Political Risk: Sovereign Action That Reaches the Cash

Political risk is the chance that a public authority's action changes what a project may charge, what it must spend, or whether it may operate at all. A single-asset vehicle cannot absorb much of it: at Tapti Crossing Infrastructure Private Limited a 10 per cent cut in revenue takes the cover from 1.36 times to 1.19 times and removes 47.5 per cent of the sponsors' cash.

Most risk lists carry a line that says political risk, followed by a colour or a word. Amber. High. Moderate. Nobody quite knows what the next column should say, so the list moves on. The column stays empty for a reason, and the reason is not that the risk is unknowable. Half of it is perfectly knowable. The knowable half is the one everybody skips.

One split matters more than any other, and it is worth holding on to before any arithmetic arrives. The size of a public authority's action can be computed, on any project, from figures the project already has. The chance of that action cannot be computed at all, not from the project's own figures and not from anything published. A risk note that reports a colour has done neither. A risk note that reports the size and says plainly that the chance is not knowable has done the whole of the work that can honestly be done.

Start with a household picture. The structure is easier to feel before it is easier to read. Think of a person whose entire income is a licence to run one stall on one stretch of pavement, granted by the body that manages that stretch. The stallholder has savings, has a rent to pay, and holds a licence that carries terms. Now suppose the managing body decides that stalls on that pavement may open four hours a day instead of ten. Nothing has been stolen. No contract has been torn up. The stall still stands, the licence still exists, and the takings have fallen by more than half. There is no second stall and no second pavement, so every rupee of that fall lands on the one person.

The stall is the whole shape of political risk in a project financing, and everything that follows adds arithmetic to it. The worked example is Tapti Crossing Infrastructure Private Limited, invented for teaching along with every figure attached to it: a single-asset toll road company formed to build and operate one crossing, with no other business and no recourse beyond the project itself. The crossing cost Rs 1,800 crore, funded Rs 1,260 crore of borrowing and Rs 540 crore of equity. In the modelled year it collects Rs 310 crore of toll receipts, spends Rs 62 crore to operate, and therefore earns Rs 248 crore of earnings before interest, tax, depreciation and amortisation (EBITDA). Its debt service in that year is Rs 182.70 crore, being Rs 119.70 crore of interest at the project's own contracted rate of 9.5 per cent on Rs 1,260 crore, plus Rs 63 crore of scheduled principal.

What is political risk here, and how is it different from ordinary risk?

Define it by the route it travels rather than by a catalogue of things that might happen. Political risk at a project is any action by a public authority that changes what the project may charge, what it must spend, or whether it may operate at all. Three routes, and everything else is a variation on one of them.

Why define it that way instead of listing events? Because a list of events is a list somebody else wrote, for somebody else's project, in somebody else's market. Such a list cannot be checked, cannot be extended, and yields nothing that can be computed. A route is different. A route ends somewhere specific in the figures the project already has. A route can therefore be traced through and given a number at the end. A number at the end is what separates a risk register that reads like a warning label from one that reads like a working paper.

Now the second half of the definition. The second half is what makes this risk its own subject rather than a corner of operating risk. Sovereign actionAn act of a state or of a public body exercising state power, such as a change to the law, a change to a licence, or a decision by the body that granted a right. comes from the same party that granted the right the project runs on. A road company does not collect toll because it built a road. The company collects toll because a public authority granted it the right to collect, and the grant is the asset. When the party that granted the right takes an action affecting the right, the project is not facing a counterparty in a market. The project is facing the source of its own permission.

Hold the contrast with ordinary operating risk clearly. The two are often lumped together. If traffic across the crossing is lighter than modelled because a rival route opened, that is ordinary demand risk: the market moved. If the toll the crossing may charge is capped or deferred by the granting authority, the market did not move at all. The same number of vehicles crosses, and the receipts still fall. Nothing about the physical project changed, and there is nothing operationally the project could have done differently.

The routes also deserve to be traced one at a time rather than mixed. Two of them reduce cash and one of them can stop it. Reducing and stopping are not the same shape of problem, they do not have the same arithmetic, and they are not treated the same way in the documents. A risk note that folds all three into a single word has thrown away the only distinction that would tell a reader which problem is being described.

A ring-fenced vehicle servicing borrowing out of one asset's cash behaves the same way in any market, so the mechanism holds wherever the road is.

Why does the same action hurt this crossing more than an ordinary company?

Because of concentrationHaving a large share of something, income or exposure, resting on a single source rather than spread across several.. Four separate things are concentrated here, and the vehicle has one of each.

One asset: a single crossing, with no second road and no second business. One right: a single grant, without which the tolls cannot be collected at all. One granting authority: a single public counterparty whose actions can move the terms of that grant. And one source of cash: whatever the crossing collects, with nothing else arriving from anywhere. There is no second source to dilute against, so an action that a diversified business would spread across many sources reaches this vehicle at full strength.

Set that beside an ordinary corporate borrower for scale. Harivansh Packaging Limited, also invented, makes rigid and flexible packaging for food and personal care customers, and its position looks nothing like the crossing's. On revenue of Rs 3,180 crore it earns EBITDA of Rs 477 crore, and its finance cost for the year is Rs 60 crore. The finance cost is 12.58 per cent of that EBITDA, committed to servicing its borrowings. At Tapti Crossing Infrastructure Private Limited, debt service of Rs 182.70 crore against EBITDA of Rs 248 crore is 73.67 per cent committed. Nearly three quarters of the crossing's cash is spoken for before anybody upstairs decides what to do with the rest.

Read those two percentages together and the exposure argument makes itself. Harivansh Packaging has 87.42 per cent of its EBITDA left after servicing its borrowings, several customers, more than one product and more than one place it operates in. A public action affecting one product in one place moves a slice of a slice. The crossing has 26.33 per cent of its EBITDA left after debt service, one product, one place and one right. A public action affecting that one right moves everything.

Say the uncomfortable part plainly. The concentration is not a defect in the structure. The concentration is the structure working as designed. The whole point of putting one asset into a ring-fenced vehicle is that the project stands on its own cash and nothing else. Standing on its own cash is what lets lenders underwrite the asset instead of a balance sheet, and what stops a failure at the crossing reaching the sponsors' other businesses. The exposure is the price of the design rather than a sign that somebody did the design badly.

Which routes does sovereign action actually travel to reach the cash?

Three, and they are worth naming as destinations rather than as causes. Route one changes what the project may charge, and lands on the revenue line. Route two changes what the project must spend, and lands on the operating cost line. Route three changes whether the project may operate as expected, and does not land on a line at all: it interrupts the cash.

EBITDA is revenue less operating cost, so routes one and two arrive at the same place and a coverage ratio cannot tell which of the two lines moved. The convergence saves a great deal of argument in a risk discussion. A toll deferral of Rs 20 crore and a new maintenance obligation of Rs 20 crore produce identical arithmetic below the EBITDA line. The deferral and the obligation differ in how fast they arrive and in how large they can get, and not at all in where they end up.

Route three is a different animal, and the documents treat it differently for that reason. A project that may not operate does not have a smaller EBITDA. The operating cost line does not obligingly fall to zero when the receipts do, and debt service is a date in a schedule rather than a share of whatever came in. The project has an EBITDA problem of a different kind.

Route one: what happens when somebody else sets what the project may charge?

Route one is the fastest route to the cash and the one with the least warning. A toll or a tariff that is set, capped, deferred or revised by a party other than the operator moves the revenue line directly, on the date the decision takes effect, without anything physical about the project changing at all. The road is the same road. The traffic is the same traffic. The receipts are different.

Separate two forms of route one. The two feel similar and behave differently. A tariff revisionA change to the price a regulated or contracted operator may charge, usually made by the body that set the original price or the formula behind it. changes the rate itself. A deferral leaves the rate alone and postpones the date on which it may be applied. A postponement sounds gentler and reaches the same cash in the year affected. Both show up as a smaller revenue number, and neither of them shows up as anything else.

Work it on the crossing. The arithmetic is one subtraction and the result is not intuitive. One point of toll receipts at Tapti Crossing Infrastructure Private Limited is Rs 3.10 crore, and because the operating cost and the debt service both hold, the whole of that Rs 3.10 crore comes off the sponsors' residual and none of it off anything else. Rs 3.10 crore against a residual of Rs 65.30 crore is 4.75 per cent. So a one point move in what the crossing may charge is a 4.75 point move in the sponsors' cash for the year.

Now the same one point measured against the cover. The cover is the number the lenders watch. Rs 3.10 crore against EBITDA of Rs 248 crore is exactly 1.25 per cent, so the cover reading moves 1.25 per cent while the residual moves 4.75 per cent. The residual moves 3.80 times as fast as the cover does. The 3.80 is not a coincidence and not a market fact. The ratio is EBITDA divided by the residual, Rs 248 crore over Rs 65.30 crore, and it will be different on any project with a different structure.

One finding matters more than anything else in route one. The two parties are reading two different instruments off the same event, and one instrument is far more sensitive than the other. When a sponsor says a toll decision matters a great deal and a lender says it is manageable, they may both be reading their own number correctly.

Route two: what happens when the action raises what the project must spend?

A change in lawA new or amended legal requirement that applies to the project after it was financed, such as a new standard it must meet or a new obligation it must carry. is the usual carrier here: a changed standard, a new safety or environmental requirement, an added obligation, a service the project must now provide that it did not have to provide before. The action does not touch the toll. The action touches the cost of running the crossing.

Two properties make this route behave unlike the first. Route two is slower. A new requirement usually takes effect after a period and often needs work done before it bites. Route two can also be larger. A requirement can add a cost that was never in the model at all, where a toll decision is bounded by the toll. EBITDA is revenue less operating cost, so the slower route reaches exactly the same place and the coverage ratio has no way of telling which of the two lines moved.

Compute the size rather than assert it. The scale is genuinely surprising. To take the crossing's cover from 1.36 times to 1.00 times through route two alone, the operating cost has to rise from Rs 62 crore to Rs 127.30 crore. The rise is Rs 65.30 crore, or 105.32 per cent. The cost of running the crossing has to more than double. Through route one, the same 1.00 times is reached when toll receipts fall Rs 65.30 crore. The same fall is 21.06 per cent of Rs 310 crore.

The same Rs 65.30 crore, arriving from either side, and two completely different percentages. The reason is the margin. The crossing earns 80.0 per cent on revenue, ordinary for a road and extraordinary anywhere else. A margin that wide makes the revenue line five times the size of the operating cost line. A given rupee amount is therefore a small percentage of the big line and a very large percentage of the small one. Percentages are not comparable across lines of different sizes, and a risk register that ranks a 20 per cent cost rise above a 10 per cent revenue fall has compared two percentages that were never on the same base.

Route three: what happens when the project may not operate as expected?

The third route is the one that changes the shape of the problem rather than the size of it. Permissions that are delayed or withdrawn, access that is disputed, land that is subject to compulsory acquisition of landThe power of a public body to take land for a public purpose, following its own process and usually against payment, whether or not the holder agrees., or a public counterparty that does not meet an obligation it took on. In each case the question stops being how much the crossing collects and becomes whether it collects.

The operating cost line does not fall to zero when the receipts do, and the debt service date does not move at all, so a reduction and a stoppage are different problems. If receipts stop for a quarter, the crossing does not save Rs 15.50 crore of quarterly operating cost by not operating: some of that cost is there whether or not vehicles cross. And the quarterly debt service of Rs 45.68 crore, being one quarter of Rs 182.70 crore, arrives on its date regardless.

A stoppage of that kind is where a relief eventA defined event under a contract which, when it happens, changes what a party must do or by when, rather than paying that party money. earns its place in the documents. A stoppage caused by the granting authority's own action is usually dealt with separately from an ordinary shortfall precisely because it is a different shape: it is treated by adjusting obligations and timing rather than by adjusting a price. How such a provision is drafted, what triggers it and how it would be enforced is covered separately.

Arithmetic can say how much funded patience the crossing has, and the answer comes from the reserve. The project holds a debt service reserveCash set aside in a designated account to meet scheduled debt service if the cash coming in falls short. The cash sits outside what the sponsors may take out. equal to two quarters of debt service, which is Rs 91.35 crore. Two quarters is the whole of it. The account holds no more than that, and whether two quarters is enough depends on how long a stoppage runs.

Push the arithmetic one step further, and name the assumption out loud. A plausible number gets invented at exactly this kind of step. Suppose the reserve had to carry both the debt service and the full operating cost with nothing coming in. The quarterly burden is then Rs 45.68 crore plus Rs 15.50 crore, or Rs 61.18 crore. Rs 91.35 crore divided by Rs 61.18 crore is 1.49 quarters. Whether the operating cost would actually run in full during a stoppage is not something the record settles, so treat 1.49 quarters and 2.00 quarters as the two ends of a range and not as a forecast.

What do the contracts usually do about this, and where is the limit?

In general terms, a concession commonly provides for some form of compensation or relief where the granting authority's own action is what caused the loss. The compensation term is the ordinary mechanism, and it is the reason a project of this kind is financeable at all. Without something of the sort, lenders would be underwriting the discretion of a counterparty rather than the cash of an asset.

Now the limit that readers skip, and it is the whole of this section. A compensation provisionA term in a contract requiring one party to make good a defined loss suffered by the other. The term creates an entitlement to be paid, not a payment. is a claim against a counterparty, so it converts a revenue problem into a receivable and a timing problem, and a project with two quarters of reserve has two quarters of patience.

The provision does not put cash in the account on the day the toll is capped. The provision creates an entitlement. Between the entitlement arising and the money arriving there is a process, and the process takes time that nobody can put a number on in advance. A schedule does not pause for a claim under discussion, so the debt service dates keep arriving. So the practical question about a compensation provision is never only whether the project is entitled. The question is also how long the project can carry the gap, and the reserve gives the answer.

The household version is exact. An employer accepts that it owes an employee two months of unpaid salary and will pay it. The entitlement stands, the amount is agreed, and the month still has to be got through with the rent due on the fourth. Between now and then the entitlement does not matter. The balance in the account does. A project sits in precisely that position. Its account holds a stated number, Rs 91.35 crore, and its date is a fixed schedule.

How far can revenue fall before this project's cover is gone?

Here is where the argument stops being an essay. Only the revenue line moves, the operating cost holds at Rs 62 crore and the debt service at Rs 182.70 crore, and the arithmetic makes the case. Every row below is one subtraction and one division.

Take the last row on its own. The last row is the most useful single output of the whole table. With operating cost held at Rs 62 crore, toll receipts can fall 21.06 per cent before EBITDA reaches Rs 182.70 crore, the cover reaches exactly 1.00 times and the sponsors receive exactly nothing, and all three of those things happen at the same instant. They happen together because they are the same event described three ways: EBITDA equal to debt service is a cover of 1.00 times is a residual of zero.

Derive the figure rather than quote it. The derivation is a single line that runs on any project. The headroom is EBITDA less debt service. Rs 248 crore less Rs 182.70 crore is Rs 65.30 crore. The headroom expressed as a share of the line being moved, revenue, is Rs 65.30 crore over Rs 310 crore, or 21.06 per cent. Headroom over the line being moved is the whole method.

And now the trap that catches almost everybody. A cover of 1.36 times sounds like 36 per cent of room, so 36 per cent is the first instinct. The first instinct is wrong. The 35.74 per cent is the amount by which EBITDA exceeds debt service, measured on debt service, and an excess measured that way is not a fall in anything. The second instinct corrects for the base and asks how far EBITDA can fall. Rs 65.30 crore on Rs 248 crore is 26.33 per cent, and 26.33 per cent is a true statement about EBITDA. But a public authority usually moves the revenue line, and revenue is bigger than EBITDA, so the same rupee headroom is a smaller percentage of it. Three different percentages, 35.74, 26.33 and 21.06, are all true statements about the same Rs 65.30 crore, and they differ only in what they are measured against.

One ceiling has to be stated before anybody carries these numbers anywhere, and it makes them worse rather than better. The record for this crossing carries no tax charge and no maintenance spending. Missing deductions mean every cover figure in the table above is the highest it could be, and 21.06 per cent is the largest fall this project could survive rather than the fall it can survive. Insert the deductions the record does not carry and every row moves down. Read the sensitivities as a ceiling, never as a forecast.

What happens to the sponsors long before the lenders are touched?

Read the table again down the last two columns instead of across, and the finding is unmistakable. At a 10 per cent reduction, the project lenders have lost nothing at all: they were owed Rs 182.70 crore and they received Rs 182.70 crore. Over the same event the sponsors went from Rs 65.30 crore to Rs 34.30 crore. Nearly half their cash for the year, 47.5 per cent of it, is gone.

The sponsors absorb the whole of the first 21.06 per cent of any fall in toll receipts, and the lenders begin to be touched only after that. The order is not a matter of fairness or of who negotiated better. The order follows entirely from the lenders' claim being fixed and the sponsors' claim being whatever is left. Both sides agreed to that arrangement at the start and priced for it.

The sensitivityHow much one number moves when another one is changed by a stated amount, worked out by changing the input and recomputing rather than by estimating. arithmetic makes the asymmetry precise, and it is worth writing the ratio down. One point of toll receipts moves the cover 1.25 per cent and the residual 4.75 per cent, so the sponsors' number moves 3.80 times as fast as the lenders' number. The two parties watching the same crossing on the same day are watching instruments with different gearing, and neither of them is misreading.

One more asymmetry is easy to miss, and the figure below is drawn for it. The two measures do not reach trouble at the same moment. At a 20 per cent reduction the cover still reads 1.02 times, and 1.02 times by itself sounds survivable. The sponsors' cash for the year has meanwhile fallen from Rs 65.30 crore to Rs 3.30 crore, 94.95 per cent gone, a cash return of 0.61 per cent on Rs 540 crore of equity. A ratio above 1.00 times can sit quite comfortably alongside an equity position that has effectively been emptied for the year.

Which part of political risk cannot be sized at all?

Everything above sized a consequence. Nothing above put a chance on one, and no honest treatment of political risk ever will. A consequence can be computed from figures the project already holds, and likelihoodHow probable something is, usually written as a chance or a frequency. Likelihood is a claim about the future that needs a source, not a calculation. cannot be computed from anything at all, so an honest note sizes the consequence and declines the chance.

The reason matters. A claim that something is not known carries weight only alongside what would have to be true for it to be known. Putting a probability on a public authority capping the toll at this crossing would need a defined population of comparable events, a defined period, and a count. No such population exists for this crossing, no such count has been made, and no published figure would produce one. Inventing a number in that position is not conservatism and it is not judgement. An invented number is arithmetic with nothing underneath it.

And inventing a number costs more than leaving the space empty. A risk note that puts a 10 per cent chance in the column looks quantified. The 10 per cent goes into a weighted average, gets multiplied by a consequence, and produces an expected figure that then travels through a decision as though it were computed. Every reader downstream inherits an invented input dressed as a result. A probability nobody can source replaces a known unknown with an invented number, and the invented number is harder to argue with than the honest blank.

So what does an honest note look like? Two lines. The first sizes the consequence in the project's own figures: a 10 per cent reduction in toll receipts takes the cover to 1.19 times and removes 47.5 per cent of the sponsors' cash for that year. The second states the position on chance: the likelihood of such an action cannot be estimated from anything published, and no number is offered. A note of that shape can be checked by somebody who disagrees, extended by somebody with better figures, and revisited when the figures change. A colour can do none of those things.

How does a lender, a sponsor's finance team or an investor actually use this?

Three readers, three different first questions, and it is worth watching how differently the same table is used. None of them is reading it wrongly.

The project lenders read the table for one thing: the distance from 1.36 times to 1.00 times, expressed in the units of whatever they think can move. A public action usually reaches the revenue line, so the lenders express that distance as 21.06 per cent of toll receipts. The lenders then ask what sits between the project and that fall. The reserve of Rs 91.35 crore is part of the answer. Whether the concession carries a compensation mechanism is part of the answer. The lenders are sizing the distance to the point where their own claim stops being met in full, not the chance of an action.

The sponsor's finance team reads the same table down the residual column and reaches a different conclusion from the identical figures. Their number moves 3.80 times as fast, so the reductions that a lender would call manageable are the reductions that empty their year. The sponsors are also the party that has to fund anything the reserve cannot. A sponsor's own treasury planning therefore starts at reductions far smaller than the ones the lenders discuss.

An infrastructure investor considering the equity reads the table as a question about what the residual is worth, and immediately hits a ceiling. The Rs 65.30 crore is one year's cash on Rs 540 crore of equity, and the record carries no concession period, no debt tenor and no traffic forecast. So there is no return over the life to be computed here, no payback, and no coverage ratio for any year other than the one modelled. An investor takes the sensitivity method and runs it on a record that does carry those things.

Take the household version once more. The reflex is the same at every scale. Where a household's income comes from one contract with one counterparty, the useful question is never whether that counterparty is likely to change the terms. The counterparty's intentions cannot be known. The useful question is what happens to the month if they do, and how many months of savings sit behind the answer. A project asks the same two questions with bigger numbers and a written reserve account.

Where do the rules about any of this actually live?

The mechanism above is arithmetic on an invented record, and it holds wherever the road is. The requirements, on the other hand, are set by authorities and by the concession document itself, and they are exact, they change, and they belong to the bodies that write them.