Preferred Return: The Hurdle Before Carry Begins

A preferred return is a threshold, not a payment. Nobody is owed it as a debt and nobody receives it as income. A preferred return is an amount the investor class must have received before the manager's share of profit begins, and if the fund never produces it, it is never paid. Nilgiri Growth Partners Fund II, invented, has accrued Rs 2,29,98,27,451 of it at Year 9 Quarter 2 and paid none.

The word itself misleads on both halves. Almost every reader arrives at this subject having been misled by it at least once, so start with the word. A preferred return is not a return in the ordinary sense of something a person earns and then holds. The word preferred does not mean ranking ahead of somebody who lent money. A preferred return is a splitting rule: a level of money that has to have reached one side of a contract before the other side is allowed to share in anything, and it is not a return that anybody ever actually receives as a separate cheque. Everything that follows is a consequence of that one sentence.

Is a preferred return a payment, a promise or a threshold?

Picture two people opening a juice cart at the end of a busy street. One of them puts up the money for the cart, the crates, the licence and the first month of fruit. The other one stands there twelve hours a day and runs it. The two of them write down how they will split what comes in, and the sentence they write is this: the one who stood there gets a share of the profit only after the one who paid has taken back every rupee they put in, and then a further amount worked out at eight rupees a year for every hundred rupees that was tied up, for as long as it was tied up.

The further amount worked out at eight rupees a hundred is a preferred returnAn amount the investor class must receive before the manager shares in profit.. Notice what it is not. The preferred return is not a wage, and it is not a fee. Nobody has agreed to hand it over out of their pocket. The amount exists only inside the arithmetic of splitting whatever the cart happens to make. If the cart makes very little, the one who paid takes back what they can, the one who stood there gets nothing extra, and neither of them has broken the agreement. If the cart is a disaster, the money is gone and there is nobody to sue. The preferred return did not promise a result; it only decided who gets what if a result arrives.

Run that sentence again as a private fund and nothing changes except the scale. Nilgiri Growth Partners Fund II is a closed-end growth and buyout fund managed by Nilgiri Alternatives Advisors Private Limited, invented. In the vocabulary the documents actually use it is set up as a trust, with Nilgiri Trusteeship Services Private Limited, invented, as trustee and Nilgiri Financial Holdings Private Limited, invented, as sponsor, and the role that global practice calls the general partner is discharged between the manager and the trustee. The contract is a trust deed and a contribution agreement rather than a partnership agreement, and the economic vocabulary used throughout is the vocabulary the reader will actually meet in those documents.

Twelve investors committed Rs 4,90,00,00,000 to that fund and the manager committed a further Rs 10,00,00,000 of its own, for total commitments of Rs 5,00,00,00,000. The record date is the end of Fund II Year 9 Quarter 2. By that date Rs 4,80,00,00,000 had been drawn from those commitments, and every figure that follows is struck at that one date. The fund contracts a preferred return of 8.0 per cent a year, compounded annually. The 8.0 per cent is this invented fund's own contracted term and nothing else, neither usual nor market, neither high nor low.

So what is it? A preferred return is one line in a distribution orderThe sequence in which a fund pays out what it has received.: the agreed sequence in which money coming into the fund is handed out. The sequence for this fund puts the return of every rupee ever drawn first. Only when that is complete does the preferred return line become live at all, and only when that line is satisfied in full can the manager begin to share in profit. Because the line sits there and has to be got past, the same thing is also called a hurdleThe same thing, named for what it does rather than what it is.. A hurdle is obviously a thing to be cleared rather than a thing that is handed over, so hurdle is the better name.

The single most useful test is this: ask who would be in default if the preference were never paid. The honest answer is nobody, and that is exactly why a preferred return is not a debt. A lender who is not paid has a claim. A supplier who is not paid has a claim. An investor whose preferred return is never reached has a position in a queue that the money never got to, and there is no claim anywhere in that sentence.

What does the preference accrue on, and from which day?

Two questions hide inside that heading and they have different answers. The first is what the 8.0 per cent is charged against. The second is when the clock starts. Get either one wrong and the number that comes out is wrong by crores.

Take the base first. The preference does not accrueTo build up over time without being paid. on what an investor promised. The preference accrues on what the investor actually handed over and has not yet had back, an amount called unreturned capitalDrawn capital that has not yet been distributed back to investors.. A commitment that has never been called has cost the investor nothing, so nothing accrues on it. Money that has been called and then returned has stopped being tied up, so it stops accruing from the moment it comes back. Nothing accrues on an uncalled commitment, so the fund's unfunded commitment of Rs 20,00,00,000 at the record date, being Rs 5,00,00,00,000 less the Rs 4,80,00,00,000 drawn, plays no part in any of the arithmetic below.

Now the clock. A private fund does not take its money in one go. Nilgiri Growth Partners Fund II drew its Rs 4,80,00,00,000 in seventeen separate calls spread across more than eight years, each one issued to every investor in proportion to what they had committed, and each one met in full. The first was Rs 13,10,00,000 in Fund II Year 1 Q1 for organisational expenses and the first year of fee and expenses. The seventeenth was Rs 2,20,00,000 in Fund II Year 9 Q1, for fee and expenses and nothing else. The preference accrues on each rupee from the day that rupee was drawn. Seventeen calls therefore start seventeen separate clocks, and no single number of years describes the fund.

The picture kills the most common shortcut in the subject. Taking the full Rs 4,80,00,00,000 and charging 8.0 per cent against it for the 8.50 years since the fund closed gives Rs 3,26,40,00,000. The figure is wrong, and wrong in an interesting direction. Most of that Rs 4,80,00,00,000 simply was not there for 8.50 years, so the shortcut runs far too large. Thirteen of the seventeen calls fall inside the fund's five-year investment period and account for Rs 4,55,50,00,000 of the total, but even those arrived across five years rather than on day one, and the last four calls, spread over three and a half years of fee and expenses, have barely had time to accrue anything at all.

There is a clean way to say how much time the money has actually been outstanding. Add up every rupee multiplied by the years it has been tied up, and divide by the total. On the straight-line basis that produces the equivalent of the whole Rs 4,80,00,00,000 having been outstanding for 4.5702 years, not 8.50. The 4.5702 years is why the shortcut fails, and it is a reasonable thing to work out for oneself the next time somebody quotes a hurdle as though a fund had one start date.

The fund writes every part of this down in its day-count conventionThe agreed rule for how an accrual is measured over time.: the agreed statement of how time is measured, where each cash flow is treated as falling, and what the accrual is charged against. In this fund a flow in Year n Quarter m is treated as falling at n minus one, plus a quarter of m, years after the final close, so Year 1 Q1 sits at 0.25 years and Year 9 Q2, the record date, sits at 8.50 years. The convention is not a detail buried in a schedule but a term of the contract, and a great deal of money rides on it.

Does it compound, and what does compounding actually add?

The preferred return of Nilgiri Growth Partners Fund II is compounded annuallyAccruing each year on the previous balance as well as on the original amount.. At the end of each year the preference that has built up but has not been paid stops being a separate figure sitting to one side and joins the balance the next year's 8.0 per cent is charged against. Unpaid preference itself starts accruing preference.

Everybody thinks they understand compounding until they are asked to say what it costs, so here is the cost, on one invented fund, at one date. Charge the same 8.0 per cent, on the same seventeen calls, on the same dates, and count it as a straight line with nothing ever added to the base, and Nilgiri Growth Partners Fund II has accrued Rs 1,75,49,60,000 at its record date. Compound it once a year, as the contract says, and the same fund has accrued Rs 2,29,98,27,451. The gap between those two readings is Rs 54,48,67,451, more than the fund's entire management fee of Rs 70,20,00,000 across most of its life, and the compounded reading is 1.31 times the straight line one on identical facts.

The lower panel of that picture contains the thing readers get wrong in the opposite direction, so it deserves more attention than the upper one. At the end of Year 1 the straight line reading is Rs 1,88,60,000 and the compounded reading is Rs 1,84,68,509. The straight line is ahead, by Rs 3,91,491. Compounding has not helped the investor class in the first year at all; it has cost them a little.

The reason is simple once seen. Annual compounding credits nothing part way through a year. An annual rest waits for the year to close and only then adds what has built up to the base. A straight line does not wait; it credits a proportion of the year as the year runs. Over any period shorter than a whole year, an annual rest gives less than a straight line, and the fund's first calls all landed mid-year, so most of Year 1 is exactly that kind of fractional period. By the end of Year 2 the position has reversed: Rs 14,02,62,709 compounded against Rs 13,97,00,000 straight, so the compounded reading is now Rs 5,62,709 ahead, and from that point it never trails again. The gap then widens every single year, and that widening is the whole of what compounding does.

Two things in that table are worth naming out loud. The first is that the difference between the two columns grows faster than either column does. Compounding ordinarily behaves that way, and it is why a hurdle on a fund that runs long is a very different animal from a hurdle on a fund that runs short. The second is that neither column is a return anybody in this fund has received. Both are readings of a threshold. Nothing in that table has been paid to anybody, and the fund's investors would be no richer if the number in the left column were twice as large.

Which convention does this fund use, and how was it proved?

Here is a claim that sounds like pedantry and is not: the sentence "a preferred return of 8.0 per cent a year, compounded annually" does not, on its own, determine what any number will be. At least four different readings of that sentence are perfectly grammatical, they are all in use somewhere, and on this manager's own record they are wrong by tens of crores relative to one another.

The four readings are these. One, compound once a year on unreturned capital plus unpaid preference together, so the preference itself accrues. Two, compound once a year on unreturned capital alone, so the unpaid preference sits to one side and never earns anything. Three, no compounding at all: a straight line on the balance. Four, compound four times a year rather than once, on the same balance.

The way to settle which one a fund means is not to argue about it but to find a case where the answer is already known and see which reading reproduces it. The manager of Nilgiri Growth Partners Fund II has one. Nilgiri Growth Partners Fund I, invented, wound up at its own Year 10 Q4, and its record is complete: Rs 2,40,00,00,000 paid in, Rs 4,80,00,00,000 distributed, and a preferred return of Rs 1,18,00,00,000 actually paid out across its last two distributions. Run all four readings against that fund's ten drawdowns and four distributions and see which lands.

Compounding once a year on capital plus unpaid preference together gives Rs 1,17,99,99,738 against a known Rs 1,18,00,00,000. The miss is Rs 262 on a figure of a hundred and eighteen crore, rounding and nothing else. Compounding on capital alone gives Rs 87,06,90,000, wrong by more than a quarter. A straight line gives Rs 1,19,07,02,530 and quarterly compounding gives Rs 1,23,13,14,130, both close enough to look plausible at a glance and both wrong by crores. Three of the four readings are near enough to the right answer that nobody would catch them by eye, and that is precisely why the convention has to be read out of the document rather than assumed.

Two traps are worth carrying away from that exercise. Both of them were live during it. The first is that a control which passes does not prove the case that actually matters is right. Fund I's own distribution dates were entered correctly, so an engine that placed all four of Fund II's distributions one quarter early matched Fund I to the rupee throughout, and it still came out Rs 9,58,69,253 low on Fund II. A test that only exercises the part that was already correct reveals nothing about the part that was wrong.

The second is subtler. Fund I's last event is its own wind-up, so the accrual always runs to a date on which something happens. Fund II's last event is drawdown 17 at 8.25 years, and the record date is 8.50 years, so the whole balance sits accruing across a quarter with nothing happening in it. The quiet quarter alone adds Rs 5,18,29,978, taking the balance from Rs 2,66,79,97,473 at drawdown 17 to Rs 2,71,98,27,451 at the record date. A fund whose last event is its wind-up can never expose that kind of stub-period mistake. Whenever an accrued preference is quoted, establish the date it accretes to. A figure struck at the last transaction and a figure struck at the reporting date are different numbers.

What do nine years of unpaid preference look like?

Now the worked case, on this fund's own convention, with every figure belonging to Nilgiri Growth Partners Fund II at the end of its Year 9 Quarter 2. The fund has drawn Rs 4,80,00,00,000. The fund has distributed Rs 4,38,00,00,000 in four payments. Its unreturned capital is Rs 42,00,00,000 and its accrued preferred return is Rs 2,29,98,27,451, of which not one rupee has been paid.

Read those last two figures against each other for a moment. The pairing is the point. The accrued preference is 5.48 times the unreturned capital it is currently sitting on. Nine years of an unpaid, compounding threshold does that to a number, and none of it is owed. An investor looking at that Rs 2,29,98,27,451 is not looking at a receivable. The investor is looking at how far the fund would have to get before the split changes.

The most surprising feature of that line is what it does not do. Four distributions fall inside it: Rs 63,00,00,000 in Fund II Year 6 Q4, Rs 2,03,00,00,000 in Year 7 Q3, Rs 1,50,00,00,000 in Year 8 Q2 and Rs 22,00,00,000 in Year 8 Q4. Between them they moved Rs 4,38,00,00,000 out of the fund and into investors' hands, including one payment of over two hundred crore rupees. The preference line does not step down at any of them. Not once.

The reason is the sequence. Every rupee of every distribution went to returning capital first, and the fund has still not finished returning capital, so no distribution has ever reached the preference line at all. A payment that reduces unreturned capital slows the accrual down, and next year's 8.0 per cent is charged on a smaller balance. The payment does not reduce the preference that has already built up. Money leaving the fund and the preference falling are two different events, and on this fund only the first has ever happened.

Every unreturned capital figure in that table is simply the cumulative paid in less the cumulative distributed, checkable across a row, and the accrued preference column is what 8.0 per cent compounded annually does to that base over those dates. Notice Year 7 in particular. Cumulative distributions jumped from Rs 63,00,00,000 to Rs 2,66,00,00,000 in a single year and unreturned capital fell by nearly two hundred crore rupees, and the accrued preference still went up, from Rs 1,52,56,54,899 to Rs 1,93,35,89,579. A big distribution slows the accrual; only a distribution that actually reaches the preference line reduces it.

The last two readings on that picture are worth a paragraph on their own. Between the end of Year 8 and the record date, unreturned capital went up. Unreturned capital rose from Rs 39,80,00,000 to Rs 42,00,00,000, a distance of Rs 2,20,00,000, and nothing bad happened in those six months. Drawdown 17 of exactly Rs 2,20,00,000 was called in Fund II Year 9 Q1 for fee and expenses, and no distribution was made in either quarter. The fund got further from returning its investors' capital by doing nothing except paying its own running costs. The last stretch of a closed-end fund looks exactly like that. The rise is about one pixel at the honest scale of that plot. The panel on the right of it therefore magnifies those two readings about fifty three times rather than stretching the main scale and overstating a small real movement.

How far does this fund still have to travel?

A threshold is only meaningful as a distance, so measure it. The fund's contract is whole-of-fundEvery rupee of capital ever drawn must come back before any profit share is paid., meaning every rupee ever drawn from investors has to come back before any profit share arises, and that includes capital drawn for the management fee and for expenses just as much as capital drawn to buy a company.

Three marks on one scale. The fund stands at Rs 4,38,00,00,000 of cumulative distributions. Every rupee ever drawn has come back at Rs 4,80,00,00,000, a further Rs 42,00,00,000 on. The preferred return has been met in full at Rs 7,09,98,27,451, another Rs 2,29,98,27,451 beyond it. Only at Rs 7,09,98,27,451 of cumulative distributions does the manager receive its first rupee of carried interestThe manager's share of a fund's profit, as distinct from its management fee., and that level is 1.48 times what investors paid into this fund. The total remaining distance from where the fund stands is Rs 2,71,98,27,451.

Rs 2,71,98,27,451 is a large number and it deserves an honest frame rather than a dramatic one. The fund's total value at the record date, counting the Rs 4,38,00,00,000 already distributed plus the Rs 2,82,00,00,000 of holdings still carried on its books, is Rs 7,20,00,00,000, or 1.50 times paid in. So the hurdle sits Rs 10,01,72,549 below the fund's own stated total value. But the Rs 2,82,00,00,000 half of that stated value has never been sold to anybody. The Rs 2,82,00,00,000 is five unrealised holdings carried at valuations, one of them already written down, and what any of them will realise, when, and whether the hurdle will ever be cleared are all unknown. A distribution order pays on cash that has actually arrived. A carrying value is an estimate. Cash and an estimate are not the same object, and putting the two beside each other is the honest reading rather than the encouraging one.

One more thing that number is not. The remaining distance is not a target, a projection or a statement about what this fund or any fund will do. The distance is arithmetic on two figures that are already fixed: a hurdle level struck at a stated date, and a distribution total struck at the same date.

What happens if the fund never earns it?

The answer to that question is the reason a preferred return is not a debt.

Follow the right-hand branch carefully. If the fund's remaining holdings realise less than the distance requires, or realise nothing at all, then the accrued Rs 2,29,98,27,451 is simply never paid. There is no claim against Nilgiri Alternatives Advisors Private Limited, no claim against the trustee, no claim against the fund and no claim against anybody else. In the ordinary legal sense there are no arrears, so no interest runs on them. Nothing has fallen due. The figure was never a liability of anything; it was a level in a splitting rule, and the money never got to it.

An investor who has accrued Rs 2,29,98,27,451 of preference has accrued a position in a queue, not a receivable, and the difference between those two things matters more than anything else about a preferred return. A receivable can be enforced, sold, discounted, written off against tax and shown as an asset. A position in a queue can be none of those things. The position exists only for as long as there is a queue, and the queue exists only for as long as money is arriving to be split.

Go back to the juice cart. If the cart shuts after two bad years, the person who put up the money does not get to send a bill to the person who stood there for eight rupees per hundred per year. The share of profit that person would have taken simply never arises, and the person who paid takes back whatever the crates and the licence are worth. Neither of them has been wronged. The arrangement did exactly what it said: it decided a split, and it never promised an outcome.

The left-hand branch matters too. If cumulative distributions do reach Rs 7,09,98,27,451, the preferred return has been met in full out of money the fund actually received. Every rupee of it went to the investor class as part of a distribution. The rupees after that one, and the way the manager's share is then struck, belong to a different mechanism with its own arithmetic, covered separately.

Preferred return ranks ahead of what, exactly?

The other half of the word is as misleading as the first half. Preferred sounds like a rank, and in most of finance a rank settles who stands where when there is not enough money to go round. Preference shares rank ahead of ordinary shares. A first charge ranks ahead of a second. A secured lender ranks ahead of an unsecured one. Readers arrive here carrying all of that, and it does not apply.

The preferred return ranks ahead of the manager's share of profit and ahead of nothing else at all in the world. It is a term in a contract between a fund and its own manager about how to divide money that has already arrived. The term says nothing whatever about anybody outside that contract.

Working down the chain shows how many things get paid first. Kaveri Renewables Private Limited, invented, is holding 8 of this fund's nine holdings. If that company has borrowed from a bank, the bank is paid on its own terms out of the company's own cash, and the fund's preferred return has no bearing on it whatever. The company's suppliers are paid. Its staff are paid. Its tax is paid. Only whatever is left over at the end of all that has any chance of becoming a sale price, and only a sale price can become cash in the fund. The whole of this guide's arithmetic starts after every one of those payments has already happened.

Then, inside the fund itself, more things come first. The management fee of Rs 70,20,00,000 drawn to the record date has already been paid to the manager. Fund expenses of Rs 9,80,00,000 have already been paid to administrators, valuers, auditors and lawyers. Fee and expenses add to Rs 80,00,00,000, and neither was paid out of the preference or after it. Both were drawn from investors as capital, so both sit inside the Rs 4,80,00,00,000 that has to come all the way back before the preference line is even reached. And the return of that Rs 4,80,00,00,000 itself comes first.

What does somebody actually do with this on a Tuesday?

Nobody meets this subject in the abstract. The people who need it are the ones who open a quarterly report and find a line called accrued preferred return with a large number beside it. The work is a sequence of six questions taken in order, and every one of them has an answer available in the document.

A person on the investment team at investor 4, the treasury of a domestic bank that committed Rs 60,00,00,000 to this fund, runs exactly that sequence when the pack arrives. So does the fund controller at the administrator, Kolar Fund Services Private Limited, invented, though from the other side. The controller has to produce the figure rather than interrogate it, so the controller is the one who must read the day-count convention out of the trust deed and apply it the same way every quarter. And so does anybody sitting on the investor advisory committee, chaired for this fund by Meera Sathe, invented, representing investor 1.

The household version of the same discipline is smaller and identical in shape. When a relative asks for money to put into a shop and offers the first eight per cent of the profit each year before taking anything, the six questions are the same six. Is the eight per cent charged against the money put in, or against that money plus whatever went unpaid last year? Does last year's unpaid amount itself start earning? What happens if the shop earns four per cent, and what happens if it earns nothing? And, the one everybody forgets: is there a written agreement, and does the written agreement say what the conversation said?

Who does the hurdle protect, and from what?

The hurdle protects the investor class, and it protects them from exactly one thing: a manager taking a share of profit before the investors have had their capital back and a stated amount on top of it. The protection ends there.

A term with the word preferred in it invites a reader to feel safer than they should. Be blunt, then, about everything the hurdle does not do. The hurdle protects nobody from loss, makes no fund safer than it would otherwise be, and does not improve any outcome by a single rupee: it decides how an outcome is divided, and only that. If every one of this fund's remaining five holdings were written off tomorrow, the preferred return would not return one rupee to anybody. The write-offs would simply mean the manager gets nothing either.

The hurdle does change timing and behaviour, and that is not nothing. Because the hurdle is whole-of-fund, and because it compounds, every year that a fund holds an investment without selling it raises the level at which the manager begins to share. On this fund the level rose by Rs 12,42,53,027 in the last two quarters alone, from Rs 6,97,55,74,424 at the end of Year 8 to Rs 7,09,98,27,451 at the record date. Rs 2,20,00,000 of that is drawdown 17, being fee and expenses that have to come back like any other capital, and the other Rs 10,22,53,027 is nothing but two quarters of accrual on a balance nobody paid down. A manager sitting on unrealised holdings in year nine of a ten-year fund is watching that number move away from it. Whether that is a good design or a bad one is a separate question, and it is certainly not a reason to think well or badly of any structure. The arithmetic simply behaves that way.

The last honest thing to say is about the gap between two true sentences. The fund's total value is 1.50 times what investors paid in, and its net internal rate of return, counting the unsold holdings at their carrying values, is 8.3 per cent, above its own 8.0 per cent hurdle. And this fund has not paid one rupee of preferred return and has not reached its hurdle. Both sentences are true at the same date about the same fund. A distribution order pays on cash that has arrived. A rate of return counts an estimate of what has not been sold. A fund can be above its hurdle on one measure and Rs 2,71,98,27,451 away from it on the one that actually governs the split, and nothing has gone wrong in either sentence.