How a Benchmark Gives a Fund Return Its Real Context

A benchmark gives a fund return its context by supplying a yardstick declared before the year started, so a bare figure has something to be measured against. The comparison works only when four things are checked: that the yardstick was declared in advance, that both figures cover the same dates, that both sit on the same cost basis, and that the index plausibly measures what the scheme was told to do.

Here is what sits underneath that. Any return figure is a measurement of one thing across one window. The figure is complete, it is arithmetically exact, and standing alone it is very nearly mute. Setting a second measurement of the same window beside it is what turns a measurement into a reading, and the four checks below are the whole of what makes that second measurement usable rather than decorative.

One scheme carries the whole walk-through. The Girnar Large Cap Equity Fund is an open ended equity scheme run by Girnar Asset Management Limited, holding net assets of Rs 4,200 crore against 120.00 crore units in issue, so one unit is worth Rs 35.00 once the first figure is divided by the second. Its expense ratio of 1.65 per cent takes Rs 69.30 crore out of that pool across a year. Kalyani Bhagat is the fund manager and operations sit with Sohail Merchant. Across the one stated year the scheme returned 13.4 per cent net. The benchmark it had declared returned 12.1 per cent, with nothing at all charged against it.

Why does a return figure need something to sit beside it?

The number begins entirely on its own. The Girnar Large Cap Equity Fund returned 13.4 per cent net across one stated year. Read by itself, that sentence supports exactly one statement: the value of a unit finished the year 13.4 per cent above where it began, after everything the scheme paid out of its own assets along the way. The sentence contains nothing for that year to be ordinary or unusual against, so whether the year was ordinary or unusual cannot be said.

Consider a shopkeeper who reports that takings rose 8 per cent last year. The next question is automatic, and it is not about the shop. It is about the street. If every other shop on that street rose 20 per cent, an 8 per cent rise is one story. If the street fell 5 per cent, the identical 8 per cent is a completely different story. Nothing about the shop changed between those two sentences. The change is that a second reading of the same year arrived alongside the first.

A second reading of that kind is what a benchmarkA published index a scheme names in advance as the yardstick its own result will be read against. supplies, and a benchmark supplies three specific things a bare figure cannot. A benchmark supplies a scale, so a reader can tell whether 13.4 per cent net is a large movement or a small one for that kind of holding. The index lived through the same months the scheme did, in the same weather, so the index supplies a control on the window. And the index was computed by somebody who had never heard of the Girnar Large Cap Equity Fund and had no reason at all to flatter it, so it supplies something a third person can check.

A benchmark turns a number into a comparison, and a comparison is still not an explanation. The distinction between a comparison and an explanation runs through everything below. Even a comparison surviving all four checks gives the size of a difference and stays completely silent about where the difference came from. Both halves have to be held at once, and dropping the second half is the most common route by which a careful reader publishes something that was never meant to be said.

The four checks run in a fixed order, and the order is not a matter of taste. Each one can destroy the value of every check after it, so the cheapest and most destructive check is placed first. Two further steps follow the four: writing down what the checks produced, and naming what the whole exercise still could not establish.

What is the first check on any benchmark comparison?

Whether the yardstick was declared in advanceNamed before the period being measured began, rather than picked once the result was already known.. Step one is that question and nothing more, and it is deliberately the cheapest check on the list.

The action is to find the benchmark named in the papers of the scheme itself, rather than in the comparison somebody is presenting, and to confirm that the naming predates the period being measured. Two different places, and the difference between them is the entire check. A benchmark appearing only in the comparison, and nowhere in what the scheme published before the year began, has not been declared. The benchmark has been selected.

The advance-declaration check already runs in ordinary life without being called one. When a student comes home and announces that 62 out of 100 was a good score, the first thing anybody sensible asks is who decided what counted as good, and when. Decided beforehand by the school, it is a standard. Decided afterwards by the student, holding the mark, it is a description of the mark wearing the clothes of a standard.

Step one runs first because failing it makes every later step pointless. If the yardstick was picked after the year had run, then confirming that both figures cover identical dates, restating them onto one cost basis, and testing whether the index matches the job the scheme was set are all careful work performed on something that was never a comparison. The arithmetic will be flawless and the conclusion will be worthless.

Which benchmark a scheme must declare, where it must declare it, and what it must disclose about the comparison afterwards are all set by the Securities and Exchange Board of India (SEBI), and they are exactly the kind of requirement that gets revised. The current position is published at sebi.gov.in and is read on the day it is needed. Step one is not a mechanism that can be reasoned out from first principles. Step one is an action to perform rather than an idea to understand.

How are the two figures confirmed to cover the same dates?

By reading both ends of both figures, not one end of each. Step two is a labels job rather than a numbers job.

The two failures this check catches are a scheme figure for a financial year set against an index figure for a calendar year, and a scheme figure measured to a recent date set against an index figure measured to a month end. Both produce two perfectly correct numbers describing two different stretches of time. Subtract one from the other and the result is a difference between periods dressed up as a difference between a scheme and its yardstick.

The household version is an electricity bill. One cycle reads higher than the cycle it is held against, and the conclusion drawn is that the household is using more power. Then the two cycles come into view: one covers 28 days and the other covers 35. The two amounts were never comparable, and nothing in either amount said so. The dates said so, and only the dates.

A date mismatch is invisible in the numbers themselves and shows up only in the labels. Catching it is a step rather than an instinct. A gap of 1.3 points looks precisely like a gap of 1.3 points whether the two windows line up or not. There is no ragged edge, no rounding oddity, no smell. Unless the start date and the end date are read on both sides, this error passes the analyst every time, and it passes the person the note is sent to as well.

How are both figures put on the same cost basis?

By restating one of them, keeping both rows, and never quietly replacing one with the other. Step three has two limbs, and the cost limb comes first because it moves a number.

Here is the rule the whole comparison stands on, written out rather than assumed. The Girnar Large Cap Equity Fund returned 13.4 per cent for the one stated year, and that figure is net: it is worked out from unit values into which the 1.65 per cent charge has already been folded, a little on each day, against the scheme's Rs 4,200 crore of assets. No fee comes off it later. The benchmark it declared returned 12.1 per cent over the same year, and that figure carries no cost at all. An index is not a thing anybody can actually hold, so there is no holder for a charge to fall on. Taking the second away from the first sets a net result beside a costless one, so the 1.3 points that subtraction produces is not a like for likeCompared after both figures have been put on identical rules, so the difference between them is not partly an artefact of how each was built. reading of anything.

The everyday shape of this is a fare. One driver quotes Rs 340/- and the other quotes Rs 310/-, and the second sounds cheaper until the first quote turns out to include the toll and the second not. Neither driver lied. The two numbers were built to different rules, and the only honest move is to put the toll on both sides before comparing, then to say out loud that it was done.

The action on this limb is to restate the scheme onto the costless basisThe set of rules a figure was built under: what it includes, what has been taken out of it, and what it is measured against. of its benchmark, and to keep both rows rather than replacing the published one. Adding the 1.65 per cent charge back onto the 13.4 per cent net gives 15.05 per cent, so the restated gap against a costless 12.1 per cent is 2.95 points rather than 1.3. The arithmetic checks backwards before it is trusted: 15.05 less 1.65 returns 13.4, and 12.1 plus 2.95 returns 15.05. Both arithmetic chains close exactly.

Now the part that has to be said out loud instead of buried. The add-back is arithmetic, and the charge is not. The charge builds up a little on each day, on an asset base that itself moves each day, so its effect compounds instead of adding. A strict restatement divides instead: 1.134 over 0.9835, exactly 324 over 281, giving a gross equivalent of about 15.30 per cent and a gap of about 3.20 points. Both of those are rounded down from 15.30249 and 3.20249. Even the strict route pretends that a charge taken in daily slices was one deduction made once at the end, so it is an approximation too. The distance between the two routes is roughly a quarter of a point, the word about belongs on both of them, and neither figure is forced to tie exactly. Either route comes through with a finding and not a figure. The honest gap sits somewhere close to three points, more than twice what the published 1.3 shows.

Why is the income question part of the basis question?

Because income is the second way two numbers can be built to different rules, and different rules is what the basis check exists to catch.

Income the scheme collected during the year landed in the pool and moved the value of a unit, so a scheme return already has that income inside it. An index has a choice. A total return indexAn index computed as though every dividend and payout it receives is put straight back into the index. is computed as though every payout it received went straight back in. A price indexAn index computed from prices alone, with any dividend or payout it receives simply left out. is computed from prices alone and leaves the payouts out. Setting a scheme that received income against an index computed on price alone credits the scheme with income the index was never given. The fault is exactly the cost limb wearing different clothes.

The record does not say which way the stated benchmark of the Girnar Large Cap Equity Fund treats income, so the check is recorded as unresolvedA check that was run properly and came back without an answer, because the information needed to settle it is not in the record. and nothing is assumed in either direction. Say what would change if it went each way, and then stop. If the benchmark is a total return index, both sides carry their income and this limb passes cleanly. If it is a price index, then part of the 1.3 points, and part of the restated gap as well, is income the index simply never counted, and the honest gap shrinks by an amount this record cannot size. Naming both branches and refusing to pick one is the output of the check.

How is it checked that the index measures what the scheme was told to do?

By reading the job the scheme was set, then asking whether an index built on different rules could run ahead of it or fall behind it for reasons that have nothing to do with the person managing it. Step four is mandate fitWhether the yardstick covers roughly the same job the scheme was set, so a difference between them is capable of meaning something., and it is the only check on the list that asks a question about meaning rather than about construction.

The parallel is a vegetable seller judged against the takings of a street full of phone shops. Both are honest counts of real money across the same months. Neither count is wrong. But whatever difference emerges speaks to vegetables against phones, not to the seller, and no amount of care with the dates or the costs repairs that.

The action is to read the stated mandate and ask what the index would have to do to move for a reason unconnected with the manager. If the answer is obvious and large, the comparison is measuring two different jobs. Whether any particular index suits any particular scheme is a judgement about a real product, so the question matters more than any answer a record of this kind could give. On the record as it stands the Girnar Large Cap Equity Fund is an equity scheme with a stated benchmark, and the record carries no index construction rules at all, so this check is recorded as not verifiableA check that cannot be completed because the information it needs is absent from the record entirely, rather than merely unclear. on this record. A blank of that kind is an honest output of the procedure and not a hole in it.

Two measuring sticks that are not the same stick

Girnar Asset Management also runs the Girnar Broad Market Index Fund, and its numbers sit close enough to the equity scheme's to be worth separating deliberately. Across the one stated year the benchmark declared by the Girnar Large Cap Equity Fund read 12.1 per cent, with nothing charged against it. Over that very same year, the unnamed broad index tracked by the Girnar Broad Market Index Fund read 12.40 per cent, again with nothing charged against it. The record never said those are the same index, and a large capitalisation benchmark and a broad market index are built to measure different things whatever their two figures happen to look like in any single year. The scheme was never measured against that stick, so never merge them, never subtract one from the other, and never set the equity scheme's 13.4 per cent net against the costless 12.40 per cent of the broad index.

One step along sits a near miss that catches readers the same way, so it gets named rather than hidden. The Girnar Broad Market Index Fund's own return for the same year was 12.12 per cent, net of its 0.20 per cent charge. The net figure sits one fiftieth of a point away from the 12.1 per cent costless stated benchmark of a completely different scheme. The closeness is a coincidence of arithmetic and nothing more: one figure is a scheme result after its charge, the other is a costless yardstick for a different part of the market, so two separate mismatches sit between them rather than one. No arithmetic is performed between those two numbers. The separation is settled by what each figure is a measurement of, not by the distance between two printed numbers. The rule holds unchanged in a year when the two read alike and in a year when they do not.

What gets written down once all four checks are done?

Two gap figures, each carrying its basis, and every check with the result it actually returned. Step five is where the work becomes a document, and the document is the output.

A reader handed one number will use that number, so writing down both gaps is the result of the procedure rather than a presentation choice. Putting only the published 1.3 points in front of somebody means they carry 1.3 points into their note, their summary and eventually their decision, and the basis label does not travel with it. Putting only the restated figure of about 2.95 points in front of them replaces what the scheme published with an approximation. The opposite error is different in kind and just as hard to undo. Both rows, both labels, every time.

What is still unexplained after an honest comparison?

Everything about how the gap came to exist. By now the work feels finished and a cause feels like the natural last sentence, so naming what is unexplained is the step people skip.

A clean comparison establishes a size. A clean comparison does not say which holdings produced it, whether more risk was taken to get it, or whether anything like it will happen again. The remaining work is attributionBreaking a difference in return down into the specific decisions and holdings that produced it, which needs data a headline comparison never contains., it needs holdings and weights and a risk reading, and this record holds none of those. The cricket score is known. Who scored the runs, on what pitch, and against what bowling are not.

A procedure that ends by naming what it could not establish is finished, and one that ends with a verdict has quietly added a step nobody can audit. Notice where the temptation lands: exactly here, at the end, when four careful checks have earned a feeling of authority that the fifth sentence then spends. The discipline is to write the gap, write the bases, write the two open rows, and stop.

What does the whole walk-through look like run end to end?

Here is every step in one place, on the Girnar Large Cap Equity Fund, with the output of each step shown rather than described. Read downwards, the result column has two rows that contain no number at all. A completed comparison has exactly that shape.

The two routes through step three disagree, and the walk-through leaves them disagreeing. About 2.95 points on the add-back and about 3.20 points on the strict division sit a quarter of a point apart, and neither figure is adjusted to make them meet. Both routes leave the same finding, and no single number: the honest gap sits near three points, more than double what the published 1.3 shows. If somebody needs one number from this, they need the sentence instead.

And here is the limit, set alongside the figures themselves. The record covers twelve months of one scheme. No second year exists, no month by month series and no sibling scheme. Twelve months will not stretch into a rate for a longer stretch, will not run backwards, and cannot be laid beside any actual fund or actual index. Nor does it show what picking holdings achieves, or what following an index achieves. Every subtraction above is exact and reaches almost nowhere, and both of those hold at the same time.

Girnar Asset Management Limited, the Girnar Large Cap Equity Fund, the Girnar Broad Market Index Fund, the stated benchmark, the unnamed broad index, Kalyani Bhagat and Sohail Merchant are invented.
Educational material. Not advice on any investment, tax, budget or market position.