Fund Accounting vs Fund Valuation: The Two Questions

Valuing the holdings of a scheme settles what they would fetch right now. Keeping its books settles what happened, what is due to the scheme and what the scheme owes. The valuing feeds the bookkeeping and the traffic runs one way: the books collect the day's figures, add what was earned, take out what was incurred, and divide what remains by the units in issue. One published number carries both and reveals neither.

Every figure below belongs to the Girnar Large Cap Equity Fund, an invented open ended equity scheme operated by Girnar Asset Management Limited. The scheme reports net assetsWhat a scheme holds after taking out what it owes. The figure the unit count is divided into. of Rs 4,200 crore and has 120.00 crore units outstanding, and one over the other settles a value per unitNet assets divided by the number of units in issue, published for the day. of Rs 35.00 with nothing left over. There are 3,80,000 folios behind that pool. Kalyani Bhagat picks what the scheme holds and Sohail Merchant runs the operating side.

Two more things are settled and taken as read. A written method exists for putting a figure against each kind of thing the scheme holds, and a working cycle exists for writing that day into the books. Both are covered separately. The far more useful question is which of the two is actually in view when a published number is in dispute. A reader who cannot tell the two apart will spend an afternoon checking the half that is almost never wrong.

What question does each of the two actually answer?

Start with valuing, and start outside the scheme. Valuing a holding is a statement about the world: at a stated instant, what would this thing fetch. Nobody inside Girnar Asset Management decides that. The world does, and the job is to observe it where the world is talking loudly and to estimate it, by a written method, where the world is quiet. The output is one figure against one holding, and that figure is called its carrying valueThe figure a holding stands at in the books of the scheme on a given day. once it lands in the books.

Consider a household with a gold chain in a locker. Asking what a buyer would pay for the chain today is a question about the world outside the house, and about what the metal fetches. Asking instead what the household paid for the chain, when, and whether the last instalment on it has been cleared is a question about the household itself. The two questions are different, and no amount of care with the second one answers the first.

Now the books. The books of the Girnar Large Cap Equity Fund hold a record of the history of the scheme: what came in, what went out, what is due to the scheme and not yet received, and what the scheme has incurred and not yet paid. The last pair is what an accrualRecognising an amount as earned or as incurred on the day it arises, whether or not any cash has moved. is for, and it is why a charge can appear in the books days before any money leaves. A scheme is a pool, and the number of claims against the pool is part of its history, so the books also carry the unit count.

The two are different kinds of question and not two views of one question. Valuing produces an estimate about the outside world. The books produce a record of what the scheme itself did. Nothing in that record shows what a holding is worth, and nothing in a valuation shows whether the scheme received a payment on the day it was due.

Which of the two runs first, and can the order ever reverse?

Valuing runs first, the books follow, and the arrow never turns round. The one way direction decides more than any other fact about where an error can hide. A total is an addition and an addition needs numbers to add, so the books cannot start until a figure stands against every holding. So the day is worked in one direction: the holdings are valued, the values go into the books, the books add what was earned and take out what was incurred, and what is left is divided by the units in issue.

Test the direction the other way and it collapses immediately. Could a valuer produce a figure for a holding without opening the books at all? Yes, comfortably. The question is about the world outside, and the written method points at the world outside. Could a bookkeeper produce net assets without any values? No, not even approximately. There is nothing to add. One of the two survives on its own and the other does not, and dependence is exactly that.

Because the dependence runs one way only, an argument about a published figure that opens in the books has opened downstream of where the answer probably is. That is not a claim that the books never go wrong. The claim is about where to look first, and it is worth more than most of the technique on either side. The reader who has the direction right searches a small space. The reader who has it backwards searches a large and mostly clean one.

What is each one good at, and what does that strength cost?

Valuing is good at being current. The written method reaches out and takes today rather than yesterday. A holder transacting today should transact on today, and that is exactly what the method delivers. The cost of that strength is that the world does not always supply a clean answer. Where the holding is quiet, or thinly traded, or has not moved recently, the method stops observing and starts estimating, and an estimate is a considered figure rather than a fact. A written method makes that estimate consistent and accountable. A written method does not make the estimate correct, and no hidden true figure sits behind it waiting to be found.

Bookkeeping is good at completeness and consistency. Everything that happened is in there, once, on the day it arose, in the same shape every time. Sameness of shape is what lets one day be compared with the next and a total be tested. The cost of that strength is uncomfortable and worth saying plainly: a complete and perfectly consistent set of books will record a wrong value faithfully, carry it forward every day, and never once object to it. Completeness is a property of the record, not of the figures inside it.

Notice that neither weakness is repairable by the other side. The estimate is about the world and the books are about the scheme, so no amount of bookkeeping discipline turns an estimate into an observation. A holding that was valued beautifully and then never entered is simply absent, so no amount of valuation care makes a record complete. Two unrepairable weaknesses are the whole reason the two exist as separate work rather than one, and why a scheme keeps two kinds of control rather than one bigger one.

How does a valuation go wrong, and how do the books go wrong?

A valuation failure has exactly one shape. Some holding stands at a number that re-running the written method would not return. Somebody reached for a source the method does not point at, or skipped a step, or applied the method to most instances and not to one. The record is otherwise immaculate. The figure is simply not the figure the method gives.

A bookkeeping failure has three shapes and they are all about presence rather than size. Something is not there. Something is there twice. Something is there on the wrong day. A charge that should have been recognised was not recognised, an entry was put through twice by two people who each thought the other had not, a receipt was written into the day before or the day after. None of the three has anything to do with whether the figures involved were right.

Two practical tests pull them apart cleanly. Run the written method over the holding a second time and a valuation failure turns up. Set one record beside another built separately and a bookkeeping failure turns up. Hold on to the two words. Re-application and matching are the whole distinction. Re-applicationRunning the written method over a holding again and comparing the figure it produces with the figure standing in the books. takes the method and the evidence about the holding, and asks whether the figure comes out the same. MatchingSetting one record against a second record built separately, and asking whether the two agree line by line. takes two records built by different parties and asks whether they agree.

Different work, different evidence, and in a well arranged operation, different people. The separation of the two tests is not bureaucracy. The two errors leave different traces, and a check that hunts for one kind of trace is blind to the other by construction rather than by carelessness.

Why do two unlike mistakes land on the same published number?

Because there is only one road out. Whatever went wrong, and on whichever side, net assets move. Net assets are then divided by the units in issue and the result is published. Two errors with nothing in common, found by different work, discovered by different people using different evidence, both pass through one division and come out looking exactly alike.

Most readers of a published figure go astray at exactly this point. The published value per unit keeps no record of which kind of error moved it, so a figure that is out cannot serve as evidence about which of the two disciplines slipped. There is no field in the number for cause. Somebody who looks at a wrong figure and announces that the bookkeeping must have slipped has skipped the only step that could have told them anything. The skipped step is testing each side against its own evidence.

What do the two errors look like when both are worked on one scheme?

Work both on the Girnar Large Cap Equity Fund and the distinction stops being an argument and becomes arithmetic. Start from the base, stated once and then held still. Net assets are Rs 4,200 crore, or Rs 42,00,00,00,000/- in full rupees, standing against 120.00 crore units, or 1,20,00,00,000 of them. The division gives Rs 35.00 a unit exactly, and it is exact because 4,200 divided by 120 is 35 with nothing left over. Rs 35.00 a unit is the settled starting figure every line below points back to.

The first error is a valuation error. One holding is carried Rs 1.20 crore above the figure the written method would have produced. Everything else is perfect. Net assets are then Rs 4,201.20 crore, and spreading Rs 1.20 crore over the 120.00 crore units comes to exactly Rs 0.01, so the published figure surfaces as Rs 35.01 instead of Rs 35.00. One paise. As a share of net assets that is one thirty fifth of one per cent, or 0.02857 per cent at five decimal places. On the average folio it works out at about Rs 31.58, and that line appears in the table below with its working.

The second error is a bookkeeping error. The mechanism is the point rather than a contest, so the two sizes are deliberately not matched. Every holding is carried at exactly the figure the method requires. A charge is missing: the running cost of the scheme for one day is not put through. The annual charge is divided into 365 equal parts, an assumption rather than a rule taken from anywhere, and one such part goes unrecorded. Take 1.65 per cent of the Rs 4,200 crore base and the year comes to Rs 69.30 crore, of which one 365th is about Rs 0.18986 crore. Divided by 120.00 crore units that is about Rs 0.0015822 a unit, about a sixth of one paise.

Two of the rows above are the same equation rearranged and cannot disagree, so neither is a check: dividing the error by the units and adding the result to Rs 35.00 is the identity (N plus e) over U minus N over U equals e over U, and it holds for every N, every U and every e. Running it twice proves nothing at all. So here is a check that can genuinely fail, and does. The average folio splits 120.00 crore units among 3,80,000 folios, or 60000 over 19 units, that is 3,157.894736 recurring. The count 3,80,000 carries a factor of 19, so the division never terminates. Round that half up to 3,157.89 units and give every folio the same figure, and the register holds 1,19,99,98,200 units against a true 1,20,00,00,000. The signed residue is minus 1,800 units, about Rs 63,000/- at Rs 35.00 a unit. A register of equal rows is arithmetically impossible on this scheme.

The same kind of check bites on the other side. Rs 69,30,00,000/- split into 365 equal whole rupee parts gives Rs 18,98,630/- each, and 365 of those is Rs 69,29,99,950/-, a signed residue of minus Rs 50/- for the year. Neither residue is a mistake by anybody. Both are what happens when a total that does not divide is forced into equal parts, and hiding them would teach that the arithmetic is tidier than it is.

Now the two readings the comparison exists to produce. The first reading: one published figure swallowed two unlike errors and showed the same face for each. Both arrived through the same division, so a reader looking at a figure that is out cannot tell from the figure whether a valuation was stretched or a charge was missed. The second: the work that finds each is entirely different. The valuation error surfaces when the written method is run over that one holding again. Every record agrees on the wrong figure, so no quantity of matching records would ever bring it up. The missed charge is found by setting what was put through against what should have been put through. Every holding is already right, so no amount of re-valuing would ever surface it. Different work and different evidence are why the two are kept apart rather than merged into one team with one checklist.

One honest note about size. The arithmetic invites a conclusion it cannot support. Reaching the Rs 1.20 crore of the valuation error would take about 6.32 of those daily amounts, exactly 1460 over 231 of them, so at these particular magnitudes the valuation error is the larger of the two. The ranking holds for those two amounts and nothing wider. Ranking the two disciplines by the size or the frequency of the errors they produce would need a full list of holdings, a run of daily figures and a log of what actually went wrong, and two chosen amounts cannot stand in for any of the three.

Who tests each of the two, and what evidence does each test need?

Testing the valuation means running the written method again from the start. The work needs three things: the method itself, the evidence about the holding that the method points at, and a list of every instance the method covers. The question is not only whether the method was applied but whether it was applied to all of them rather than most of them. A method followed nine times out of ten is a method that produced one figure nobody can account for.

The books are tested by matching. Matching needs a second record built by somebody else, and that is why the structure around a scheme has other parties in it at all. The custodianThe party holding the securities of a scheme and settling what the scheme trades. keeps its own record of what is held. The registrar and transfer agent keeps its own record of folios and units. Where units sit in dematerialised form there is a depository record as well, at the National Securities Depository Limited (NSDL) or Central Depository Services Limited (CDSL). Each of those was built independently, and independence is exactly what makes a mismatch informative and an agreement worth something.

Above both sits the trustee companyThe party that holds a scheme on behalf of its unit holders and answers for it without running any of the work., answerable for both and producing neither, and beside it the auditor, who tests rather than makes. The roles are distinct. Whoever answers for the work never performs it, and attaching a name to any of the roles drags attention towards an organisation and away from that point.

The two tests draw on different evidence, so an operating setup able to perform one of them but not the second is checked in one respect and left unchecked in the other. Half checked is the more honest word. A strong matching record says nothing whatever about whether the written method was applied. Matching was never asking.

Which of the two is a reader usually arguing about?

The valuation, almost every time, and almost always while believing the argument is about the books. The reason is a trick of language. Arithmetic is the visible part of the chain and the part a reader feels able to check, so a dispute arrives phrased as arithmetic. Run the additions again, the request goes. Show the workings.

Run them, and they will almost certainly be right. Arithmetic on a computed record is the most reliable link in the whole chain. The real dispute sits one step upstream: the figure standing against a holding, and whether the written method would have produced it. A figure standing against a holding is a question about worth, and questions about worth are not settled by addition.

Here is the tell. If what is disputed is what something is worth, the disagreement is about valuing and has merely borrowed the vocabulary of the books, and no quantity of adding up will close it however carefully the adding is done. Recognising that early moves the question to where evidence about it exists, and that turns a long argument into a short one.

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