Fund Operations: From Order to Statement, Step by Step
Fund operations is everything that happens between an instruction and the record a holder can read back. An application is checked against the register that names the folio, the money is confirmed, the day's value per unit is struck from what the scheme holds, units follow from that value, the register is written, and only then does a statement report it. Four separate parties touch that line.
One observation makes the subject worth setting out on its own. Everything a holder ever sees about a scheme is an output. The value per unit on a screen is an output. The unit count printed on a statement is an output. The money that lands in a bank account after a redemption is an output. A holder reading a statement is inspecting the last step of a long line and nothing at all before it. Inspecting only the last step is safe so long as somebody else inspected the rest. Upstream inspection by other parties is what fund operationsThe working machinery that turns instructions and holdings into records: checking, confirming, valuing, computing, writing and reporting. means.
One scheme carries every figure below. The manager is Girnar Asset Management Limited, an invented firm, and the scheme is the Girnar Large Cap Equity Fund, an open ended equity one. Its net assetsWhat a scheme holds after deducting what it owes, which is the amount divided by the units in issue. stand at Rs 4,200 crore against 120.00 crore units in issue. Do the division yourself instead of taking the result on trust. Put Rs 4,200 crore over 120.00 crore units and what falls out is Rs 35.00 a unit, exact, with no remainder at all. The scheme is held across 3,80,000 folios. Sohail Merchant is the head of operations at the manager, and the equity portfolio there is run by Kalyani Bhagat.
Four other parties appear in this guide, each by task rather than by name. There is a custodian. There is a registrar and transfer agent. There is a trustee company. There is an auditor, and where one was involved, a distributor. The separation of those tasks is the entire teaching, and the moment a task is attached to a firm name a reader starts thinking about firms instead of about who is standing outside whose work.
What sits inside fund operations, and where does it start and stop?
Fund operations is easier to define by its two ends than by a list of desks. The line starts the moment an instruction reaches the scheme: somebody wants to put money in, take money out, or move it between schemes. The line finishes when the record of that instruction can be read back by the person who gave it. Between those two moments sits everything that turns an intention into an entry, and an entry into a document.
A wedding caterer has the same shape, and nobody finds it puzzling in that setting. An order comes in for a certain number of plates. Somebody checks the order against what was actually agreed. Somebody confirms the advance has landed. Somebody counts what is in the store and prices it. Somebody works out how many plates the store supports. Somebody writes the day's book. Somebody hands over a bill at the end. Nobody would accept a bill from a caterer who also owned the store, wrote the book and audited the count, and a scheme is arranged so that no one party does all four either.
Between the two ends sits a small number of repeating tasks, and they run every working day whether or not any particular holder did anything. The holdings have to be valued. The value per unit has to be computed from them. The unit registerThe record of who holds how many units of a scheme, kept folio by folio. has to absorb the day's purchases and redemptions. The securities have to be held somewhere and moved when they are traded. The books of the scheme have to be kept. Records held by different parties have to be matched against each other. And every one of those has a deadline attached.
How Mutual Fund Orders Move Through the Fund-Operations System: what happens first?
Walk it once, in outline, in the order it happens. An instruction arrives and is checked against the register for the folio it names. A request to redeem units that a folio does not carry has to stop there, and a request to add to a folio nobody has opened yet has to become a new folio first. The money is then confirmed as actually received by the scheme. Money arriving is a different event from the instruction arriving, and it is treated as a different event throughout.
The day's value per unit is then struck from what the scheme holds, not from what any holder did. The value per unit is a property of the scheme, computed once, and every instruction being processed against that day meets the same number. Units follow by division. An amount divided by a value per unit gives a unit count, and the allotmentThe act of creating and assigning units to a folio, which is what turns money received into a holding. is the moment those units are recorded as belonging to a folio. The register is written. Only after the register carries the new position can a statement be produced that reports it.
Every part of that is a statement about order and not about schedule. Which day's value per unit an instruction meets, how long the scheme has before that value must be published, and the period inside which redemption money must reach a holder are all set by the market regulator, and they all move. The sequence is a property of the arithmetic rather than of any rule, so the sequence is what holds steady.
Why can the units for a purchase not be computed before the day's value per unit has been struck?
Who does each step, and who is in a position to check it?
Four parties, four different tasks, and the separation is doing real work. The custodian keeps the securities that belong to the scheme. In practice custodyHolding securities on behalf of somebody else and moving them only on instruction, without any claim on them. means holding them, releasing them only on instruction, and having no view whatever on what they are worth. The registrar and transfer agent keeps the register of who holds units, folio by folio. Valuing the holdings and working the value per unit out from them falls to the operations team at the asset manager. The trustee companyThe body standing between the asset manager and the people who hold units, answerable for whether the scheme was run as its documents say. and the auditor test the result rather than producing any part of it.
Now say who can change what. The answer is the sentence worth carrying away. The registrar and transfer agent can change the register and can change nothing else. The custodian can move securities and can change nothing on the register. The operations team at the asset manager can change the valuation record and neither of the other two. The trustee company and the auditor can change none of the three, and can require any of the three to be explained. Because the party that computes the figure is not the party holding the securities and is not the party keeping the register, a value per unit can be tested by somebody who did not produce it.
The separation is also why the matching exercise between those records exists at all. ReconciliationSetting one record against another built independently, so that a difference between them shows up as a difference rather than staying hidden. works only when the records were built separately. Set a register total against a computed unit count, or a custody position against a holdings list, and a mismatch is information. Merge the two record keepers and the two records agree by construction. Agreement by construction is not the same thing as being right, and the matching exercise quietly becomes theatre.
Suppose the party computing a scheme value per unit also held its securities and kept its unit register. What has been lost?
Move one task at a time, and watch what stops being checkable
The control below moves a single task off the party that currently performs it and onto the party that computes the value per unit. Nothing else changes: the same three records exist at every position, the same work gets done, and the same documents come out. Only who built each record moves.
Educational illustration. Move it and watch what stops being checkable. The red bar down the left marks the records that have come under a single party.
AMC vs Trustee: who does the work, and who answers for whether it was done?
Written as two businesses this comparison teaches nothing, so write it as a division of one task instead. The asset manager does the work. The manager values the holdings, computes the value per unit, instructs trades, instructs the record keepers and produces the documents. The trustee company answers for whether that work was done the way the scheme documents said it would be. One produces; the other is answerable for the production. The two are not versions of the same job, and reading them as versions of the same job is where most confusion about this pair begins.
The trustee company does not produce a second value per unit, and the bar is structural rather than one of effort or cost. A checker that recomputes the answer has become a second producer of the answer, and once there are two producers and no one standing outside, which of them to believe has no answer inside the arrangement at all. Testing is a different activity from producing. Did the work follow the method as written? Were the inputs the ones that method calls for? Do the separate records agree? Was anything done that the scheme documents never allowed? Testing asks those four questions, and none of them produces a rival figure.
The everyday version is a school examination. The teacher sets and marks the paper. The examination board does not sit the paper again to see whether it gets the same marks; it checks that the marking scheme was applied, that the marks were added correctly, and that nothing was awarded outside the scheme. If the board sat the paper itself, there would be two sets of answers, no third party, and a disagreement with nowhere to go.
Should the trustee company recompute the value per unit itself as its way of checking?
How Fund Valuation Challenges Can Affect Reported NAV: where does the figure actually come from?
Most of what an equity scheme holds has a price that can be observed on the day, and for those holdings the valuation step is closer to reading than to deciding. Some holdings, at some times, do not have one. Trading in a security can be suspended. A security can go long stretches without a trade at all. An instrument can be of a kind that simply does not trade on a screen. The division cannot be performed with a blank in the numerator, so in each of those cases a value still has to reach the published figure. The value that reaches it comes out of a written method rather than off a screen.
The published value per unit is therefore partly an observation and partly the output of a policy, and the number itself does not show which parts came from where. That is not a defect and it is not concealment. A scheme must be able to strike a figure every day it is required to, including on days when part of what it holds has no observable price, and the method for doing so is written down in advance precisely so that it is not being decided in the moment by the party that benefits from the answer.
Two consequences follow and both matter. The first is that a reader who treats the published figure as wholly observed has misread what kind of number it is. The second is that this is exactly why the checking is separated: where a figure comes out of a method rather than off a screen, the useful question is whether the written method was followed, and that question can only be answered by somebody who did not apply it. The market regulator sets what the method must contain and what a scheme must do when a price is unavailable. The method itself is covered under the valuation policy.
The Girnar Large Cap Equity Fund publishes a value per unit of Rs 35.00. Is that an observed fact?
Why can these six steps not be run in a different order?
Because each step consumes the output of the one before it. The six steps are a chain, not a checklist. A checklist can be worked in any order and still be complete. A chain cannot. A step with no input has nothing to do.
Take them one at a time. The computation is a division and the value per unit is the divisor, so units cannot be computed before a value per unit exists. A value per unit struck after units have been issued would be struck on a pool of assets that already includes the money those units were issued for. The figure would depend on its own output, and new money would quietly take a share of value that existed before it arrived. The unit count is what gets written into the register, so the register cannot be written before the unit count exists. A statement is produced from the register and reports what the register says rather than what anybody intended it to say, so a statement cannot report a register that has not yet been updated.
The constraint is arithmetic rather than administrative, so nothing in that ordering is convention and none of it could be rearranged by agreement between the parties. Two parts of the process genuinely are decided. How long each step may take, and when the clock starts, are set by the market regulator and revised by it.
Why is a small operating error never a small amount of money?
Here is the arithmetic that the whole subject rests on, and it needs no holdings data, no daily series and no flow figures to work. Suppose something anywhere in the chain moves the published value per unit by one paise. Not by a rupee. By the smallest amount the figure can move at all.
One paise, on 120.00 crore units in issue, is Rs 0.01 multiplied by 1,20,00,00,000 units. The product is Rs 1,20,00,000/-, or Rs 1.20 crore. One paise and Rs 1.20 crore are the same event described twice. On a statement it is one paise and it is beneath the notice of every one of the 3,80,000 holders. Across the scheme it is Rs 1.20 crore of somebody else's money that has moved to somebody, or moved away from them. The per unit expression of an operating error is what makes it invisible, and the total expression is what makes it matter, and both descriptions are exactly correct at the same time.
Saying which holding caused a particular miss would need a breakdown of what the scheme holds, a month by month series and a record of flows in and out. The sensitivity needs none of that. The sensitivity falls straight out of the unit count, and anyone learning this subject should be able to compute it unaided.
Before reading on: something in the chain moves the published value per unit of this scheme by one paise. Across all 120.00 crore units, what is that worth?
What does the whole line look like, run once on one scheme?
Run it end to end on the invented record, with every figure computed here rather than quoted. Net assets are Rs 4,200 crore and units in issue are 120.00 crore. Divide Rs 4,200 crore by 120.00 crore units and the value per unit is Rs 35.00 exactly. Written in whole rupees, that is Rs 42,00,00,00,000/- divided by 1,20,00,00,000 units, and it still comes to Rs 35.00 with no remainder.
Spread the scheme over its 3,80,000 folios and an average folio comes to 1,20,00,00,000 units divided by 3,80,000. The division reduces to 60,000 over 19, about 3,157.89 units. At Rs 35.00 a unit that average folio is worth about Rs 1,10,526/-. Now put the one paise miss through the same three lenses, and keep the exact fraction beside the convenient percentage in each case.
| Reading | How it is built | Result |
|---|---|---|
| Across the scheme | Rs 0.01 a unit multiplied by 1,20,00,00,000 units | Rs 1,20,00,000/- |
| Against net assets | Rs 1.20 crore divided by Rs 4,200 crore, exactly 1 over 3,500 | about 0.02857 per cent |
| Against one unit | Rs 0.01 divided by Rs 35.00, exactly 1 over 3,500 | about 0.02857 per cent |
| On the average folio | About 3,157.89 units multiplied by Rs 0.01 | about Rs 31.58 |
| Against a year of expenses | Rs 1.20 crore divided by Rs 69.30 crore, exactly 4 over 231 | about 1.7316 per cent of a year |
| Check that can fail | Rounded Rs 31.58 multiplied back by 3,80,000 folios | Rs 1,20,00,400/- |
The second and third rows of that table are one equation rearranged, and saying so out loud is the difference between a check and a decoration. Write the unit count as U. Then the scheme wide miss is Rs 0.01 times U and net assets are Rs 35.00 times U, so the ratio is (0.01 times U) divided by (35 times U), and U cancels. The two rows cannot disagree for any U whatever. Agreement between them proves nothing at all. The fifth row is the same rearrangement wearing a different hat. The annual expense of Rs 69.30 crore is 1.65 per cent of Rs 35.00 times U, so U cancels there too.
The last row is a check that can genuinely fail. A check that can fail earns its place on the table. Rebuilding the scheme wide figure from the folio side, using the rounded numbers a reader would actually write down: Rs 31.58 on each of 3,80,000 folios comes to Rs 1,20,00,400/-, against the Rs 1,20,00,000/- the unit count gives. The residue is plus Rs 400/-, and it does not cancel. The exact unit count of 60,000 over 19 rebuilds the total to the paise, so the residue is rounding and nothing else. It is still reported as a signed row rather than quietly absorbed. A treatment that hides a residue of Rs 400/- teaches a habit that hides larger ones later.
One paise works out at about Rs 31.58 on the average folio here. Why is that the wrong way to size the problem?
Where do the timings and the deadlines live, and why does a summary of them go stale?
Every step described above has a deadline attached to it, and every one of those deadlines is set by the Securities and Exchange Board of India (SEBI), the market regulator, in the master circular it maintains for mutual funds. The timing at which an application counts as received, the value per unit that then applies to it, the period inside which redemption money must reach a holder, the deadline by which a value per unit must be published and the requirements a valuation method must satisfy all sit there. So does the treatment of a day on which a price cannot be observed.
Why a deadline has to exist at a given point in the chain is teachable and is set out here; where that deadline currently falls is checkable at sebi.gov.in. The distinction is not fussiness. A summary carrying the current window would not become merely old on the day the window moved. A stale window becomes wrong rather than merely old. A reader who had learned the number from it would carry a false fact forward with full confidence, and that is worse than carrying no fact at all.
Reasoning through why the deadlines have to exist predicts the shape of them without knowing any of the values. A value per unit has to be published within some stated period, or holders could not price their own holdings. An application has to be tied to a stated moment, or two holders applying on the same day could meet different prices for reasons neither of them controls. Redemption money has to arrive within some stated period, or an open ended scheme would not be open ended in any sense a holder could rely on. Each of those is a requirement whose necessity is derivable and whose value is not.
Where is the timing at which an application counts as received, and the deadline by which a value per unit is published, to be found?
Who reaches for this on a working day, and what do they do with it?
Three people use the sensitivity arithmetic and none of them is doing it out of interest. The head of operations here, Sohail Merchant, reaches for it to size an exception before deciding how hard to chase it: a discrepancy that could move the published figure by a paise is Rs 1.20 crore on this scheme and is chased tonight, not tomorrow. A lender taking units as security uses the same logic from the other side. The collateral is valued off a published figure produced by somebody else, so who tested that figure is a fact about the collateral rather than about administration.
An analyst comparing two managers cannot see either operating record from the outside at all, and this is the honest part. The arrangement is visible: whether the computing, the holding and the register keeping sit with separate parties, and whether the testing sits with a party that produces none of the three. The arrangement is observable and the error record is not. Supervision therefore fixes the arrangement.
A household holding units in one folio can do none of this and does not need to. The useful thing for a holder is smaller and sharper: knowing that the statement is the last step, that it reports the register rather than testing it, and that the reason to be relaxed about that is the separation upstream and not the neatness of the document.
The misreading that costs the most, and what it actually costs
A holder treats the operating chain as bookkeeping, on the reasonable ground that the numbers on a statement move in paise, and concludes that if something went wrong it would be small. The arithmetic runs the other way. One paise of the published value per unit on this scheme is Rs 1,20,00,000/- across the 120.00 crore units in issue. A miss of that size is exactly why the checking is spread across four parties instead of being left with the one that computes.
The cost of the misreading is not that the holder loses money on the spot. The real cost is that the holder has no way of telling a well operated scheme from a badly operated one by looking at a statement, and will keep believing that they do. A statement is produced from the register at the end of the chain, so it reports the register rather than testing it, and a document that is internally perfect tells nothing whatever about the five steps that fed it.
The fix is not vigilance. Nothing on the document gives vigilance anything to work on. The fix is knowing where the assurance actually comes from: the party computing the figure holds neither the securities nor the register, so any one of the three records can be set against the other two, and a party that produces none of them tests all three. The separation is what makes the figure worth something, not the arithmetic on the statement.
A holder wants to know whether a scheme is being well operated. Does the statement tell them?
Who fixes the timings, the periods and the valuation requirements?
SEBI fixes them, and they are set out in the master circular it maintains for mutual funds, together with the requirements a valuation method must meet, the treatment of a holding whose price cannot be observed, the conditions attaching to a record date, the rounding conventions a scheme follows and the frequency at which it must disclose. Whatever the position is on the day it matters, sebi.gov.in is where it is found.
The Association of Mutual Funds in India (AMFI), at amfiindia.com, puts out material about how the industry operates and figures at the level of individual schemes. AMFI is a place to look and decides nothing. A holding kept in dematerialised form is recorded at a depository instead, either the National Securities Depository Limited (NSDL) at nsdl.co.in or Central Depository Services (India) Limited (CDSL) at cdslindia.com. The mechanism above holds in any market, so a second market would only bring a second set of timings.
References
| Where to read it | Which body | What it settles |
|---|---|---|
| sebi.gov.in | Securities and Exchange Board of India | The master circular maintained for mutual funds, in which the timing at which an application counts as received, the value per unit that then applies, the period inside which redemption money reaches a holder, the deadline by which a value per unit is published and the requirements a valuation method must satisfy all sit. |
| amfiindia.com | Association of Mutual Funds in India | Material on how the industry operates, and the place scheme level figures get published. |
| nsdl.co.in | NSDL | One of the two depositories at which units held in dematerialised form are recorded |
| cdslindia.com | CDSL | The second depository at which units held in dematerialised form are recorded |
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.
