Valuation Lag: Why Private Marks Move After Public Ones

A public price absorbs information as it arrives. A private mark is built from information that reaches it quarterly and annually. So a private carrying value changes later, and it changes in both directions. Holding 3 of Nilgiri Growth Partners Fund II, invented, was carried at 2.00 times cost at Fund II's Year 7 year end and sold at 2.50 times one quarter afterwards.

Start with something visible from an ordinary window. Suppose a plot of land sits at the end of a street, and everybody in the neighbourhood has a settled view of what it is worth. The neighbourhood's view updates roughly never. Then one morning a plot four doors down changes hands, and by the evening every conversation on the street has moved. Nothing physical happened to the first plot. No wall fell down, no road was widened, no new school opened. A transaction is what produces the number, and transactions arrive rarely. So a figure that had been sitting still for two years moved a long way in one day.

Now put a listed share beside that plot. The share has a price at eleven in the morning and a different one at half past eleven, and neither of those prices waited for anybody to decide anything. Neither the plot nor the share is valued better than the other. Information simply reaches the two of them on entirely different timetables. Carried into a private fund's reporting pack, that difference in timetables is valuation lag.

Why does a private mark move later than a public price does?

Because of two clocks that were never set to each other. A listed instrument sits inside a machine that turns every new fact into a price within the day, and often within the minute. Somebody somewhere is willing to act on the fact, and the act itself is the price. Nobody has to convene, nobody has to sign, and nothing has to be circulated for review.

An unlisted holding sits inside a different machine, and that machine runs on a calendar. In Nilgiri Growth Partners Fund II an independent valuation agent values every unrealised holding once a year, the manager makes the mark in the quarters between those annual valuations, and the administrator takes those marks and strikes the fund's net asset value. Three parties, one hand-off, and the timetable those hand-offs run on is fixed by the arrangement rather than by events. Who each of those parties is, what independent means when it is applied to one of them, and what each is contracted to do, are settled separately and are used here without being re-explained.

So consider a fact that comes into existence in the middle of a quarter. A competitor announces a price cut. A large customer renews for three years. A regulator opens a consultation. For a listed instrument that fact is absorbed the same afternoon. For an unlisted holding the fact simply waits, not because anybody is hiding it and not because anybody is being slow, but because the next occasion on which a number gets made is the next quarter end. The distance between the day a fact exists and the day a private mark absorbs it is the entire mechanism, and no part of that distance is anybody's misconduct.

What is valuation lag, and what is it not?

Valuation lagThe gap between new information existing and a private mark absorbing it. is the gap between a fact existing and a private mark absorbing it. The definition carries no judgement in it. Lag is not a measure of how good anybody's valuation work is, and not a signal about the holding. The timetable sets its size, not the effort, so lag does not get smaller when people try harder.

Precision about what is being compared matters here. A loose version of this idea does a lot of damage. The lag is not the gap between a private mark and some true value that exists somewhere and is being missed. There is no such number sitting in a drawer. The lag is the gap in time between two moments: the moment a fact comes into existence and the moment a figure produced on a calendar takes account of it. Lag is a property of a timetable, not a verdict on a number.

Two consequences follow immediately, and both matter more than they look. The first is that the lag exists even when every party in the chain does everything right. The second is that the lag has no direction built into it. A fact that arrives mid quarter might be good for the holding or bad for it, and the calendar does not care which. Readers reliably get the second consequence wrong, and the two worked cases below take it in both directions.

What happened when holding 3 was sold one quarter after it was marked?

Holding 3 of Nilgiri Growth Partners Fund II is Tungabhadra Logistics Private Limited, an invented road freight operator. The fund entered it in Fund II's Year 2 Q2 at a cost of Rs 60,00,00,000. At Fund II's Year 7 year end it was carried at 2.00 times that cost, a carrying valueThe value a holding is recorded at, which is an estimate rather than a price paid. of Rs 1,20,00,00,000. Two times sixty is one hundred and twenty, and that is the whole of the arithmetic behind the reported figure.

In Fund II's Year 8 Q1, one quarter later, the position was realised through an initial public offering followed by a sell-down after the lock-in ended. Total proceeds were Rs 1,50,00,00,000, being 2.50 times the Rs 60,00,00,000 of cost. The realisationThe sale of a holding, turning a carrying value into a receipt. moved the recorded value by Rs 1,50,00,00,000 less Rs 1,20,00,00,000, a difference of Rs 30,00,00,000. As a share of the carrying value that is 30 over 120, being 25.0 per cent exactly. As a share of the cost it is half a turn: 2.50 times against 2.00 times.

Half a turn of cost appeared in the record of Nilgiri Growth Partners Fund II between Fund II's Year 7 year end and its Year 8 Q1, and nothing about the underlying business changed on the day of the sale. The lorries were the same lorries, the customers were the same customers and the depots were in the same places. The machine producing the number changed. Up to that day the number was an estimate made on a calendar. On that day it became a receipt.

What happened when holding 6 was carried at cost while it deteriorated?

Now the other direction, on the same fund, four years earlier. Holding 6 of Nilgiri Growth Partners Fund II is Vaigai Edutech Private Limited, an invented education company, entered in Fund II's Year 3 Q3 at a cost of Rs 30,00,00,000. Through Fund II's Year 5 it was carried at that cost. At Fund II's Year 6 it was written to 0.70 times cost, being Rs 21,00,00,000, and by then the deterioration in the business had been visible for three quarters.

The write-downA reduction in the value a holding is carried at, with no sale involved. is Rs 30,00,00,000 less Rs 21,00,00,000, or Rs 9,00,00,000. Against the Rs 30,00,00,000 it had been carried at, that is 9 over 30, being 30.0 per cent exactly. Notice how differently this one arrives. Nothing was sold. No cash moved. A figure that had sat still for a long stretch simply became a different figure at a year end, and the fact that produced it had been in the room for three quarters before it reached the report.

One sentence here catches almost every reader the first time, and it is worth pausing on. Carrying a holding at what it cost is not the absence of a valuation. Marked at costCarried at what was paid for it, which is a valuation decision and not the absence of one. is a conclusion: on the evidence available at the valuation date, the price paid remains the best estimate of what the position is worth. A holding carried at cost has been valued, and the later write-down of Rs 9,00,00,000 on holding 6 of Nilgiri Growth Partners Fund II at its Year 6 year end is therefore a change of view rather than the first appearance of one. Think of somebody who keeps saying their scooter is worth what they paid for it. The scooter owner is not staying silent about the value. The owner is making a claim about it, and a claim can be wrong in exactly the way any other claim can.

How does a single mark reach a figure the fund has already published?

How Valuation Lag Can Affect Alternative-Investment Reporting

Here is the join that makes the lag matter to a reader rather than only to a valuer. A private fund's net asset value is not an opinion about the fund. The net asset value is a sum of the carrying values of the holdings still held, one line added to the next. And the headline multiple a reader is handed sits directly on top of that sum. Total value to paid in is cumulative distributions plus net asset value, all divided by cumulative capital paid in. Two of those three quantities are cash that has already moved and cannot be argued with. The third is a stack of estimates.

So follow the chain in one direction. One holding's mark changes. The net asset value is the sum, so it changes by exactly the same number of rupees. Paid in is the denominator, and it does not move when a mark moves. So the published multiple changes by that same number of rupees divided by capital paid in. The effect of one holding's mark on a whole reported multiple is the rupee change in that mark divided by cumulative capital paid in, and there is no other route by which it travels.

The identity settles in advance how much a lag can possibly matter on a given fund, and it is worth holding on to. A large fund with a small holding will barely notice. A fund whose largest holding is a third of its net asset value will notice a great deal. At Fund II's Year 7 year end, holding 3 was carried at Rs 1,20,00,00,000 against a fund net asset value of Rs 3,59,50,00,000. Holding 3 alone was 33.4 per cent of everything the fund then said it held. One holding of six, and a third of the estimate.

What would the Year 7 multiple have read, had holding 3 carried what it fetched?

Now the arithmetic in full, and it is a counterfactualA worked case of what a figure would have been under an assumption that was not true. from beginning to end. Nilgiri Growth Partners Fund II actually published total value to paid in of 1.33 times for Fund II's Year 7 year end, built from Rs 2,66,00,00,000 of cumulative distributions plus Rs 3,59,50,00,000 of net asset value, over Rs 4,71,50,00,000 of cumulative capital paid in. Six hundred and twenty five crore fifty lakh over four hundred and seventy one crore fifty lakh is 1.3266, written 1.33.

Holding 3 was Rs 1,20,00,00,000 of that net asset value, leaving Rs 2,39,50,00,000 across the other five holdings then held. The Rs 2,39,50,00,000 is derived by subtraction rather than being a figure the record states on its own. Had holding 3 instead carried the Rs 1,50,00,00,000 it fetched one quarter afterwards, the Year 7 net asset value would have been Rs 3,89,50,00,000 and the published multiple would have read 1.39 times. Six hundred and fifty five crore fifty lakh over four hundred and seventy one crore fifty lakh is 1.3902.

The published Year 7 figure for Nilgiri Growth Partners Fund II and its sale-informed counterfactual are 1.33 times and 1.39 times. A completed sale superseded that mark three months later, and the two figures differ by 0.06 times. And that 0.06 is exactly the identity from the block above doing its work: Rs 30,00,00,000 of movement divided by Rs 4,71,50,00,000 of capital paid in is 0.0636, which is the whole of the gap. The counterfactual figure was never reported by anybody and is not a corrected version of anything. The counterfactual demonstrates how much of a headline multiple can hang on one line made on a calendar.

What does the Year 5 counterfactual do that the Year 7 one cannot?

The Year 5 counterfactual moves the published figure the other way, and that is its entire job. At Fund II's Year 5 year end, Nilgiri Growth Partners Fund II reported total value to paid in of 0.95 times, being Rs 4,34,00,00,000 of net asset value over Rs 4,55,50,00,000 of cumulative capital paid in, with nothing distributed to anybody at that date. Four hundred and thirty four over four hundred and fifty five crore fifty lakh is 0.9528.

Holding 6 sat inside that Rs 4,34,00,00,000 at its Rs 30,00,00,000 of cost. One year afterwards holding 6 received a mark of 0.70 times cost. Had it carried that mark at Fund II's Year 5 year end, the Year 5 net asset value would have been Rs 4,25,00,00,000 and the published multiple would have read 0.93 times. Four hundred and twenty five over four hundred and fifty five crore fifty lakh is 0.9330. The 0.93 times is a counterfactual too, was never reported by anybody, and is not a corrected figure.

The move is 0.02 times, against 0.06 times four years later, and the difference in size is not the point. One lag pushed a published figure for Nilgiri Growth Partners Fund II down at Fund II's Year 7 year end, and the other pushed a published figure for the same fund up at its Year 5 year end. The pair is the teaching, not either one alone. An account carrying only holding 3 would leave a tidy and false rule behind: private marks lag on the low side, so a reader should mentally add something. Holding 6 is the counterexample, on the same fund, from the same reporting machinery.

What does the lag not establish?

Two things, and both are worth stating rather than leaving to be inferred. The first: valuation lag does not make private marks wrong. A mark made at a valuation date on the evidence then available is a mark made at a valuation date on the evidence then available. A mark can be superseded by later evidence, as holding 3 of Nilgiri Growth Partners Fund II was superseded by its own Year 8 Q1 sale, and being superseded is not the same as having been wrong at the time.

The second sounds sophisticated, and is the more tempting for it. The lag does not make private returns look smoother in a way that is an advantage. Whether a reported series moves less from one quarter to the next than some other series is a fact about a series. Turning that into a benefit requires a claim about what smoothness is worth to somebody, and there is no basis anywhere here for that claim. Naming what a mechanism does is teaching. Naming what a mechanism is worth is not.

Underneath both sits one structural fact, and it is the reason the pair of worked cases above are drawn on the same scale and in the same ink. The lag has no sign: on this one fund it ran below on holding 3 and above on holding 6. Any adjustment for it would need to know each holding's direction in advance, and that is exactly the information nobody has. Colouring the upward case green and the downward one red quietly delivers a verdict the arithmetic does not support.

What does a reader actually do about a lag nobody can remove?

Read one line before reading any number. The valuation dateThe day a figure is as at, which is often earlier than the day a reader receives it. is the day the figures beneath it are as at. It is almost never the day the pack reached the reader. For Nilgiri Growth Partners Fund II that date is the end of Fund II's Year 9 Quarter 2, and nothing was revalued in the two quarters after Fund II's Year 8, so the Rs 2,82,00,00,000 of carrying value is the Year 8 year-end mark carried forward. Both of those sentences are in the record. Neither of them is in the headline.

The valuation date buys the only part of the lag that is actually visible: the distance between the day the numbers were made and today. Nothing settles which way an individual mark will move, and the date does not either. The date states how much world has gone by. Everything else follows from it. If a great deal has happened since the valuation date, a reader knows that the figures are describing an earlier state of affairs, and can say so out loud instead of pretending otherwise or quietly inventing a replacement.

Three more questions belong to the same one minute of reading, and none of them requires any arithmetic. Who made the mark, and when did the independent one last happen? Is any figure in the report a valuation the fund itself made rather than one somebody else made? And how much of the stated value has actually been sold to anybody? For Nilgiri Growth Partners Fund II at the end of Fund II's Year 9 Quarter 2 that is Rs 4,38,00,00,000 realised in cash against Rs 2,82,00,00,000 never sold, or 39.2 per cent of the Rs 7,20,00,00,000 of total value. The practical handling of a lag is a date and three questions, not an adjustment. The adjustment is the one move the evidence never supports.