The Project Lender: What They Underwrite and Require

A project lender underwrites a cash stream it cannot verify from history, so it underwrites the arrangements that produce it instead: who builds, who operates, who pays and on what terms. In return it requires a separate vehicle, a cover ratio it sets, a funded reserve, control of the accounts and a limit on what the sponsors may take out.

The lender has been asked for Rs 1,260 crore. The borrower was incorporated recently, has never sold anything, has no customers, no employees to speak of and one asset: a crossing that has not been built yet. There is no five year record to read because there has been no five year anything. Every ordinary credit habit, reading a trend in revenue, testing whether margins hold when volumes fall, comparing this year's borrowings against last year's, has nothing to work on. The file a lender would normally open is empty.

So what does a lender read instead? A lender's reading is the whole of project finance underwritingThe work a lender does before committing money: deciding what it is exposed to, what could go wrong, and on what terms it is willing to lend at all., and the answer is not a clever one. The lender reads the arrangements. Somebody has been granted the right to charge for the crossing. Somebody has committed to build it for a stated price. Somebody will run it once it opens. Somebody, or a great many somebodies, will pay to use it. And a stack of contracts states who carries each thing that could go wrong. A project lender cannot verify a cash stream from history, so it verifies the arrangements that manufacture the cash instead. Its questions sound legal for that reason, even though its exposure is entirely financial.

The worked case throughout is Tapti Crossing Infrastructure Private Limited, an invented single asset toll road company formed to build and operate one crossing, with no other business and no recourse to anybody beyond the project itself. The project costs Rs 1,800 crore, funded Rs 1,260 crore of debt and Rs 540 crore of equity, a 70 to 30 split. In the year this record models it collects Rs 310 crore of revenue against Rs 62 crore of operating cost, so it produces Rs 248 crore of cash before financing. Its own contracted interest rate is 9.5 per cent, a term fixed in this project's own documents rather than read off a market.

What is a project lender actually underwriting when there is no trading history?

Not a company. Every instinct a credit reader has is trained on companies, so giving up the idea of a company is the first move and the hardest one. Tapti Crossing Infrastructure Private Limited is a legal container. The container has a board, a bank account and a set of filings, and the lender is exposed to none of them. The lender's exposure is a ring of arrangements, and the container merely holds those arrangements in one place.

The household version makes the shape obvious. A cousin asks for a loan against a tea stall not yet built. There were no takings last year, so there is nothing to read. So what would a lender actually ask? Whether the landlord has agreed in writing to let the stall stand there. Who is building the counter, for how much and by when. Who is going to stand at the stall every morning. Where the customers come from, and whether the office building next door is actually opening. And, if the counter is late or the office never opens, whose problem that becomes. Every one of those is a question about an arrangement, not about a record. Asking those five questions is underwriting a project.

The project version has the same five questions with heavier paper behind each one. There is a right to collect, granted by a granting authority, without which the crossing may be built and may still not be allowed to charge anybody. There is a construction arrangement, under which a party has committed to deliver the crossing for a price. There is an operating arrangement, under which a party will run and maintain it after it opens. There are the users, whose payments are the entire Rs 310 crore of revenue. And there is an allocation of risk, agreed in advance, stating who carries a cost overrun, who carries a delay, and who carries a change in the terms of the grant.

Notice what the lender has done here. The lender has replaced one unanswerable question, whether this business will do well, with five answerable ones, each with a counterparty and a document attached. The substitution is not a trick, and it makes the exposure legible rather than smaller. The lender is not underwriting a business at all, it is underwriting a set of promises made by identified parties, and its whole task is to find the one question no promise covers. Whatever is left uncovered after all five arrangements have been read is the thing the lender is genuinely carrying, and that residue is what the rest of its requirements are built around.

Why does completion come before every other question?

Because until the crossing is built and open, there is no cash. Not a small amount of cash, not an uncertain amount of cash. None. The Rs 310 crore of revenue has not started, the Rs 248 crore of operating cash has not started, and the Rs 182.70 crore of debt service that has to come out of it has nothing whatsoever to come out of. Every arrangement on the previous figure is about a crossing that works. Completion riskThe risk that the asset is not finished, or is finished late, or costs more than planned, so that the cash it was meant to produce arrives late or not at all. is the risk that the project never reaches the part where those arrangements matter.

Completion risk gives the construction stretch a strange property. The construction stretch is the period with the largest amount of money going out and the smallest amount of information coming back, and it is also the period in which a half finished crossing is worth remarkably little to anybody. A completed toll road that under performs still collects something. A crossing that stops at seventy per cent collects nothing at all, and it cannot easily be sold to somebody else. What would anybody else do with it? The asset only becomes an asset at the moment it opens.

The ring around a project is not equally tight at every stage. The construction stretch is the one period in a project's life where the sponsors are usually asked to stand behind it in some agreed form, and the ring closes fully only once the crossing opens. The clean picture, of a vehicle that has no recourse to anybody and services its debt purely out of one asset's cash, is a picture of the operating stretch. Before completion there is nothing else to reach for, so most structures reach back to the sponsors in one way or another. The ring around the project is not absolute from the first day.

The household picture again, and it is exact. If a cousin borrows to open the tea stall, there are no takings during the fortnight the counter is being built, so no lender is lending against them. During that fortnight the lender is lending against the cousin. Once the stall is open and the takings are real, the conversation can move to the takings. Same money, same stall, two entirely different exposures separated by one event.

Why does the lender insist on a separate vehicle instead of lending to the sponsors?

Ask the question in reverse and it answers itself. Suppose the project lenders had instead lent Rs 1,260 crore to the sponsors directly, against the sponsors' general credit, and the sponsors then spent it on building the crossing. What would the lender then hold? A claim on the sponsors, ranking alongside everybody else who has a claim on the sponsors. The crossing's Rs 248 crore a year would arrive into the sponsors' general accounts and mix with everything else they do, and would be available to whoever the sponsors chose to pay with it. The lender would have financed one asset and taken exposure to a whole set of activities it never examined.

Putting the crossing into a special purpose vehicleA company formed to hold one asset or carry out one project and nothing else, so that its cash and its obligations stay separate from everything its shareholders do. does the opposite in both directions. The sponsors' other activities cannot reach the crossing's cash, and the crossing's cash cannot leak into the sponsors' other activities. Most readers meet this idea from the sponsor's side and file it as a protection for shareholders, a way of keeping a large borrowing away from the rest of what they do. The sponsor's reading is not wrong, but it is half the picture, and it is the half that matters less here.

The same wall that stops the sponsors' troubles reaching the crossing also stops the crossing's cash going anywhere the lenders did not agree it could go, so the vehicle protects the project lenders at least as much as it protects the sponsors. The lender did its work on five arrangements attached to one asset. The vehicle is what makes sure those five arrangements are the only things it is exposed to. Without it, the underwriting the lender performed would not describe the exposure it actually took.

There is a plain household version. Ten shops in one market building can share a single service road, a single power supply and a single landlord, so a lender to one of those shops is quietly exposed to nine others. Move the shop into a separate arrangement with its own takings, its own account and its own landlord agreement, and the lender's exposure narrows to the thing it looked at. Nothing about the shop changed. What changed is how much else can reach its till. The nature of a special purpose vehicle in its own right, and how one is formed and held, is covered separately.

How does a required cover ratio decide how much can be borrowed?

This is the part most readers have backwards, and getting it the right way round changes how every project financing reads. The usual mental picture is: a certain amount of debt gets put into the project, and afterwards somebody computes a cover ratio to see how comfortable it looks. Cover, in that picture, is a grade awarded to a decision already made. The usual picture is inverted.

The lender decides, before anything is committed, how much spare cash it requires the project to produce over and above its obligations. The level of spare cash it requires is the required cover ratioThe minimum multiple of a year's debt obligations that a lender insists the project's cash must reach, set before the money is committed rather than measured afterwards.. Once it is set, the arithmetic runs in one direction only, and the amount of debt falls out at the end as a result rather than being fed in at the start.

Run the sequence on Tapti Crossing Infrastructure Private Limited. Step one is the cash: Rs 310 crore of revenue less Rs 62 crore of operating cost is Rs 248 crore. Step two is the requirement the lender sets. Step three divides the cash by the requirement to get the debt service the project can carry. Step four converts that debt service into an amount of debt. On this project's own contracted terms every rupee of debt costs 9.5 per cent of interest plus scheduled principal at 5.0 per cent of the amount drawn, or 14.5 per cent of the drawn amount in a year.

Take a requirement of 1.20 times, purely as an illustration of the mechanism. Rs 248 crore divided by 1.20 is Rs 206.67 crore of debt service the project could carry. Divide that by 14.5 per cent and the debt that fits is about Rs 1,425 crore. Now take the cover this structure actually produces, 1.3574 times before rounding. Rs 248 crore divided by 1.3574 is Rs 182.70 crore, and Rs 182.70 crore divided by 14.5 per cent is Rs 1,260 crore, exactly what was lent. The arithmetic closes on the real number, and that closing confirms the sequence is the right way round.

The amount of debt is an output of the cover requirement, not an input tested against it, so a difference of 0.16 in a ratio nobody outside the room ever sees moves roughly Rs 165 crore of real money into or out of the same crossing. And because the project cost is fixed at Rs 1,800 crore, the money the requirement moves comes straight out of the equity cheque. At the structure as financed, the sponsors put in Rs 540 crore. At a 1.20 times requirement they would have put in about Rs 375 crore, and the funding split would have moved from 70 to 30 towards roughly 79 to 21. The lender's requirement did not merely grade the structure. It wrote it.

What can be said about a requirement this record never states?

Nowhere in the record of Tapti Crossing Infrastructure Private Limited does anybody say what cover the project lenders required. The record carries the project cost, the funding split, the revenue, the operating cost, the contracted rate, the scheduled principal and the reserve. The record does not carry the requirement. So any account stating that the lenders required 1.30 times, or 1.25 times, or any other tidy figure, has invented it.

Reading the structure backwards produces a boundA statement that a quantity lies above or below some level, without saying exactly where it sits. Less than a figure, and considerably more than a guess. rather than a figure. Rs 248 crore of cash against Rs 182.70 crore of debt service is 1.36 times, rounded from 1.3574. The lenders committed Rs 1,260 crore at that cover. Whatever they required, it cannot have been more than what the structure delivers. Had they required 1.50 times, only about Rs 1,140 crore would have fitted and this structure would not exist. So the requirement was at or below 1.36 times, and that is the whole of what is knowable.

Do not skip past that as a weak answer, because it is not one. A bound is a real result, and swapping it for an invented figure trades something known for something unknown. The bound shows the structure was sized to leave roughly a quarter of the cash spare rather than a half: Rs 65.30 crore of the Rs 248 crore survives debt service, and that Rs 65.30 crore is 35.7 per cent of the Rs 182.70 crore obligation. It also shows the lenders were not sizing to a level that would have left the sponsors putting in far more than Rs 540 crore. Ruling out a large part of the space of possible answers is exactly what a bound is for.

The same discipline applies to what 1.36 times itself is allowed to mean. It says the project covered its obligations with about 36 per cent to spare in the year this record models. It says nothing whatsoever about any other year. This record contains no concession period, no debt tenor, no traffic forecast and no year by year schedule. Payback, the return over the life of the concession and cover for year seven each need a tenor, a forecast and a schedule, and none of the three has been supplied. Naming that absence is the correct move. Estimating around it is not.

What is the reserve buying, from the lender's side of the table?

The structure carries a debt service reserveA sum set aside in a separate account and held there, available only to meet the project's debt obligations if the project's own cash falls short in some period. of Rs 91.35 crore. Where does that figure come from? It is exactly two quarters of debt service. A year of debt service is Rs 182.70 crore, a quarter is Rs 45.675 crore, and two quarters is Rs 91.35 crore. Nothing about it is arbitrary and nothing about it is a percentage of anything. It is a length of time expressed in rupees.

From the project's side the reserve just looks like money it cannot spend. Read what it does from the lender's side. Suppose the crossing has a bad quarter. A stretch of approach road is being repaired, traffic diverts, and the quarter's collections come in far below the run rate. In a project with no reserve, that quarter's instalment cannot be paid, and the moment an instalment is not paid, the whole structure moves into a completely different conversation, one about default and remedies and enforcement rather than about a bad quarter. A traffic problem has become a credit event by nothing more than timing.

With Rs 91.35 crore sitting in a reserve, the instalment is paid out of the reserve and the conversation stays about the approach road. The lender is buying time rather than money, and two quarters is precisely how much time it bought: long enough to find out what went wrong and do something about it before anybody has to reach for a remedy. Money is exactly what a reserve does not create. The reserve is the same cash, held back rather than released, and the project is no richer for having it. What changes is that a timing problem is now a funded problem.

Every household understands this without the vocabulary. Keeping two months of loan instalments in a separate box is not extra income, and nobody is wealthier for it. The box means that the month the shop is shut for repairs, the instalment is still paid and nobody has a conversation with the lender. Same money, different position. The record here does not say whether this reserve was funded out of the drawdownThe act of actually taking the borrowed money from the lender, in one or more instalments, as the project needs it. at the start or built up out of early operating cash. That distinction matters a great deal to the sponsors and not at all to the arithmetic worked here.

What does control of the accounts actually give a lender?

Every project financing sends the revenue into accounts the project lenders have a claim over, and money leaves those accounts in an order agreed in advance. The order of payment is covered separately, under debt service. The prior question is why account controlAn arrangement under which the borrower's receipts land in accounts the lender has rights over, so the borrower cannot freely move that money elsewhere. is worth anything at all when the project is already contractually obliged to pay.

A promise to pay is an undertaking that, on a date, somebody at the borrower will initiate a payment. On every good day that is indistinguishable from an actual payment. The difference appears on exactly one kind of day: the day somebody at the borrower would rather use the money for something else. A contractor is threatening to stop work, a repair cannot wait, a shareholder needs cash. On that day, a promise is a claim the lender has to assert and cash in a controlled account is a claim the lender already holds.

Account control does not create a single rupee of cash, it decides where the cash already is when the obligation falls due, so the lender stops depending on anybody choosing to pay. The Rs 310 crore of revenue arrives where the lenders can see it and cannot be quietly redirected. The Rs 182.70 crore of debt service is met from money that never left the perimeter. Nothing has been added and nothing has been taken. The sequence of custody changed, and that turns out to be most of what a security package is doing.

The everyday version is a landlord who asks that rent be paid straight into a designated account rather than handed over in cash. The tenant owes exactly the same rent either way. What changed is that the landlord no longer relies on the tenant deciding, each month, to walk over with it. Nobody thinks the landlord got richer by asking. Everybody understands why they asked.

Why is there a limit on what the sponsors may take out?

Do the subtraction first, because the number is what makes the argument. Rs 248 crore of cash less Rs 182.70 crore of debt service leaves Rs 65.30 crore in the year this record models. On the face of it that money belongs to the sponsors. The sponsors put in Rs 540 crore of equity, the lenders have been paid everything they were promised, and what is left over is the return on the equity cheque. A distribution restrictionA term limiting when and how much cash a borrower may pay out to its shareholders, so that money stays inside the project until agreed conditions are met. stops them simply taking it, and readers usually find that harsh on first meeting.

Then set the two figures beside each other. The reserve is Rs 91.35 crore. The residual is Rs 65.30 crore. The reserve is 1.40 times the entire residual of the year. Which means that if the reserve had not been funded already, one full year of everything left for the sponsors would not have been enough to fund it. These are not two separate pools with a rule sitting awkwardly between them. The reserve and the next instalment are claims on the same cash as the residual, and the sponsors are the only party in the structure who can be asked to wait.

Being the only party who can be asked to wait is the whole of it. The contractor cannot be asked to wait, because work stops. The operator cannot be asked to wait, because the crossing stops being maintained. The lenders cannot be asked to wait, because being asked to wait is precisely the event everything here is built to prevent. The sponsors can be asked to wait, because waiting costs them timing and not existence, and because they hold the residual claim, which is the claim that by definition ranks behind everything else. A restriction on distributions is not a punishment attached to the equity. It is what being the residual claim means, written down so that nobody has to argue about it in a bad quarter.

The household version is a shopkeeper who has borrowed against the shop and keeps two months of instalments aside. In a good month there is money left after the instalment, and the honest question is whether it can be spent on the house. The answer depends on whether the two month box is full. If it is not, the money left over is not spare money, it is the box's money that has not been put in the box yet. Nothing about that arithmetic requires a document to be true. The document only makes it enforceable.

How does a working desk actually use all of this?

Three readers, three genuinely different first questions, and it is worth watching them diverge on the same set of figures. A project lending desk starts at the far end, with the obligation, and works back: Rs 182.70 crore has to be met in a year, so how much cash has to exist, and how confident can it be that the cash exists. Its instinct is to ask what happens to that Rs 248 crore in the worst quarter it can describe, the same instinct as asking how long the reserve lasts. Every requirement described above is that desk turning a worry into a term.

An analyst on the sponsor's side reads the identical structure from the other end. The sponsor's analyst starts with the Rs 540 crore that went in and asks what came back, and the honest answer here is Rs 65.30 crore in the year the record models, restricted, with a reserve standing above it. The same analyst has a second job the lender does not have: noticing that the cover requirement decided the size of its own cheque. Push the requirement down and less equity is needed. Push it up and more is. The negotiation over a ratio is a negotiation over how much of the sponsor's own money is committed.

A credit reader coming from ordinary corporate lending has the hardest adjustment. Their reflexes are all trained on Harivansh Packaging Limited and companies like it: a trading record, several product lines, a set of customers, and the ability to sell one thing to pay for another. Every one of those reflexes is useless here. There is no second product, no other customer set and no other asset. The reason a project lending desk asks so many questions that sound like a lawyer's questions is that its entire exposure is a set of documents, and reading them is the only work there is.

The habit worth carrying away is small and portable. Given any project structure, the thing to read is not the list of protections but the level attached to each one. How much cover was required. How many quarters the reserve funds. And what has to be true before a distribution may be made. Those numbers are the structure. The names of the protections are just how the numbers were filed.

Where are all of these requirements written down?

Every requirement described above has a documentary home. The commitment to lend Rs 1,260 crore, the conditions that must be satisfied before any of it may be drawn, the cover the lender requires and how it is measured, the reserve and how it is funded and topped up, the accounts and who has rights over them, and the conditions that must hold before the sponsors may take anything out, all sit in a facility agreementThe contract between a borrower and its lenders setting out how much may be borrowed, on what terms, subject to what conditions and with what consequences if the terms are breached. and the security documents that go with it.

How a covenant is drafted, how it is tested and on what information, what constitutes a breach, what remedies follow and how any of it is enforced, is covered separately in this subject area. A project lender's wants and the reasons behind them are one subject; how the wanting becomes an enforceable obligation is another, and that one is covered separately.

The reason the split is drawn there is that the two are genuinely different skills. A facility agreement can be read fluently by somebody who still does not know why a two quarter reserve is two quarters, and the reason the reserve exists can be understood exactly by somebody who could not draft the clause that funds it. The reason behind a requirement is the part that transfers to every project a reader ever sees, including the ones documented in a form they have never met.