Portfolio Holdings: Where the Line-by-Line List Comes From

Nobody writes a scheme holdings list. A holdings list is assembled: the custody record supplies each quantity, the valuation supplies the amount each thing is carried at, and the two halves are joined into one row. Add cash, add the amounts owed to the scheme, take off the amounts it owes, and the total is the net assets that the unit count divides into the published value per unit.

The assembly is the operational answer, and it differs from the answer that follows from asking what a disclosure means or what a scheme is obliged to publish. Reading a published list, and the obligation to publish one at all, are both covered separately. What follows stands on the production side of the counter: which record every column is drawn off, what has to be true before a single line can go out, and how the operations chain gets from a custody position to a printed row.

Everything computed below stands on one made up record. Girnar Asset Management Limited, an invented asset manager, runs an open ended equity scheme called the Girnar Large Cap Equity Fund, and the record used here states it at net assetsThe scheme total once everything it is liable for has been taken off, which is the figure the published value per unit is struck from. of Rs 4,200 crore against 1,20,00,00,000 units in issue, being 120.00 crore of them. Put the first over the second, exactly as the price sequence settled, and one unit works out at Rs 35.00 on the nose. Behind it sit 3,80,000 folios, and the ratio charged against those assets is 1.65 per cent. The portfolio is managed by Kalyani Bhagat; operations are headed by Sohail Merchant.

One more fact about the record has to land before the mechanism starts. The record used here carries no holdings at all for this scheme: no security, no issuer, no quantity, no weight, no rating and no sector split. The record states net assets, units, folios and a ratio, and it stops there. So the shape of a holdings list gets drawn below and not one cell of it gets filled. The absence is worked into the teaching: by the end, any discrepancy in any such list can be sized without knowing a single thing the scheme holds.

What is actually in one line of a scheme holdings list?

Four things, and the useful way to describe them is not by what they are called but by where each one is drawn from. There is something that identifies the thing held. There is a quantity of it. There is an amount that thing is stated at in the books. And there is that amount expressed as a share of the scheme total. Two of those four are lifted off records kept elsewhere, one is arithmetic, and the fourth is a label.

Here is the thing most readers never notice. The quantity on a line and the value on the same line have two different parents, and a reader who knows that reads the list differently from one who does not. The quantity is drawn off a record kept by the party that physically holds the securities. The amount is produced by a method applied by the operations team at the asset manager. The quantity and the value arrive at the row from opposite directions and meet in the middle of it. The printed row gives no sign of it whatever, and that silence is the reason this has to be pointed out rather than noticed.

Think of a warehouse for a moment. A godown keeper counts the sacks and signs a stock sheet: forty sacks, sixty sacks, whatever is actually stacked there. The trader in the office puts a rate against each sack to work out the value of the stock. Two people, two records, one statement of stock value at the end. If the same person counted the sacks and set the rate and printed the statement, nobody outside could ever tell whether the statement described the godown or described a wish. A scheme holdings listThe line by line statement of what a scheme held at one stated instant, carrying a quantity and an amount against each thing held. is that stock sheet, with the counting and the rating deliberately kept in separate hands.

Which record does the quantity on a line come off?

The custody recordThe running record of securities held for a scheme, kept by the party holding them rather than by the party putting values against them., kept by the custodian. The custodian is a role rather than a named firm. The custodian holds the securities for the scheme and keeps the running account of what is there. Every line in that account is a positionOne entry in a securities record: a single thing held, together with how much of it is held, at a stated moment.: one thing held, and how much of it. Where a holding sits in dematerialised form, a depository layer stands behind the custody record, and there are two of those in India: the National Securities Depository Limited (NSDL), at nsdl.co.in, and the Central Depository Services Limited (CDSL), at cdslindia.com. Both are named so that the layer is visible; their workings are covered separately.

The operations team does not invent the quantity. It reads it. Reading the quantity rather than inventing it is the whole reason a holdings list is worth anything at all. The practice makes the quantity column a record rather than a claim. Turn it around and see how bad the alternative is: if the asset manager both held the securities and stated how much of each it held, there would be no independent thing left in the world to test the statement against. Such a list would not be a false one. It would be an untestable one. For a reader on the outside, untestable is the same problem wearing better clothes.

The separation matters; who fills either seat does not. The moment a firm name is attached to the custodian, a reader starts thinking about that firm and stops thinking about why the seat exists.

Which record does the carrying value come off?

The valuation, applied under a written policy that is covered separately and used here rather than explained again. A carrying valueThe amount a thing held is stated at in the books on a given date, reached by applying a written method rather than by asking anybody. is not observed the way a quantity is observed. A carrying value is produced. Somebody applies a stated method to a stated input and the method returns a number.

Which means the two halves of a row carry two completely different kinds of certainty, and this is worth slowing down for. A quantity either agrees with the custody record or it does not, and that is a yes or no question with an answer sitting in a file somewhere. A carrying value is right if the method was the right method and if the method was actually followed. Those are two conditions, and neither of them is a simple lookup. One printed row therefore mixes a fact that can be matched against another record with a figure that can only be judged against a procedure, and the row does not mark which half is which.

Every row looks equally solid in print. Equal solidity is a property of printing, not a property of the underlying certainty. A line whose value came straight off an active market and a line whose value came out of a policy applied to a thing that did not move that day are typeset identically. Nothing in the layout is deceptive; the layout simply has no way of showing the difference, and knowing that is part of reading operations output properly.

What has to be true before the list can be published at all?

A short chain of conditions, and each one of them is somebody in operations signing off that a thing is settled. The chain of conditions is the part of the subject that lives nowhere else: not what the list says, not that it must exist, but what has to have happened first.

A list that describes different holdings at different moments describes nothing, so the positions have to be fixed as at one stated instant. The value has to be put against every fixed position under the written policy, with the policy applied rather than improvised. None of the following lives in the custody record, so all of it has to be pulled in from the scheme books: cash, the amounts the scheme is entitled to and has not received, and the amounts it has become liable for and has not paid. Then the whole thing has to be added and set against the net assets figure the value per unit was struck from. Only when those four are done does a publishable list exist, and until they are done what operations is holding is a working file rather than a statement of position.

The chain is not a schedule. The chain fixes the order of the steps and no length of time for any of them. How long each step takes is an operating matter, and how often the finished thing has to go out is a matter for the market regulator. The order survives whatever the regulator sets, and the order is therefore the part worth learning.

How do the lines tie back to net assets and the value per unit?

Through one identity, the mechanical spine of everything below. Every line valued, plus cash, plus the amounts the scheme is due, less the amounts it must pay, comes to net assets. Net assets divided by units in issue is the published value per unitThe figure one unit of a scheme is stated at, reached by dividing net assets by the number of units in issue..

The second half runs on the scheme at hand. Net assets of Rs 42,00,00,00,000, or Rs 4,200 crore written out in full rupees, divided by 1,20,00,00,000 units, gives Rs 35.00 a unit exactly. The division from Rs 42,00,00,00,000 to Rs 35.00 is performed here rather than quoted. A figure watched being produced behaves differently from a figure handed over finished.

So a holdings list is not a separate document that sits beside the value per unit; it is the same quantity written out at a finer grain, and a list that does not tie is not a list with a small problem, it is a description of some other scheme. The claim is a strong one and it is meant to be. If the expanded version and the summary version disagree, one of them is not about this scheme on this date, and finding out which is not a rounding exercise.

There is a trap hiding in how satisfying that division looks, and it needs saying out loud. Net assets divided by units giving the value per unit, and the value per unit multiplied by units giving back net assets, are one equation rearranged, not two checks. Written out: if N is net assets, U is units and V is the value per unit, then N divided by U equals V is the same statement as V multiplied by U equals N. Multiplying Rs 35.00 by 1,20,00,00,000 units brings Rs 42,00,00,00,000 back. The return trip feels like confirmation and is arithmetically incapable of adding anything the first line did not already say. Anything incapable of coming out wrong is not testing anything.

Here, then, is one with room to go wrong, put together out of the same three record figures. The scheme sits across 3,80,000 folios. Dividing 1,20,00,00,000 units by 3,80,000 folios gives an average holding per folio of 60000 over 19 units. A denominator of 19 does not terminate, so the average comes to 3,157.8947 units and keeps going. Rebuilding the total from that average is where it breaks. At 3,157.894 units a folio, 3,80,000 folios reconstruct 1,19,99,99,720 units, a residue of minus 280 units against the true total. At 3,157.895 units a folio they reconstruct 1,20,00,00,100 units, a residue of plus 100 units. The rebuild fails in both directions. A real check looks exactly like that: it has room to be wrong, and it shows that the rounded average is a description rather than a component.

What sits inside net assets that is not a holding at all?

Three things, and they are the reason the addition a curious reader tries at home never comes out. Cash sits inside net assets and appears in no holdings row. Every receivableMoney the scheme is due and has not yet had handed over to it., meaning money the scheme is entitled to and has not been paid, sits inside net assets and appears in no holdings row. Every payableAn amount the scheme has become liable for and has not yet paid out., meaning money the scheme has become liable for and has not paid, sits inside net assets as a deduction and appears in no holdings row.

One of those payables is computable from the record used here alone, and it is the most instructive number on the scheme. The ratio charged is 1.65 per cent, its base is net assets, and applied to Rs 4,200 crore that works out at Rs 69.30 crore for a whole year. The number of days a year is treated as having for an accrualAn amount recognised in the books on the day it arises rather than on the day cash actually moves. of this kind is a convention settled elsewhere, so a single day is taken as a fraction of the year rather than as a count. On the record used here that fraction of the year comes to about Rs 0.18986 crore. The accrued amount is recognised on the day it arises, and on that day it sits among the amounts the scheme must pay.

So a holdings list on this scheme could be complete and correct to the last security and still fail to add to net assets. About Rs 0.18986 crore of that day arithmetic is a deduction with no row anywhere behind it. Nothing was concealed. The list simply never claimed to be the whole identity, and a reader who treats it as the whole identity has misread the document rather than caught anybody out.

What can a reader check in a list, and what can nobody check?

The question splits in two. The halves have different answers, and the split is the actual skill. From the printed list alone, a reader can check that the shares of the scheme total add to something sensible rather than to something absurd. A reader can check that the same identifier does not turn up twice. A repeat is a duplication error and one of the few things catchable unaided. And a reader can check that the stated total agrees with the net assets behind the published value per unit. Agreement between the two is the tie.

The printed list cannot support anything at all about the two halves of a row. A reader cannot check that a quantity agrees with the custody record. The custody record is not visible from outside and was never meant to be. A reader cannot check that a carrying value was reached by applying the stated method. The method leaves its trace in working papers rather than in the printed figure. The trustee company and the auditor exist precisely to test those two, and a reader who understands what they personally cannot check has read the list correctly rather than incompletely.

And there is a subtler version of the error that catches careful people. A list that ties has proved one thing and has silently been credited with three. The tie has shown the arithmetic is consistent. The tie has shown nothing about whether the quantities match custody, nothing about whether the values followed the method, and nothing whatever about what happens next. A holdings list is a photograph of one instant that was already in the past when it was printed, and it tells a holder nothing about what the scheme will do or what any holding will be worth later.

Why is an invented holdings line worse than an empty one?

Because there is none to print, and because printing one anyway would be the worse of two errors. The record on this scheme amounts to net assets, a unit count, a folio count and a ratio. The record carries no security, no issuer, no quantity, no weight, no rating and no sector classification, and it never did.

A made up row would look exactly like a real one, would be quoted onward as though it meant something, and would teach nothing that the tie does not teach better. The reason is exactly that plain. There is no typographic device that marks an invented holding as invented once the row is screenshotted, pasted into a message, or read by a machine that keeps the row and drops the caveat. A number that cannot survive being separated from its warning should not be printed with a warning; it should not be printed.

So the shape gets drawn and the cells stay empty. The refusal costs almost nothing: every calculation above works at the level of the scheme, from Rs 4,200 crore down to Rs 31.58 on an average folio, and none of it needed a single holding. The interesting mechanics of a holdings list live in the identity it belongs to rather than in the rows themselves.

How big is a discrepancy once it reaches the published figure?

Divide it by the units in issue, and that is the entire method. Whatever the error was and wherever it sat, in a quantity, in a carrying value, in a receivable or in a payable, it enters net assets as a rupee amount and leaves as a rupee amount per unit, and the road out is the same road for all four.

Work it on this scheme. An error of Rs 1.20 crore is Rs 1,20,00,000 against 1,20,00,00,000 units. The result is Rs 0.01 a unit: one paise on a published figure of Rs 35.00. Push it up. Rs 3.00 crore is Rs 0.025 a unit, Rs 6.00 crore is Rs 0.05 a unit, and Rs 60.00 crore is Rs 0.50 a unit. Pull it down. Rs 0.60 crore is Rs 0.005 a unit, and that does not even reach a paise. Four different errors, one division, and the answer scales in a straight line because a division by a fixed unit count cannot do anything else.

Now put it where a person feels it. The average folio on this scheme holds 3,157.8947 units, so one paise a unit is about Rs 31.58 on that folio. Run the rest of the rungs through the same multiplication and Rs 0.005 a unit is about Rs 15.79, Rs 0.025 a unit is about Rs 78.95, and Rs 0.05 a unit is about Rs 157.89. The same division gives a way of sizing any discrepancy in any such list, on this scheme, without knowing one single thing about what the scheme holds.

Where do the frequency, the format and the deadline come from?

All four come from the market regulator. How often a scheme holdings list has to go out, in what shape, by what point, and carrying which contents are all matters the Securities and Exchange Board of India (SEBI) settles, and all four are revised from time to time. A reproduced figure would do worse than go stale on the day of a revision. A reproduced figure would state something untrue. Stating something untrue to somebody who trusted the figure is a far heavier thing to do.

So they get pointed at rather than printed. The current position sits at sebi.gov.in, to be read on the day the question arises, and the industry level picture of how such disclosure is laid out from one scheme to the next sits at amfiindia.com, a publisher of that material and not a maker of the rule. Taught here in place of those four is the part that survives every one of their revisions: a list has to be dated, has to state the instant it describes, and has to tie. Change the frequency, change the format, move the deadline, and those three conditions are exactly where they were.

Can the whole thing be laid out in one place, start to finish?

The whole thing can be laid out in one place, and the table below runs it that way on the scheme, joining the published end to the list end. Work down the rows and watch closely for the point at which the arithmetic simply stops. That break is the honest edge of what the record will support.

Four of the nine rows say the same thing, and it is the thing worth carrying away. The identity is fully specified and mostly unpopulated, and it still yields a working sizing rule. An absent holdings breakdown therefore costs a reader far less than it first appears to.

Girnar Asset Management Limited, the Girnar Large Cap Equity Fund, Kalyani Bhagat and Sohail Merchant are invented.
Educational material. Not advice on any investment, tax, budget or market position.