose entire ability to repay depends on how many vehicles use one crossing on one road. There is no second product, no history, and no other business standing behind it. A project lender sits in exactly that position, and the position is why project finance looks and behaves so differently from an ordinary borrowing.

Here is the household version, and it is closer than it looks. A trader borrows Rs 8 lakh to fit out one shop. In the usual arrangement, the lender can come after the trader's house and the savings if the shop fails, so the lender is really lending against the household, not the shop. Now suppose the lender signs a document saying that if the shop fails, it may take the shop, the fittings and the till, and nothing else. The house and the savings are out of reach in writing. The moment that sentence is signed, the lender's whole attention shifts to one question: how much does this shop take in every month, and how reliable is it? The shift from asking who is behind the borrowing to asking what the asset itself collects is the whole of project finance.

Tapti Crossing Infrastructure Private Limited, an invented company, is the worked instance throughout, and every rupee attached to it was invented for the teaching. The company was formed to build and operate one toll crossing and holds no other business. The cost was Rs 1,800 crore, funded with Rs 1,260 crore of debt and Rs 540 crore of equity. In the year modelled here, it collected Rs 310 crore of toll revenue and spent Rs 62 crore running the crossing. The 9.5 per cent it pays on its borrowing is the rate contracted for this project rather than a rate read off a market, and a market rate would move every figure that follows from it.

What is project finance actually lending against?

Money lent to a company formed to build and run one asset, repaid out of what that asset earns. The definition is that short, and everything below follows from it. A great deal of practice follows from two properties, and both are worth holding in mind at once.

The first property is that there is only one asset. Tapti Crossing Infrastructure Private Limited holds a crossing. The company holds no second crossing, no packaging plant and no portfolio of investments. A company built this way is a special purpose vehicleA company incorporated to do exactly one thing and hold exactly one set of assets, so that its results and its obligations cannot be mixed with anybody else's., incorporated for exactly one purpose and kept clear of every other activity. The reason it exists is not tidiness. The company exists so that the cash the crossing collects arrives in an account with no other claim on it, and so that the obligations of the crossing cannot be mixed with the obligations of anything else.

The second property is that there is nothing behind it. The people who put the equity in are the sponsorsThe parties that promote a project, put in its equity and usually build or operate it. They stand outside the project company and are not the borrower., and they usually have other businesses of their own. The other businesses of the sponsors are not part of this arrangement. If the crossing collects less than it needs, the sponsors are not obliged to make up the difference out of anything else they hold. The lenders hold a claim on the crossing and on the company that owns it, and nothing more.

Taken together, those two properties make an arrangement that is best pictured as a ring. Inside the ring sits one company, one crossing, one revenue line and one loan. Outside the ring sits everything else in the world, including everything the sponsors hold. The ring-fenceThe legal and contractual boundary that keeps one set of assets, cash flows and obligations inside a single company, so that claims cannot travel across it in either direction. is the boundary itself, and it is built out of the incorporation of the company, the security taken over its assets and the terms of the agreements it signs. How those documents are drafted is covered separately. The ring's effect on the arithmetic is the subject here.

Notice how the ring changes the question a lender must answer. A lender to an ordinary trading company can ask a soft question: is this a decent business that will keep going? A lender inside the ring has to ask a hard one: in each year, will this one asset collect enough cash to pay what falls due in that year? Nothing else will pay it. The answer cannot be borrowed from anywhere.

What does no recourse beyond the project mean when the cash runs short?

Consider the worst year imaginable for Tapti Crossing Infrastructure Private Limited. Traffic on the crossing falls, the toll collected drops, and the cash available is less than the amount due to the lenders. Where does the shortfall land?

The project lenders can take what the crossing has. The lenders can enforce their securityThe specific assets a lender may take control of and sell if the borrower fails to pay, recorded in a charge over those assets. over the assets of the company, take control of it, and apply whatever cash and value they find there. And then they stop. The sponsors never promised to make up a difference, so the lenders cannot present themselves and ask for one out of the other businesses the sponsors hold. The lenders' claim ends at the boundary of the project company, and that single property is what makes this a project financing rather than an ordinary borrowing. Take the property away and every other answer in this guide changes.

Set that beside an ordinary corporate borrowing so the contrast is concrete. Harivansh Packaging Limited, invented, is a packaging maker that holds borrowings of Rs 740 crore. The Rs 740 crore is not written against one machine or one plant. The borrowings are written against the whole company: several plants, a spread of customers, an inventory, a receivables book and years of trading history. If one plant has a poor year, the rest of the company still pays the interest. A lender to Harivansh Packaging is lending against a portfolio of earning things that happen to sit inside one company. A lender to Tapti Crossing Infrastructure is lending against one crossing, and there is no portfolio behind it.

Now follow the consequence, and the shape of every project financing follows with it. If the lenders cannot reach anything outside the ring, then everything they will ever recover has to be visible inside the ring before they hand over a single rupee. There is no possibility of lending first and looking to a parent later. So the work moves forward in time. The lenders study the construction contract, the operating arrangements, the right to collect the toll and the cash the crossing is expected to produce, and they satisfy themselves before the drawdownThe act of actually taking money under a loan that has been agreed. A facility can be signed long before any of it is drawn, and conditions usually have to be met first., not after it.

A project financing therefore feels heavy at the start and comparatively quiet afterwards. In an ordinary borrowing, a lender can review a company every year and form a fresh view. Inside the ring, the lender's protection is very largely built before construction begins. Afterwards there is one asset in one place and the money has already gone into concrete. The requirements the lenders impose, and how they record them, are set out under the project lender's requirements. The reason the requirement exists at all is the ring.

How is the split between debt and equity decided, and who decides it?

Tapti Crossing Infrastructure Private Limited cost Rs 1,800 crore to build. Of that, Rs 1,260 crore came from the project lenders and Rs 540 crore came from the sponsors as equity. Work the proportions rather than reading them: Rs 1,260 crore divided by Rs 1,800 crore is 70.0 per cent exactly, and Rs 540 crore divided by Rs 1,800 crore is 30.0 per cent exactly. Set one against the other and the debt to equity ratio is Rs 1,260 crore over Rs 540 crore, or 2.33 times.

One question matters more than the numbers. Who chose 70 to 30? The natural assumption is that the sponsors did, the way a household decides how much of a house purchase to borrow. The assumption is wrong, and the decision in fact runs the other way round.

The split is not a preference the sponsors expressed; it is what the cover requirement leaves room for. The arithmetic shows why the direction of the decision has to run that way. The cash the crossing collects is what it is: Rs 310 crore of revenue less Rs 62 crore of operating cost. The cash figure does not move when the funding moves. But both the interest and the instalment are struck on the loan, so the amount that has to be paid to the lenders in a year moves in direct proportion to how much was borrowed. So the more of the Rs 1,800 crore that is borrowed, the larger the annual obligation standing against a fixed amount of cash, and the thinner the covering gets.

The causal order follows from that. Start from the cash. Decide how much room the lenders require between that cash and the annual obligation. The room required fixes the largest annual obligation the project can carry. Working backwards from the obligation gives the largest loan, and whatever the loan does not fund, the sponsors must fund with equity. The equity is the residual, not the starting point. A household hitting a lending limit meets exactly the same logic in reverse: told that the instalment cannot exceed a certain part of the monthly salary, the household is being told what it may borrow, and the down payment is simply the rest of the price.

Two consequences follow, and both are worth naming. The first is that the Rs 540 crore of equity is not merely a source of money; it is the buffer that stands between a poor year and the lenders. The equity is put in first, is paid last, and absorbs a shortfall before the lenders feel one. The second is that the sponsors cannot lift their share of the project by borrowing more. Borrowing more changes the covering, and changing the covering is precisely what the lenders will not allow. How much a project can carry, and how a lender fixes that requirement, is covered separately. The direction of travel is settled: from the cash to the obligation to the loan, with equity as the residual.

How is one year of debt service built out of interest and principal?

Two amounts fall due in the year, and they come from different places. Build them separately.

The first is interest. Tapti Crossing Infrastructure Private Limited pays its own contracted rate of 9.5 per cent on the Rs 1,260 crore it drew. The interest is 9.5 per cent of Rs 1,260 crore, or Rs 119.70 crore. The 9.5 per cent is the rate this project contracted for, not the going rate for Indian project lending, and a rate a point higher or lower would change every figure in this section.

The second is scheduled principalThe part of the loan itself that the agreement requires to be repaid in a given period, as distinct from the interest charged for the use of it.. The agreement requires the company to give back a stated slice of the loan in the modelled year, and that slice is Rs 63 crore. Notice what it is as a proportion: Rs 63 crore on Rs 1,260 crore is exactly 5.0 per cent of the amount drawn. The instalment is not a charge for using the money. The instalment is the money itself, going back.

Adding the two gives the debt serviceEverything the borrower must actually pay the lenders in a period: the interest charged plus any principal the agreement requires to be repaid in that same period.: Rs 119.70 crore plus Rs 63 crore is Rs 182.70 crore. Debt service is the amount the crossing has to find in the year, in cash, before anything else happens with the money.

Here is the thing readers skip, and it is worth slowing down for. Principal is not a cost, and it still has to be paid in cash, so any measure that leaves it out is answering a different question from the one a project lender is asking. Interest is the charge for using money in the period, so a profit statement records it. The same statement does not record the repayment of the loan. Giving back borrowed money is not an expense; it is a reduction of a liability. Both facts are correct accounting. Both leave the company Rs 182.70 crore lighter.

Any household with a home loan already knows this in the body. The instalment that leaves the account each month has two parts inside it. One part is interest, the price of the loan. The other part reduces the loan itself. The bank balance falls by the whole amount either way, so nobody paying that instalment feels the two parts differently. Only the recording changes between them, and the recording is exactly what misleads a reader who works from a profit statement alone.

What does a cover ratio of 1.36 times actually say?

Now set the cash and the obligation side by side. The crossing collected Rs 310 crore and spent Rs 62 crore running itself, so EBITDAEarnings before interest, tax, depreciation and amortisation. What the trading of the business produced before any financing, tax or accounting charge is taken out. is Rs 310 crore less Rs 62 crore, or Rs 248 crore of earnings before interest, tax, depreciation and amortisation. The margin is Rs 248 crore over Rs 310 crore, or 80.0 per cent.

The margin is the most misread figure in infrastructure, and it deserves a line of its own. An 80.0 per cent margin is ordinary for a road. The asset is built once, at enormous expense, and then it mostly sits there collecting, with a modest crew and modest running costs. The same margin in a packaging business or a restaurant would be extraordinary and probably a sign of an error in the accounts. So the 80.0 per cent is a property of what a road is, not evidence that this road is a good business. The expense a road really carries went in before the first vehicle crossed, and the whole of this guide is about servicing that expense.

The measure a project lender uses is the debt service cover ratio, and it is built rather than quoted. The cash the project produced in the year is divided by the amount the project had to pay in the same year. Rs 248 crore divided by Rs 182.70 crore is 1.357417, written here as 1.36 times.

Now say what that figure means, in both halves. Readers usually stop at the first half. The first half: in the modelled year, Tapti Crossing Infrastructure Private Limited covered its obligations and had Rs 248 crore less Rs 182.70 crore, or Rs 65.30 crore, still in hand. Against debt service of Rs 182.70 crore, that spare is 35.7 per cent, so a plain way to say 1.36 times is that the cash was about 36 per cent more than the amount that had to be paid.

The second half: it says nothing whatsoever about any other year. Not the year before, not the year after, not the tenth year. The ratio was computed on one year's revenue against one year's obligation, and both halves of it will be different numbers in a different year. Traffic changes. The interest is struck on a loan balance that reduces as principal is repaid. The schedule itself, over whatever tenorThe length of time a loan runs before it must be fully repaid. A longer tenor spreads the same principal over more years and makes each year's instalment smaller. the loan runs, can require a different instalment in a different year. A cover ratio is a photograph, not a description of a life.

Naming the base and the period is not a caveat attached to the ratio, it is the ratio's definition. There is no such thing as a cover ratio in the abstract. There is a cover ratio of this cash over this obligation in this period. Two people can compute an honest cover ratio on the same project and get different answers simply because one struck it on the cash before maintenance and the other struck it after, or because one used the year just finished and the other the year ahead. Neither is lying. The two people defined different things. Agreements between borrowers and lenders therefore write out the definition of the ratio at length rather than using the name alone.

What is Interest Coverage, and why is it not the test here?

Interest coverage asks a narrower question than the cover ratio does: how many times over could the cash produced in the period pay the interest charged in that period? Compute it on this project. Rs 248 crore of cash divided by Rs 119.70 crore of interest is 2.07 times.

Look at the two figures together. The contrast is the lesson. The same project, in the same year, on exactly the same Rs 248 crore of cash, reads 2.07 times on one measure and 1.36 times on the other. On the first it looks comfortable. On the second it looks thin. Nothing about the crossing changed between the two divisions.

One measure counts the Rs 63 crore of scheduled principal and the other does not, and that Rs 63 crore is the entire distance between 2.07 times and 1.36 times. Nothing else is in the gap. Not a different cash figure, not a different year, not an adjustment. One division has Rs 119.70 crore under the line and the other has Rs 182.70 crore, and the difference between those two denominators is the instalment.

So why is interest coverage not the test a project is judged on? Because the project must pay both amounts, and a measure that counts only one of them describes a smaller obligation than the one the company actually faces. If the crossing collected Rs 130 crore in a bad year it would clear its interest twice over on the narrower measure and still fail to hand the lenders their instalment. The narrower measure would have said comfortable while the bank account said short.

None of that makes interest coverage useless. Interest coverage is exactly the right measure for a different question: whether a business can carry the cost of the money it uses. For an ordinary trading company that borrows against its whole business, refinances as facilities mature and never faces a fixed repayment date on all of it at once, the cost of the money is the live question. Harivansh Packaging Limited, the invented packaging maker, earned earnings before interest and tax (EBIT) of Rs 339 crore against a finance cost of Rs 60 crore, so its interest coverage is 5.65 times. Nothing in its position forces a large repayment in one specified year the way a project schedule does. Inside the ring, the repayment is specified, so the wider measure is the one that binds.

One honest note before leaving this section, and it is a note about what cannot be computed rather than what can. Interest coverage is often struck on EBIT rather than on EBITDA. The EBIT version has depreciation taken out of the cash figure first. The record for Tapti Crossing Infrastructure Private Limited carries revenue and operating cost, and it carries no depreciation charge for the project company at all. So the EBIT version of interest coverage cannot be computed here. The missing depreciation is not a small or awkward gap; there is simply no figure to produce. Where the input is missing, the honest move is to say the measure cannot be computed rather than to produce a number that looks like one. The 2.07 times computed here is the EBITDA version and is labelled as such wherever it appears.

Why is 1.36 times a ceiling rather than a result?

Everything so far has been computed from what the record carries. Now look at what the record does not carry. The absences change how the answer should be read.

The record for Tapti Crossing Infrastructure Private Limited carries revenue, operating cost, the contracted interest rate and the scheduled instalment. The record carries no tax charge for the project company and no maintenance spending. Both of those are real things that a real toll crossing would meet. A road surface wears and has to be relaid. A company that earns a profit meets a tax charge on it. Neither figure is in the record, so neither has been taken out of the Rs 248 crore.

Ask which direction the missing figures push the answer. Both are deductions from the same pool of cash. Take either of them out and the numerator falls while the denominator stays where it is, so the ratio falls. Neither could make it rise. The 1.36 times is therefore a ceiling: the highest the cover ratio can be on what has been published, with the true figure at or below it, and every conclusion drawn in this guide is drawn at that ceiling.

Reading a figure as a ceiling rather than as a point is a habit worth building, and it changes what can safely be said. A conclusion that survives at the ceiling survives everywhere below it. The real figure can only be thinner, so the conclusion that the covering here is thin holds. A conclusion that the covering was generous would lean on a number that has two known deductions missing from it, and the whole conclusion could disappear when they arrive.

The household version is a lending conversation everybody recognises. Somebody says they can afford the instalment because they take home Rs 60,000 a month and the instalment is Rs 22,000. Then come the school fee, the electricity and the money sent home each month, and none of that was in the sum. The Rs 38,000 that looked spare was a ceiling on what is spare, and the actual room was much smaller. The arithmetic was never wrong. The reading of it was.

Move the control and watch what happens. The shape of the change is what is worth carrying away. The bar of cash never moves. The crossing collects what it collects regardless of how it was funded. The obligation line climbs steadily as the debt share rises, and the room between the two closes. The constraint on the funding split is now drawn rather than asserted: every extra rupee borrowed adds to an annual obligation that a fixed amount of cash has to cover, so the funding split is not a free choice.

There is a second reading in the panel that surprises people, and it is worth stating plainly with its limits attached. As the debt share rises, the cash left for the sponsors falls faster than their equity does. At 70 per cent, Rs 65.30 crore on Rs 540 crore of equity is 12.09 per cent for the year. At 80 per cent it is Rs 39.20 crore on Rs 360 crore, or 10.89 per cent. At 85 per cent it is Rs 26.15 crore on Rs 270 crore, or 9.69 per cent. The reason is arithmetic and not judgement. Debt service is 14.5 per cent of whatever is drawn. The cash produced is Rs 248 crore on a cost of Rs 1,800 crore, or 13.78 per cent of the cost. Now be careful about what that does and does not show. The observation is about cash in one year, not about returns over a life. The instalment inside debt service is money going back to the lender rather than a cost, and the record carries no other year in which to see the loan get smaller. Read it as a cash observation and stop there.

What is left for the sponsors, and over what period?

Follow the cash to the end of the modelled year. Tapti Crossing Infrastructure Private Limited produced Rs 248 crore and paid Rs 182.70 crore to the project lenders. Rs 65.30 crore remains. The Rs 65.30 crore is what the sponsors keep for the year. Against the Rs 540 crore of equity they put in, Rs 65.30 crore divided by Rs 540 crore is 12.09 per cent.

Now be very careful with what that 12.09 per cent is. The 12.09 per cent is one year's cash measured against the amount put in, and it is not a return over the life of anything. A return needs a period attached to it and a stream of amounts across that period. The computation is a single amount in a single year, divided by the original equity. A cash yield for the modelled year is the right label, and such a figure should never stand next to one that has a life attached without saying which is which.

The 12.09 per cent is also a ceiling, for exactly the reason the cover ratio is. The Rs 65.30 crore is what remains after debt service out of an EBITDA figure that has had no tax charge and no maintenance spending taken out of it. The sponsors are last in the order, so both would arrive before the sponsors see anything. So 12.09 per cent is the highest this figure can be on what is published, and the actual cash to the sponsors is at or below it.

There is one more limit worth naming, and it belongs to the shape of a project rather than to this record. The sponsors are last in every year, not just this one. The sponsors put their money in first, before a single vehicle crossed, and they are paid only after the operating costs and the lenders in each year that follows. The ordering is the price of standing behind a ring-fenced borrowing. Where in the order each payment sits, along with the reserve the lenders usually require to be funded before anything reaches the sponsors, is set out under debt service.

What can a single modelled year never show?

The useful skill is naming absences out loud instead of gesturing at them. Four things are missing from the record for Tapti Crossing Infrastructure Private Limited, and each one closes off a whole set of calculations that a reader might otherwise attempt.

There is no concession period. Nobody has said for how many years the company may operate the crossing and collect the toll. There is no debt tenor. Nobody has said over how many years the Rs 1,260 crore is repaid, only that Rs 63 crore falls due in the modelled year. There is no traffic forecast. Nobody has said how much the crossing will collect next year or in the tenth year. And there is no year by year schedule, so there is no table of what falls due when.

So take the list of what cannot be computed and hold it as firmly as the list of what can. The record holds no life for the crossing, so there is no return over that life. A payback period needs a run of years, and the record carries a single year. A total toll ever collected needs a concession period and a traffic forecast, and the record carries neither. A cover ratio for any year other than this one needs that year's cash and that year's obligation, and the record carries neither of those either.

Naming those four absences is the honest reading, and filling any of them in by division or by assumption is how a project analysis goes wrong quietly. Quietly is the important word. Nobody announces that they have assumed a twenty year concession. The assumption arrives as a small step: the analyst multiplies Rs 65.30 crore by a number of years that felt reasonable, and from that point on every figure downstream carries an invention that nobody can see. The figure looks like arithmetic. The figure is an assumption wearing the clothes of arithmetic.

What does the whole structure look like run end to end?

Run it once in order, so the sequence sits in one place. Every line below is computed from the case rather than carried over from a previous section.

Read the ladder from the top and the causal order is visible in it. The cost had to be funded, the funding fixed the obligation, the obligation stood against a cash figure the funding could not change, and what survived that meeting went to the sponsors. Nothing in the ladder is a matter of opinion, and every line reconciles: Rs 119.70 crore plus Rs 63.00 crore is Rs 182.70 crore, and Rs 248 crore less Rs 182.70 crore is Rs 65.30 crore.

Then close the run with the statement that governs how all of it should be read. No tax charge and no maintenance spending for the project company appears anywhere in this record. Both would come out of the same Rs 248 crore, and both would arrive before the sponsors do. So 1.36 times and 12.09 per cent are the highest each can be. Every conclusion drawn here is drawn at that ceiling. The ceiling is the only safe place to stand, and a conclusion drawn there can only get stronger as the missing figures arrive.

How does a project lender, a sponsor or an analyst use this?

Three readers, three different first questions. Each of them is reading the same record for something different, and the differences are worth seeing.

A project lender reads the cover ratio and immediately asks what would have to go wrong for it to reach 1.00 times. At 1.00 times the cash exactly meets the obligation with nothing left over, and the arrangement stops working there. On this project the answer is arithmetic: cash would have to fall from Rs 248 crore to Rs 182.70 crore, a fall of 26.3 per cent. So the lender's real question is whether a fall of that size in toll collection is a remote possibility or a plausible bad year, and everything the lender requires before drawdown is built around narrowing that question.

A sponsor's finance team reads the same record from the bottom up. Their money is last, so they start at the residual and work backwards to what has to hold for it to arrive at all. The finance team will notice, as the panel above showed, that borrowing a larger share of the cost does not simply hand them a larger figure on a smaller base, and that the covering thins at the same time. The same team reads the equity for what it is: money put in at the start against cash that only appears once the crossing opens.

An analyst covering an infrastructure holding reads it as a question about what may honestly be said. Their discipline is not computing the ratio, a single division, but refusing to extend one year across a life that no published figure describes. The best analytical work on a project like this is often a list of the things that cannot be computed and the reasons why. Such a list is what stops a reader from importing an assumption without noticing.

The household version of all three is the same reflex, and it is worth carrying into the finance version. When somebody says a shop makes good money, the useful questions are: over what period, before or after which costs, and what has to be paid out of it that has not been mentioned yet. A figure without a base and a period is not yet a number that can be used, so the base and the period come before any judgement about whether the figure is large.