How to map a Private Fund Governance Structure

Map a private fund by what each party can stop, not by where it sits on a chart. Ask five questions in order: who holds the assets, who decides, who consents, what needs no vote at all, and what each party can halt. In Nilgiri Growth Partners Fund II, invented, eight parties answer differently and only two of them can stop an investment.

A housing society is the familiar version of the problem. One person keeps the keys to the store room. Somebody else decides that the lift has to be replaced. A committee has to say yes before that much money leaves the account. Everybody else finds out at the annual meeting, after the cheque has been signed. Four people, four completely different powers, and a pyramid drawn with the secretary at the top would explain none of it. A private fund is that situation written down, signed, and given eight parties instead of four.

Each of the parties is a mechanism covered separately: the party that holds the assets, each of the two committees, the administrator, the valuation agent, and the clause that can suspend new investment on its own. Each name below is used as a known term. A sequence of questions put to a set of fund documents, in the order that makes each answer useful, does the work a chart cannot. One invented fund carries the sequence end to end, so each answer can be seen as it arrives.

Why does a chart of a private fund mislead a reader?

Because a chart draws position and governance is about power, and in this kind of vehicle the two come apart almost completely. The diagram that arrives has boxes stacked vertically, and the eye reads it the way it reads every other diagram of that shape: the thing at the top is in charge, the things below it do as they are told. In a private fund the party at the top of that picture holds every asset and makes no investment decision. The party that makes every investment decision sits to one side and is paid a fee for doing it.

So the first thing to do with the chart is put it down. A table serves instead, with one row for each party and one column for each kind of power. The only column that ever settles an argument is the last one, which asks what this party can stop. Everything else in a governance section is context. Take one specific worry. An analyst who can name the party that could have refused it, and say what refusing would have taken, has mapped the fund. An analyst who cannot has memorised a list of names.

What are the four powers to ask about for every decision?

Take any single decision the fund makes and ask four separate questions of it, in this order. Who proposes it. Who approves it. Who merely consents to it. Who is simply told about it afterwards. Proposing, approving, consenting and being told are four different powers, and a reader who collapses them into one has mapped nothing at all.

Go back to the housing society for a second. The four powers are easier to feel there. The resident who notices the lift is failing and writes to the secretary is proposing. The managing committee that votes to spend the money is approving. The one flat that has to be asked because the machine room sits over its bedroom is consenting, and it can refuse that one thing and nothing else. Everybody who reads the minutes afterwards is being told. Four powers, and the person being told has the loudest opinion and the least power in the building.

Now the finance version, in Nilgiri Growth Partners Fund II, invented. The investment managerThe party that makes the investment decisions for a fund and is paid for doing so. proposes. A committee inside the manager approves. A committee of investors consents, on a short written list of matters and on nothing else. Every investor of the fund is told, through the things the fund sends out after the fact. Refusing has a different consequence at each of the four: at the first two the thing does not happen at all, at the third only that one named matter stops, and at the fourth there is nothing there to refuse.

Step 1: who holds the assets, and does that party decide anything?

The answer to this first question surprises people, and getting it early stops every later mistake. Every vehicle in this worked case is settled as a trust, and the party that holds the assets is the trusteeThe party that holds the assets of a trust and has duties to the people who benefit from it.. In Nilgiri Growth Partners Fund II, invented, that is Nilgiri Trusteeship Services Private Limited, also invented. Every share of every company this fund has bought sits with it.

And it makes no investment decision. Not a joint one, not a confirming one, not a signing-off one. Holding and deciding are two separate powers held by two separate parties, and this is exactly the split that the vertical chart hides. A useful comparison: a bank locker holds the jewellery placed in it and has real duties about it, and it has never once had an opinion about whether the jewellery should have been bought.

The vocabulary in this subject travels from somewhere else, and saying so plainly once saves an argument later. The economics were designed in the form of a limited partnership and imported, so the documents and the investors here use that language: general partner, limited partner, capital account and carried interest. In this fund there is no limited partnership and no general partner as a matter of law: the general partner's role is discharged by the manager and the trustee between them, and the contract is a trust deed and a contribution agreement rather than a partnership agreement. Both vocabularies appear side by side in the same document, and neither is wrong.

Step 2: who actually makes the investment decision?

Two parties, working one after the other, and both of them sit on the manager's side of the map. Nilgiri Alternatives Advisors Private Limited, invented, is the investment manager of this fund. The investment manager decides what the fund buys and what it sells, and it is paid the management fee and the carried interest for doing so. Behind it stands Nilgiri Financial Holdings Private Limited, invented, the sponsorThe party standing behind a fund's manager, which in an Indian trust structure puts up the manager's own money in the vehicle.. The sponsor puts up the manager's own Rs 10,00,00,000 commitment in the fund.

But the manager does not decide alone, and the second half of the answer is the part readers skip. Inside the manager sits the investment committeeThe body that approves each investment a fund makes and each sale of one.: five members, four of them from the manager and one external. The investment committee approves every investment and every realisation this fund makes. Withholding its approval stops either one dead. The manager chooses what to put in front of it. The investment committee decides whether that thing goes ahead.

Two of the eight parties can stop a particular investment, and they stop it in two different ways. The manager stops anything it never proposes, silently, and that never shows up in a document because a deal that was not put forward leaves no trace. The investment committee stops something that was proposed, and that refusal at least exists somewhere as a record. Both belong in the map. A reader who lists only the investment committee has missed the larger of the two.

One invented fund's own documents carry the answers above, and every real set of documents carries answers of its own. The step is a question before it is an answer. Who proposes, and who approves.

Step 3: who consents, and to exactly what?

Now the third power, and this is where the map either becomes useful or falls apart. A consent rightA right to agree or refuse one specific named thing, and only that thing. is a right to agree or refuse one named matter. A consent right is not a general say, and not a seat at the table for everything. A consent right is a written list, and the length of that list is the whole extent of the power.

In Nilgiri Growth Partners Fund II, invented, the consenting party is the investor advisory committeeA body drawn from a fund's own investors that agrees or refuses a short written list of named matters.. The advisory committee has seven members, drawn from investors 1, 2, 3, 4, 5, 6 and 8, and it is chaired by Meera Sathe, invented, who represents investor 1. The committee consents on four things: conflicts, valuation policy, the first extension of the fund's term, and any change to the investment policy. The list of four is the entire power, and an investment is not on it.

Everything a reader gets wrong about that advisory committee comes from imagining the list is longer than it is. So the list goes down on paper before anything else is written about the committee. Consenting to a valuation policy is agreeing how marks will be produced, not agreeing a single mark. Consenting on conflicts is being asked before the manager does something where its own interest and the fund's pull in different directions. Being asked about a conflict is not a say in strategy.

The same fund's own named decisions lay out as a grid. The grid is the object this whole procedure is trying to build. Every row comes from this invented fund's own documents and none of it is a requirement stated by anybody.

Read against the six rows above them, the last two show the shape of the whole thing. The first extension needs the investor advisory committee, seven people, to write yes. The second extension needs a share of the money to say yes, and the money is spread across twelve investors who have never met. The two extensions are two different machines with two different failure modes, and this fund carries both inside one clause about extending its own life. Neither extension has been taken at the record date, which is the end of this fund's Year 9 Quarter 2.

And the fourth power, being told. Six numbered things reach an investor of this fund after the fact: 1 a capital account statement each quarter, 2 an unaudited quarterly report inside a period its own documents fix, 3 an audited annual report, 4 a letter from the manager alongside the quarterly numbers, 5 a notice for every capital call and every distribution, signed by Farida Contractor, invented, the manager's chief operating officer, and 6 an annual valuation report from the independent valuation agent. Six documents, and not one of them is a power. The six are the record of decisions other parties already took.

Step 4: what happens in this fund with no vote at all?

Every map so far has had somebody at the end of the arrow. Step 4 asks the opposite question, and it is the step that separates a reader who has understood the documents from a reader who has listed them. Some of the most consequential things in a private fund happen with nobody voting, nobody consenting and nobody being asked. They are written into the contract as conditions, and when the condition is met the consequence simply arrives.

An automatic provisionA contract term that operates on its own when a stated condition is met, with nobody voting on it. in Nilgiri Growth Partners Fund II, invented, is the key-person provision. The two key persons are Sundari Raghavan and Devendra Karnik, both invented. If both of them cease to devote substantially all of their business time to the fund during the investment period, the investment period suspends. No vote is taken. No committee is consulted. No new investment may be made from the moment the condition is met.

The next stage does need people, and it is the part worth reading carefully. The suspension lifts only if investors holding more than two thirds of commitments by value approve a replacement. If they do not, the investment period ends permanently and the fund manages what it already holds for the rest of its term. So one clause contains both an automatic consequence and a vote, in that order, and a reader who describes it as either one alone has described half of it. In this fund it has not been triggered.

Now the distinction that goes with it, and it is the second thing readers here get wrong. A key-person provision and a removal provision are not the same clause and do not behave alike. The key-person provision is automatic and it is about departure; a removal provision is a vote and it is about conduct or performance. This fund has both, written separately. There is no stated removal thresholdThe share of commitments a decision has to have behind it before it carries. for it, for the plain reason that this fund's own record does not fix one, and inventing a number that looks right is exactly how a wrong figure gets into circulation.

Step 5: for each party, what exactly can it stop?

Step 5 turns four columns of notes into a usable map, one sentence per party. Down the list, next to each name, goes the specific thing that party can halt. Where nothing can be written, the absence itself is recorded. An empty cell is a finding rather than a gap. A party with no stopping power is not a weak party, it is a party doing a different job, and mistaking the second for the first is how readers end up describing a fund's administrator as though it were a watchdog.

Here are the eight parties of Nilgiri Growth Partners Fund II, invented, numbered as its own record numbers them, with the stopping sentence written against each.

Read the eight and a shape appears that the vertical chart never showed. The party holding every rupee of asset value stops nothing about what is bought. The party paid the most stops whatever it never puts forward. The investment committee stops any investment. The investor advisory committee stops four named matters and no investment at all. And three parties hold no stopping power in this record whatsoever: the administrator, the independent valuation agent and the statutory auditor. Their job is to produce what everybody is told rather than to decide anything.

The three service parties deserve one more sentence. A reader who writes them off has lost something. Kolar Fund Services Private Limited, invented, strikes the net asset value. Palani Valuation Advisors LLP, an invented limited liability partnership (LLP), values every unrealised holding annually, and Rohit Vaz, invented, is its signing partner. The statutory auditor is named by role only here. None of them can stop an investment, and all three decide what the numbers on the record actually say. Deciding what the numbers say is a different kind of importance, and it belongs in a different column of the table.

Who holds the seven advisory committee seats, and were they given by size?

With a map in hand, the questions can start. The first one worth asking of any consenting body: who is on it, and how did they get there. Money is the tempting answer, the biggest cheques taking the seats. The temptation is that it sounds like how the world works. In this fund it is not what happened, and the record contains the counterexample directly.

The twelve investors of Nilgiri Growth Partners Fund II, invented, commit Rs 4,90,00,00,000 between them, and with the manager's own Rs 10,00,00,000 the fund is Rs 5,00,00,00,000. Seats on the investor advisory committee are held by investors 1, 2, 3, 4, 5, 6 and 8. Investor 7 commits Rs 25,00,00,000 and holds no seat. Investor 8 commits Rs 20,00,00,000 and holds one. A commitment schedule cannot reveal who sits on that advisory committee.

The arithmetic of representation is worth carrying, and it needs its denominator naming every time. The seven represented investors hold Rs 4,25,00,00,000 between them, or 86.7 per cent of investor commitments and 85.0 per cent of the whole fund including the manager's own money. The five unrepresented investors, being 7, 9, 10, 11 and 12, hold Rs 65,00,00,000, or 13.3 per cent of investor commitments. Rs 4,25,00,00,000 plus Rs 65,00,00,000 is Rs 4,90,00,00,000. The addition is the check that the two groups really are the whole of the investor base.

The seat count leads to the next question rather than to a verdict. A seat came from somewhere, and in this fund one of them came from a side letter: investor 1 has a seat written into its own side letter, alongside a most-favoured-nation right. The other six came from somewhere the record does not fix, and saying so is more useful than guessing. Whether seven is the right number, or these the right seven, is a separate judgement.

How is a consent right told apart from a veto and from a vote?

Answer this one before reading on. A prediction sticks where a read answer slides past.

Three words get used as though they meant the same thing, and they do not. A consent right is a named party having to say yes to one named matter, so refusing it stops that matter and leaves everything else untouched. A veto is the general word for a power to stop something single-handedly, and a consent right is therefore a veto over a very short list. A vote is a threshold across a body, where the holder can be outvoted, so holding a vote is not the same as holding the power to stop anything at all. The test that separates them in one move: whether a single refusal is enough, and if it is not, how much of the money has to agree with it.

Run the test on this fund. The investment committee refusing on its own stops an investment. The committee therefore holds a veto over investments and realisations. The investor advisory committee refusing on its own stops a conflict, a valuation policy, a first extension or a change of investment policy. The advisory committee therefore holds a veto over exactly four matters. A second extension is a vote, so no single investor stops it. And the same advisory committee that can refuse a valuation policy outright cannot touch one single carrying value produced under it. Nothing illustrates how narrow a consent right is more sharply.

One more line about the two committees. Their names are the whole problem. Both are called committees, both have members, both meet, and the two do entirely different jobs. Writing the number of members and the list of matters next to each of them at the first encounter keeps the confusion from forming.

How many investors does a threshold actually take?

Two of the rows in that grid ended in a share of the money rather than in a party, and a share of the money is the one thing in a governance section that can actually be computed. A threshold in this fund appears twice: the second extension needs investors holding more than half of commitments by value, and lifting a key-person suspension needs more than two thirds of commitments by value. Both come from this fund's own documents. Neither is set by anybody outside it.

Before any arithmetic, the trap that swallows most readings. A vote by value is not a vote by head. The phrase "by value" hides a second question, and the question is value of what. This fund has two possible bases: investor commitments of Rs 4,90,00,00,000, or total commitments of Rs 5,00,00,00,000 including the manager's own Rs 10,00,00,000. Two thirds of the first is Rs 3,26,66,66,667 and two thirds of the second is Rs 3,33,33,33,333. Name the base every single time. The two answers are not the same number, and a reader who quotes one while meaning the other has said something false with a correct-looking figure.

Now work it on the twelve. Take the four largest, being investors 1, 3, 2 and 4, in descending order of commitment. The four hold Rs 3,15,00,00,000, or 64.3 per cent of investor commitments and 63.0 per cent of total commitments. Both figures look close enough to two thirds to fool a fast reader. Neither is. The four fall short of two thirds of investor commitments by Rs 11,66,66,667 and short of two thirds of total commitments by Rs 18,33,33,333. The four largest investors in this fund cannot lift a key-person suspension on either base. Add investor 5 and the five largest hold Rs 3,65,00,00,000, which is 74.5 per cent and 73.0 per cent, and they clear it on both.

Now turn the same arithmetic upside down. Most treatments leave this half out. Take the ten smallest investors, meaning every one of the twelve except investor 1 at Rs 1,00,00,00,000 and investor 3 at Rs 80,00,00,000. Ten of the twelve is 83.3 per cent of the heads in the room. The ten hold Rs 3,10,00,00,000, or 63.3 per cent of investor commitments and 62.0 per cent of total commitments. The ten fall short of two thirds on either base as well. Five of twelve heads can carry that vote and ten of twelve cannot, at the same time, in the same fund.

One naming trap sits inside that sentence and it catches people every time. The ten smallest are not investors 2 through 12. Investor 3 is the second largest of the twelve. The numbering in this record is an identifier, not a ranking, and reading it as a ranking produces a group that holds a different amount of money from the group intended.

Two more readings fall straight out of the same twelve, and both of them are the kind of thing a finished map lets a reader ask. First, a simple majority. Investors 1, 3 and 2 together hold Rs 2,55,00,00,000, or 52.04 per cent of investor commitments and 51.0 per cent of total commitments. Three of twelve heads therefore carry any decision needing more than half of the value, including the second extension of this fund's term. Second, blocking. To stop something needing two thirds, more than one third has to be against it. More than one third is Rs 1,63,33,33,333 of investor commitments or Rs 1,66,66,66,667 of total commitments. Investor 1, the largest, holds Rs 1,00,00,00,000 and cannot block on either base. Investors 1 and 3 together hold Rs 1,80,00,00,000 and can, on either base. Investors 1 and 4 together hold Rs 1,60,00,00,000 and cannot.

The blocking arithmetic is the whole reason the step exists. Nothing about who can block was visible on the chart, in the party list, or in the sentence "a key-person replacement needs two thirds by value". Blocking power appears only once the threshold and the commitment schedule are set next to each other. Setting the two side by side takes about four minutes and answers a question nobody could otherwise answer.

How would somebody actually use this map in one working hour?

A whole set of fund documents can be mapped in about one working hour, and the five steps are the hour. The work is the same whether the question is what a manager may do without asking anybody, or what a single investor could have refused.

The first fifteen minutes go on steps 1 and 2 and produce four names: who holds the assets, who decides, who approves what is decided, and who stands behind the decider. The next fifteen go on step 3 and copy out the consent list word for word. A paraphrase of a consent list is where the length quietly changes. Ten minutes go on step 4, hunting for anything written as a condition rather than as a decision. Those clauses read like background and behave like triggers. Fifteen go on step 5, writing one stopping sentence against each party. The last five are for the threshold arithmetic, the only part of the whole exercise that produces a number.

Then the map gets used, and the use is always the same shape: one specific worry, walked across it. Suppose the worry is that the manager might sell a holding to a vehicle it also manages. On this fund's map that lands on the conflicts row, so the investor advisory committee consents, so refusal by that committee stops it, so the next question is who sits on that advisory committee, and the map already answers it: seven seats, investors 1, 2, 3, 4, 5, 6 and 8, holding 86.7 per cent of investor commitments between them. Four questions and about ninety seconds, and the reason it is that fast is that the map was built before the worry arrived rather than after.

Two habits make the difference between a map that works and a list that does not. The first is naming which committee every single time the word appears. No sentence produced can then be read the wrong way later. The second is writing down what the documents do not say. In this fund the record fixes no removal threshold and does not say how six of the seven advisory committee seats were allocated, and both of those absences are findings worth handing on. An empty cell that is marked empty is information. An empty cell that got quietly filled with a plausible guess is the beginning of a wrong figure with a long life.