NAV Financing: Borrowing Against the Fund's Own Holdings

A loan against a fund's own holdings is money borrowed on the security of the whole remaining portfolio rather than any single company in it, and the proceeds are then paid out to investors. The loan moves cash forward and creates nothing. The same holdings repay it, with interest, so the eventual total reaching investors is lower rather than higher.

Start with something picturable without any of the vocabulary. A household has a locker of gold it has held for twenty years. Nobody has bought that gold. The only figure attached to it is what a jeweller said he thought it would fetch. Now the household takes a loan against the locker, and cash arrives in the bank account this week. Nothing about the gold has changed. The household is not richer by a single gram. Value that used to sit entirely in the locker now sits partly in the bank account, and the locker carries a charge that has to be cleared before the gold is fully theirs again.

A loan against a fund's own holdings is that locker at scale, and every rupee of arithmetic below is that one sentence made specific. The differences are scale, the fact that the thing in the locker is five unlisted businesses rather than a set of bangles, and the fact that the person watching the bank account is not one household but twelve investors reading a quarterly statement they did not write.

What is a loan taken against a fund's own holdings, and what secures it?

A fund that has been running for years usually reaches a stretch where it has sold some of what it bought and still holds the rest. The unsold part carries a reported value. The reported value is an estimate of what businesses nobody has bought would fetch, struck on a timetable, by people whose method is covered separately. Exactly one thing is needed from it here: it is an estimate of businesses nobody has sold. Everything below follows from borrowing against an estimate and repaying out of the sales that will eventually test it.

Net asset value (NAV) financingBorrowing secured on a fund's remaining holdings taken together rather than on any single one of them. is a loan made to the fund itself, and the securityThe assets a lender can look to if the borrower does not repay. is the remaining portfolio taken as one pool. Not one company. Not the best company. Not the company most likely to be sold next. All of what is left, together, behind one loan. Because its claim sits across all of them, the lender is repaid out of whichever of those businesses produces cash first, and it does not much care which one that turns out to be.

Nilgiri Growth Partners Fund II, invented, is the fund every figure below belongs to. At its record date, the end of its Year 9 Quarter 2, it had drawn Rs 4,80,00,00,000 from its investors, returned Rs 4,38,00,00,000 to them in four cash payments, and still held five businesses reported at Rs 2,82,00,00,000 taken together. Four of the nine things it bought are gone entirely, three sold and one written off in full, and one position sits on both sides because part of it was sold and part is still held. The fund is therefore Rs 42,00,00,000 short of having returned the capital it called, and that gap is the number everything below turns on.

Now the sentence that matters most. Nilgiri Growth Partners Fund II has not borrowed anything. It has no loan, no lender and no charge over anything. A counterfactualSomething that did not happen, worked out in full to show what it would have done. is an arrangement that has not been done, worked through to the rupee to show what it would have moved. Every figure below is one.

Why is the security the whole remainder rather than one company?

Ask a lender what it is actually looking at, and the answer changes the transaction completely. Picture two shopkeepers on the same street who each want to borrow. The first pledges the stock in one shop. If that shop burns down, the lender has nothing. The second pledges the stock in all four shops he runs, and the lender is repaid out of whichever shop sells goods fastest. The second lender is not being generous. The second lender is looking at a spread of outcomes instead of a single one, and prices and structures the loan for the spread.

Lending against a pool of five businesses is a different transaction from lending against one of them, and the difference is not a matter of degree. A loan secured on one named company is repaid by that company and by nothing else, and it touches only the investors' exposure to that one name. A loan secured on the remaining portfolio taken together sits above every one of the five, is repaid out of whichever realises first, and lands on every investor's reported numbers at the same moment because every investor has a share of all five.

There is a second consequence that a reader meets less often. When the security is the whole remainder, the lender's claim is ahead of the investors on all five, not on one. Before an investor sees another rupee of the value of holding 4, holding 6, holding 7, holding 8 or the retained part of holding 9, the loan has to be cleared. In the gold locker example, the whole locker is charged and not one bangle. The household cannot quietly sell the two bangles it likes least and keep the cash.

How big is the loan here, and where does that number come from?

The size has to be picked before anything can be worked, and the choice is stated in the open. The loan worked below is a hypothetical 20.0 per cent of the reported value, and 20.0 per cent is a figure chosen for the illustration and nothing else. It is not a market convention, not a limit set by anybody, and not a figure this fund's record contains. Twenty per cent is a round number that makes the arithmetic legible. Because the relationship is a straight line, every conclusion below survives at any other size, and the control further down draws that line.

Twenty per cent of Rs 2,82,00,00,000 is Rs 56,40,00,000. Work it yourself: one tenth of Rs 2,82,00,00,000 is Rs 28,20,00,000, and twice that is Rs 56,40,00,000. Rs 56,40,00,000 is the whole of the loan in this counterfactual, and the proceeds are paid straight out to the investors in the quarter after they are drawn, the way this fund pays out everything else.

Hold on to a comparison while the arithmetic runs. The fund is Rs 42,00,00,000 short of returning the capital it called. The loan is Rs 56,40,00,000. So the borrowing is larger than the gap by Rs 14,40,00,000, and that spare amount is the reason the reported figure does not merely reach the line but steps over it.

What does it do to distributions against capital paid in?

Two figures describe where a fund stands, and both of them have the same denominator. Distributions against capital paid inCumulative cash returned to investors divided by cumulative cash they have put in. is cash actually returned divided by cash actually called. Residual value against capital paid inThe reported value of what is still held divided by cumulative cash put in. is the reported value of what is still held, divided by the same cash actually called. The denominator in both is Rs 4,80,00,00,000, the amount this fund has drawn from its investors to the record date.

Before any borrowing, distributions against capital paid in are Rs 4,38,00,00,000 divided by Rs 4,80,00,00,000. The result is 0.9125, written 0.91. Checked the slow way: Rs 4,80,00,00,000 less nine tenths of itself, Rs 4,32,00,00,000, leaves Rs 48,00,00,000, and Rs 6,00,00,000 of that is 0.0125 of the whole, so 0.90 plus 0.0125 is 0.9125. Every rupee of the Rs 4,38,00,00,000 has gone to giving investors their capital back, and the fund has not finished doing that.

Now add the loan. The fund borrows Rs 56,40,00,000 and pays it out. Cumulative distributions become Rs 4,38,00,00,000 plus Rs 56,40,00,000, or Rs 4,94,40,00,000. Divided by the same Rs 4,80,00,00,000 of capital paid in, that is 1.0300 exactly. Not approximately. Rs 4,80,00,00,000 times 1.03 is Rs 4,94,40,00,000, so the figure is exact to four decimal places.

Read what just happened in plain words. The reported figure crossed 1.00. At 1.00 the cash returned equals the cash called, and past that point everything is arithmetic the investor has not paid for yet. A reader watches for that crossing. The line was crossed, and not one of the five businesses had been sold to anybody. No buyer looked at holding 4. No auction ran for holding 8. Nothing was tested by anybody outside the fund.

What happens to the other column at the same moment?

The second column is the half a headline leaves out, and it is not a subtlety. The Rs 56,40,00,000 did not appear from nowhere. The five businesses are what the loan is secured on and what will have to repay it, so the Rs 56,40,00,000 came out of them. So the reported value of what is still held, seen from the investors' side, now carries a liability of Rs 56,40,00,000 sitting ahead of them.

Rs 2,82,00,00,000 less Rs 56,40,00,000 is Rs 2,25,60,00,000. Divided by the same Rs 4,80,00,00,000 of capital paid in, that is 0.4700 exactly. Before the loan the same figure was Rs 2,82,00,00,000 over Rs 4,80,00,00,000, or 0.5875, written 0.59. So residual value against capital paid in fell from 0.5875 to 0.4700. The fall of 0.1175 is exactly the amount the other column rose by.

Put the two movements side by side and the whole thing is visible in one line. Distributions against capital paid in went up by 0.1175, from 0.9125 to 1.0300. Residual value against capital paid in went down by 0.1175, from 0.5875 to 0.4700. One column took what the other column gave up, to the fourth decimal place, on the same day.

So what is the number that does not move?

Total value against capital paid inThe two figures above added together: cash returned plus reported value still held, over cash put in. is the two added together, and this fund's own record says the addition is the check to run. Before the loan: 0.9125 plus 0.5875 is 1.5000. After the loan: 1.0300 plus 0.4700 is 1.5000. The total did not move by a single basis point on the day the money arrived, and that single fact is the honest reading of the whole arrangement.

Why the total cannot move is worth saying, rather than merely noting that it did not. Total value against capital paid in is the total of two things over one denominator. The denominator is capital paid in, and the borrowing does not touch it: the fund called Rs 4,80,00,00,000 before and it has called Rs 4,80,00,00,000 after. The numerator is cash returned plus value still held. The loan adds Rs 56,40,00,000 to the first and takes Rs 56,40,00,000 off the second. A number plus and minus the same number is that number. There is no clever step in this and there is nothing hidden.

Where does the reported column actually cross 1.00?

The relationship between the size of the borrowing and the reported column is a straight line, and it is worth seeing why. Distributions against capital paid in equal Rs 4,38,00,00,000 plus the borrowing, all over Rs 4,80,00,00,000. The denominator never changes and the numerator moves rupee for rupee with the loan, so the figure rises in a straight line with no curve anywhere in it.

Setting that expression equal to 1.00 and solving requires cumulative distributions of exactly Rs 4,80,00,00,000, and the fund has already paid Rs 4,38,00,00,000, so the amount to be borrowed is Rs 42,00,00,000. Reaching 1.00 means having paid back precisely what was called, so the crossing point is not a coincidence and not a choice: it is exactly the shortfall. That is what return of capitalThe point at which investors have been paid back every rupee the fund has called from them. means as an arithmetic rather than as a phrase.

As a share of the reported value, Rs 42,00,00,000 over Rs 2,82,00,00,000 is 14.893617 per cent, written 14.9 per cent. So a borrowing of 14.9 per cent of the reported value takes the column to 1.00, and the illustration's 20.0 per cent takes it to 1.03. The difference between those two, Rs 14,40,00,000, is the whole of the headroom above the line.

Who repays the loan, and out of what?

Here is the sentence that finishes the arithmetic, and it is the sentence a summary tends to drop. The loan is repaid out of the same five businesses, with interest, before the investors see another rupee of their value. There is no other source. The fund is past its investment period, it has no new money coming in from anywhere, and the only cash it will ever produce again is the cash from selling holding 4, holding 6, holding 7, holding 8 and the retained part of holding 9.

Go back to the gold locker. The household did not acquire anything by pledging the locker. When the loan comes due it either sells some gold to clear it or finds the money elsewhere, and there is no elsewhere. So the gold that eventually leaves the locker leaves partly to repay a loan and partly to reach the household, and the household ends up with less gold and less cash than if it had simply waited and sold when it wanted to. The only thing it bought was time.

Between a proposal and a distribution funded by borrowing there is a process, and an investor typically sees only one step of it. The proposal is put by the manager. An independent valuation of what is being pledged is done. The investor advisory committee is asked to consent. That committee consents on conflicts rather than approving investments. The two are different rooms, and confusing them is the most common mistake made here. The loan is then drawn. The cash is distributed. And the repayment sits somewhere in a future that has not happened yet.

Why is the eventual total lower, and why does nobody yet know by how much?

On day one the total does not move. From day two it starts to fall, and the reason is a single asymmetry worth stating carefully. Interest builds up on the loan, and no interest builds up on an estimate. The Rs 2,82,00,00,000 is a figure struck for five businesses; it does not grow because a lender is waiting. The loan does grow, in the sense that a charge accruesInterest or a charge building up over time whether or not anything has been paid. against it whether or not anything is paid across in any given quarter. So each quarter that passes, a little more of what those five businesses eventually fetch has already been spoken for.

The invented record behind this fund fixes no cost of borrowing, and inventing one would turn an illustration into a claim about what such a loan costs somewhere. Arithmetic on an assumed amount can still be done honestly. Suppose the loan cost Rs 30,00,00,000 in interest and charges over whatever period it ran, and suppose the five businesses realised exactly the Rs 2,82,00,00,000 they were reported at. Then the total cash reaching investors across the fund's whole life is Rs 7,20,00,00,000 less Rs 30,00,00,000, being Rs 6,90,00,00,000, and against Rs 4,80,00,00,000 of capital paid in that is 1.4375 rather than 1.5000.

Every row of that table assumes the five businesses realise exactly their reported value, and that is the one assumption nobody can stand behind. The rows illustrate an amount rather than any rate. The general rule underneath them is simple enough to carry away. Rs 1,00,00,000 divided by Rs 4,80,00,00,000 is 0.002083, so each Rs 1,00,00,000 of interest takes 0.0021 off total value against capital paid in.

Now the harder half. Nobody knows by how much, and the reason is that nobody knows what the five will fetch. Who actually bears the interest depends entirely on that, and the three branches have three different answers.

Notice how the three branches change who is talking about what. In branch one, a reader who looked only at the reported column would have been directionally right and would still have been paid less than they would have been without the loan. In branch two, the reader was simply wrong. In branch three, the loan has changed the order of who gets paid, and the party ahead of the investors is a lender who was not there before. The arrangement does not create value in any branch, and in one branch it changes who bears the loss.

What is the difference between an arithmetic and a record?

A record says what happened, and an arithmetic says only what would have happened. The distinction is the easiest one here to lose. Nilgiri Growth Partners Fund II, invented, made four distributions to its record date, adding to Rs 4,38,00,00,000, and every one of them was the proceeds of something a buyer paid for. Rs 63,00,00,000 came from selling the whole of one holding to another fund. Rs 2,03,00,00,000 came from selling a holding to a buyer in the same industry. Rs 1,50,00,00,000 came from an offering followed by a later sell-down. Rs 22,00,00,000 came from selling 40 per cent of a position. There is no fifth distribution and there is no lender.

An arithmetic this clean is easy to mistake for a record of events. A counterfactual earns its place by showing what a decision would have moved before anybody takes it, and it loses its value the moment a reader remembers the numbers and forgets that nothing was done.

What does somebody actually do with this when a proposal lands?

A proposal on paper is where the mechanism becomes something a reader can apply. Three different people read the same proposal and each of them reaches for a different number.

A lender reads the pool and not the story. It is looking at five businesses, not one, and its question is how many independent ways cash could arrive and how soon the earliest of them might. In the counterfactual worked here it would be looking at reported values of Rs 1,08,00,00,000, Rs 21,00,00,000, Rs 39,00,00,000, Rs 81,00,00,000 and Rs 33,00,00,000, and asking what each of those figures rests on and who struck it. The concentration in that list is the first thing anybody notices: holding 4 alone is 38.3 per cent of the Rs 2,82,00,00,000 pool and holding 6 is 7.4 per cent, so the spread is a good deal narrower than five names suggests.

An analyst on the investor's side recomputes both columns before reading a word of the covering letter. The recipe is short. Take cumulative distributions and divide by cumulative capital called. Take reported value of what is still held and divide by the same cumulative capital called. Add them. If the total is where it was last quarter and the first figure has moved, cash has been moved forward rather than created, and the next question is where it came from. Recomputing both columns is also the one check that survives with no access to anything the fund has not already published.

An investor asks four questions and stops. What secures the loan, meaning is it the whole remainder or one name. Out of what will it be repaid, and by when. What does the whole arrangement cost, and over what period, so the eventual total can be estimated. And what has to be true about the five businesses for the arrangement to have cost nothing, a question branch one and branch two above answer in two different ways. A reader who asks those four has understood the arrangement whether or not they remember a single figure of it.

And for a household, the same discipline in one line: when money arrives, ask whether somebody bought something or whether somebody lent something. The bank statement looks identical. The next twelve months do not.