InvITs: infrastructure cash flows with an end date

What an InvIT actually holds

An InvIT is a listed trust that owns finished, operating infrastructure — toll roads, power transmission lines, gas pipelines, telecom towers, renewable generation — and passes most of the cash those assets produce to its unitholders. The full name is an infrastructure investment trust. Units trade on the exchange, and what you own is a claim on the cash, not on the concrete.

The reason the structure exists is worth a sentence, because it explains the shape of everything that follows. A company that builds a road has its money locked inside a finished asset for twenty years. Selling that asset into a trust and listing the trust converts the locked capital back into cash for the builder, and hands the steady operating cash flow to investors who wanted the cash rather than the construction risk. The construction risk and the operating cash have been separated on purpose, and the InvIT is the half holding the cash.

So far this describes something that sounds like a bond. It pays regularly, the payments come from a contracted stream, and the asset behind them is dull by design. That resemblance is real, and it is also the source of nearly every misunderstanding about the instrument. The rest of this article is about where it breaks — and the break is not the one most coverage names.

The same wrapper as a REIT, a different thing inside it

An InvIT and a real estate investment trust — a REIT — are built from the same parts. A sponsor contributes the assets. An independent trustee holds them. An investment manager runs them for a fee. The assets sit inside separate operating companies the trust owns. SEBI requires most of the net distributable cash flow to be paid out rather than retained, at a stated minimum frequency, which is why both throw off cash instead of compounding it internally. Both are business trusts in the tax code, and both list and trade like a share.

The wrapper being identical is exactly why the difference inside it gets missed. A REIT owns buildings. A building is still standing at the end of a lease; the tenant leaves and another one signs, at whatever rent the market gives you. The risk is that the rent re-prices or the floor sits empty for a while. What a REIT holds does not expire — it re-prices.

Many InvITs own something different: a concession. A concession is a right granted by a public authority to operate an asset and collect its revenue for a stated number of years, after which the asset transfers back to the grantor. Not sold. Handed back. The road is in fine condition on the last day and the trust owns nothing of it on the day after.

Two caveats before this hardens into a rule. Not every InvIT asset is a concession — some are owned outright, and some sit under licences that are renewed as a matter of course rather than surrendered, which makes them behave far more like the REIT column. And a trust that keeps acquiring new assets can hold a portfolio whose average remaining life stops shrinking. The structure is a per-asset question, not a per-category one, and the trust's own disclosures are where it gets answered.

A concession has an end date, and the asset is wasting by contract

Concessions come in recognisable shapes, and which one an asset sits under decides what you own the risk of. Under a build-operate-transfer road concession with toll collection, the trust collects from traffic and keeps what it collects. Under an annuity or hybrid-annuity structure, the granting authority pays a contracted sum for keeping the road available, whatever the traffic does. Under a toll-operate-transfer arrangement, an already-built public road's collection rights are sold for a fixed term. Transmission and pipeline assets have their own versions, usually a tariff for keeping the line or the pipe available to a standard.

What every one of them shares is a termination date written into the contract. That makes the asset a wasting one in a very particular sense: not worn out, not obsolete, simply out of time. Physical condition and economic life have been decoupled — a well-maintained road with two years left on its concession is worth two years of cash to the trust and nothing thereafter, and no amount of resurfacing changes that.

The trust's own accounts already say this, in a way that confuses people who read them looking for a share. The concession right is carried as an asset and written down over the concession period, so a large non-cash charge sits between cash collected and profit reported. Reported profit therefore runs well below cash generated, and a trust distributing more than its accounting profit is behaving exactly as the structure intends. It is not, by itself, a sign of anything wrong. If the difference between cash and profit is unfamiliar territory, how to read a profit and loss account sets out where non-cash charges sit and why they matter here.

This is also why InvITs report a figure called net distributable cash flow rather than earnings per unit as the headline. The industry did not invent that measure to flatter itself; profit is the wrong measure for an entity whose main charge is the amortisation of a right it is deliberately consuming. But the measure has a cost attached, and the cost is that the number now contains two economically different things. Cash returned and cash earned arrive on the same day, in the same rupee, and nothing on the payment advice separates them.

Why the distribution is not a yield

Because part of every payment is your own capital coming back, and when the concession expires there is nothing left behind to have earned it. Here is the arithmetic, on illustrative round numbers chosen so it can be redone by hand. No figure in this section is any real trust's. Suppose a unit costs ₹100, the trust distributes ₹10 a unit each year, and the concession behind it has 15 years left, after which the asset goes back to the grantor and the trust holds nothing.

The screen calls this a distribution yield of 10%. Now total it up. Fifteen payments of ₹10 is ₹150 in, against ₹100 out, and then it stops. The gain is ₹50 spread across fifteen years, earned on capital that is shrinking the whole way. The rate that makes fifteen annual payments of ₹10 worth ₹100 today is a little over 5.5% a year, not 10%. In a spreadsheet that is =RATE(15,10,-100); the 10% is simply the payment divided by the price, which is a different question with a different answer.

Run it the other way and the size of the gap is clearer still. If a buyer of that same 15-year stream wanted 9% a year, the annual payment would have to be about ₹12.40 per ₹100 — a headline distribution yield near 12.4% to deliver a 9% return. The wedge between the two numbers is not a fee, not a market view and not an opinion. It is the capital coming back.

Compare an instrument where the capital does not come back until the end. A Post Office Monthly Income Scheme pays 7.4% and hands your deposit back at maturity; the Public Provident Fund credits 7.1% and the balance is still yours. In both, the quoted rate is a return on capital, and the capital is a separate thing that survives. In a finite-life InvIT, one payment carries both, and only the trust's own disclosure tells you the split.

Those two rates are worth reading for their shape and not their level, because they are not the same kind of number as a distribution yield. A small savings rate is administered — notified by the government each quarter and the same for everyone on the day it applies. A distribution yield is a market outcome that moves every time the unit price does, and nobody sets it. Comparing the two as though they were rival offers is the beginning of the same error this section is about.

This is where the instrument's closest honest cousin becomes useful. A bond's yield to maturity already performs this adjustment: it takes the coupons, the price paid and the redemption amount and returns the single rate that reconciles all three, which is why a bond bought above par has a yield below its coupon. A distribution yield does no such reconciliation. It divides this year's payment by today's price and stops.

So give the failure mode a name, because it is the one that costs real money: the self-liquidating distribution. It is a payment that looks like income, screens like income, and is partly your own principal being handed back on a schedule. Spend all of it and you have not lived off a yield — you have drawn down a corpus while believing you left it intact. The specific mistake is lining an InvIT's distribution yield up against a deposit rate or a bond's yield to maturity in one column, as though the three numbers answered the same question.

Which is the general point this cluster keeps arriving at. A yield is a summary, and every summary discards. This one discards the end date, the split between income and capital, the variability of the payment, and everyone standing ahead of you in the queue to be paid. Four omissions in one number, and the number gives no hint that it made them. Other single-figure measures drop different things — the standard risk ratios discard the shape of the distribution and treat a good surprise and a bad one as the same event.

Where the equity risk actually lives

The title of this article says equity risk, and this is the section that earns it. Nothing about an InvIT distribution is contractual to you. The trust collects revenue, pays operating and maintenance costs, pays interest on borrowings at the trust and at the operating companies, pays tax, and distributes what remains. You are the residual claim, in exactly the sense a shareholder is, and the order of that queue is the whole risk story.

Illustrative arithmetic again, and again nobody's real numbers. Say an asset collects ₹100 of revenue, spends ₹30 running and maintaining itself, and pays ₹40 in interest. That leaves ₹30 for unitholders. Now let revenue come in 10% lighter, at ₹90. The maintenance still has to happen and the lenders still have to be paid, so ₹90 less ₹30 less ₹40 leaves ₹20. A 10% shortfall in revenue arrived as a one-third cut to the distribution.

That amplification is not a flaw in the example; it is what fixed claims ahead of a variable one do, and it runs in both directions. Revenue 10% ahead of plan, at ₹110, leaves ₹40 — a third more. A bond's coupon does neither of those things, which is what the two numbers together are for. The sensitivity scales with how much debt sits in front of you, which is why the trust's consolidated leverage is not a governance detail but the multiplier on everything else.

SEBI caps that leverage as a proportion of asset value, with a higher band available only where the trust carries a stated credit rating and unitholders approve it. The ceiling is a regulatory limit, not a comfort: near the cap, further acquisitions have to be funded with new units instead of new debt. The cap is a brake on growth as well as on risk, and a trust that has used most of its headroom has fewer ways to replace a concession that is running out.

The variability itself differs enormously by asset, and this is the distinction that a yield number cannot show you:

• Volume risk — a tolled road collects from whoever drives on it. Traffic responds to the economy, to fuel prices, to a competing road opening, and to a diversion during construction next door. You own that variability directly.
• Availability risk — an annuity road or a transmission line is paid for being available to a standard, not for how much traffic or power crossed it. The operational variability largely disappears and is replaced by counterparty risk: whether the paying authority pays, and how promptly.
• Regulated tariff risk — where a regulator sets the tariff periodically, the revenue is stable between resets and the risk is concentrated at each reset.

Two trusts can show the same distribution yield on the same screen while one owns traffic and the other owns a receivable from a state entity. Those are not variations of one risk; they are different risks with different failure modes, and only the asset-level disclosure distinguishes them. There is one more term to check while you are there: whether the tariff or toll escalates with a price index or a fixed annual step, or is flat in rupees for the term. A flat rupee payment stretching fifteen years loses purchasing power the entire time, which is the arithmetic inflation does to any fixed nominal stream.

What moves the unit price between now and expiry

A finite series of cash flows priced today is, mathematically, a bond — whatever the risk of the cash flows. So the discount rate moves the price, and it moves it through the same channel that duration describes for bonds and that interest rates describe generally. A rise in the rate a buyer demands lowers what the same stream is worth today.

There is a counterintuitive consequence worth stating, because it inverts what people expect. A self-liquidating stream returns its capital gradually, all the way through its life, while a perpetual one returns none of it. So at the same headline yield, the finite stream's cash arrives earlier on average and its price is less sensitive to rates than a perpetual holding paying the same — a REIT, for instance, whose buildings do not run out. On this one axis the concession structure is the more defensive of the two, which is not the direction the rest of this article has been pointing.

Read that narrowly, because it holds on the nominal discount rate and on nothing else. The rent behind a perpetual stream can be re-let higher; a concession payment fixed in rupees cannot be re-anything. The very feature that shortens the rate exposure — the stream running out — is the one that leaves it with no way to grow into inflation, however many years the concession has left. Less sensitive on one axis, more exposed on another, and a single yield figure shows neither.

The second mover is run-off, and it is pure arithmetic. Every year that passes removes one payment from the remaining stream. If nothing else changed, the unit price would drift down toward zero across the concession's life, because that is what a self-consuming asset does. So a unit price that holds level through years of run-off is not evidence that the assets improved. The likelier reading is that the market is pricing in acquisitions the trust has not made yet — which is a judgement about a manager, not a fact about a road.

Which brings in the third mover. An InvIT that keeps buying assets can hold its distribution level or growing indefinitely, and a trust that intends to says so in its stated strategy. But an acquisition is paid for with new units or new debt, and new units divide the same pie into more slices. Total distribution and distribution per unit are different statements, and only the second is about you. An acquisition adds value per unit only if what the assets produce exceeds what the units and debt issued to buy them cost — and the announcement rarely puts it in those terms.

The last one is dull and catches people at exit. Only publicly offered and listed InvITs trade on an exchange at all; a large part of the framework covers privately placed trusts an ordinary buyer never sees. Among the listed ones, traded volume is a fact to check on the screen before assuming a position can be sold at the quoted price, in the same way you would check it for any thinly traded listed security — the mechanics of holding and trading units are no different from a share.

What the disclosures answer, question by question

Everything above turns into a small number of things to look up. None of this is a screening rule and none of it says what to own; it is the list of questions the structure raises, matched to the disclosure that answers each one.

• Weighted average remaining concession life — how many payments are left. This is the single number that decides whether the distribution is mostly income or substantially capital.
• The distribution split by component — the trust reports each payment as interest, dividend and repayment of debt, the last being loan principal an operating company has returned to the trust and the trust has passed on. That component is the return-of-capital portion, named for you, in the trust's own filing.
• The revenue model, asset by asset — toll, annuity, availability or regulated tariff, and whether it escalates with prices or is flat in rupees.
• Consolidated leverage and the credit rating — how much fixed claim sits ahead of the residual one, and therefore how hard a revenue miss lands.
• The major-maintenance schedule and any reserve against it — see below, because this is the one that hides.
• Related-party acquisitions and the approvals taken — the trust often buys from its own sponsor, which is a conflict written into the structure rather than a scandal, and is handled by unitholder approval above stated thresholds.
• Sponsor unitholding and lock-in — how much of the outcome the party that sold you the assets still shares.

The maintenance point deserves its own paragraph, because it is the least visible thing in the whole structure. A road concession does not spend evenly. Resurfacing and major repair fall in specified years and arrive in lumps, sometimes very large ones, while the distribution is paid out smoothly every quarter. Those two facts cannot both be true without something reconciling them. Either cash is being set aside now, which lowers today's distribution below what the road is currently earning, or it is not, and a future year absorbs the whole bill.

So a distribution that sails through a scheduled overlay year without a dip is a fact that needs an explanation, and the explanation is in the reserve or in the borrowing that replaced it. Smoothness is a choice somebody made, not a property of the asset, and it is the sort of thing that only becomes visible in the year the bill lands.

How the money is taxed on the way to you

The tax treatment is not a footnote here. It is the clearest external confirmation that the argument in this article is right, because the tax code has already decided that part of an InvIT distribution is not income.

A distribution arrives split into components, and each is taxed differently in your hands. An interest component is taxable as interest. A dividend component's treatment depends on a choice made at the operating-company level about which corporate tax regime it sits in. And the component reported as repayment of debt — loan principal coming back to the trust from an operating company it lent to, and passing straight through — is not taxed as income when you receive it at all. Instead it reduces the cost you are treated as having paid for the unit, so it comes back later as a larger capital gain — and beyond a point, once cumulative receipts of that kind exceed what the unit was issued at, the excess becomes taxable in the year you receive it.

That cost-of-acquisition reduction is the argument in statutory form. The law is not taxing that money as income because it has concluded it is not income. It is your capital, returning, and the tax merely waits.

The consequence at exit is the part that surprises people, so take it on illustrative numbers. Buy a unit at ₹100. Suppose ₹4 of each year's ₹10 distribution is the return-of-capital component. After ten years you have received ₹40 of your own money back and your cost is treated as ₹60. Sell at ₹100 — the price you paid, no gain in any ordinary sense — and there is a taxable capital gain of ₹40. The gain was real; you collected it a bit at a time and were not taxed on it then.

On the units themselves, listed units of a business trust sold on an exchange take the same capital-gains treatment as listed equity: 12 months of holding to count as long-term, then 12.5% on gains above the annual exemption of ₹1.25 lakh, and 20% if sold sooner. The general machinery is in capital gains tax; what is specific to this instrument is that the cost side of the calculation has been quietly moving the entire time you held it.

One honest gap. The threshold at which the return-of-capital component turns taxable is written against the unit's issue price, which is not the same figure as what somebody who bought on the exchange years later actually paid. How that lands for a secondary-market buyer is a detail worth confirming for a specific holding rather than assuming from any general description, including this one.

What the structure is for, and what it costs

Set beside the alternatives, an InvIT is doing something none of them quite do. It converts a lump of capital into a contracted operating cash stream from an asset class an individual has no other practical way of owning directly, with a professional operator running it and a regulator requiring most of the cash to be paid out rather than reinvested at the manager's discretion. That mandatory-distribution rule is a genuine governance feature, and it is a rule, not a market outcome.

What it costs is equally statable. You take equity-like variability on the cash while being paid in a form that reads as fixed income. You sit behind lenders whose claims do not flex. You accept a manager's fee and a manager's judgement on acquisitions that will decide whether your per-unit stream survives the run-off. And with a concession asset, you accept that a portion of what arrives each quarter is your own capital, so the payment cannot be read as income without checking the split. None of those costs is hidden — every one of them is in the trust's own filings — but none of them is in the yield number either.

Where a holding of this kind sits, or whether it sits anywhere, is a portfolio question rather than an instrument one, and it turns on what the rest of the portfolio already owns and what the money is for. That is the subject of asset allocation, and it is deliberately not the subject here. This article's job was narrower: to make sure that when the yield number is compared against something, it is compared against the right thing.

Where to look at this

Most of the arithmetic above is a spreadsheet exercise on documents the trust already publishes, and no screen does it for you. The concession life, the component split of the distribution and the maintenance schedule come out of the trust's own reports; the annuity rate is one RATE formula once you have them.

The part that is screen work is the ordinary listed-security part. A listed InvIT unit has a price and volume history like any other exchange-traded instrument, and FNOTrader's Market Pulse Stocks Scanner charts and screens listed securities with their traded volumes, which is where the liquidity question in the previous section gets answered rather than guessed.

Nothing there computes a distribution's income-versus-capital split, because nothing can: that number lives in a filing, not in a price series.

Common questions

Continue reading

More in Advanced Investing · App: Market Pulse Stocks Scanner · Definitions: glossary · Free tools: calculators · All: every article