The Fair Value Hierarchy: Level 1, 2 and 3 Explained

The fair value hierarchy sorts measurements by the quality of the evidence behind them, not by the quality of the holding. Level 1 is a quoted price for the identical thing in an active market. Level 2 uses other observable inputs. Level 3 uses inputs nobody can observe, so a model and a judgement supply them. A Level 3 holding is not a worse asset, only one measured on evidence an outsider cannot check.

Here is what sits underneath that. Fair value is one idea, established earlier: what a holding would fetch if it changed hands on the date the accounts are drawn. The trouble is that the idea says nothing whatever about how anybody arrived at the figure. A balance sheet has one column for the amount and no column for where the amount came from, so a number read off a screen in four seconds and a number built over three weeks from a forecast and a rate look identical once they are printed. The fair value hierarchyA three level ranking of fair value measurements according to how observable the inputs behind them are. The ranking applies to the evidence, not the holding, and the table sits in the notes. is the device that puts the missing column back.

Three things are already in hand. Fair value as a word was established earlier, as were the ideas of a custodian and of collateral. The choice between carrying something at cost and carrying it at a current value was worked through when inventory was measured. And the tests that classify an instrument were settled when debt and equity were separated. The three levels, the rule that places a measurement built from several inputs, the meaning of a transfer between levels, the Indian standard that carries all of this and an empty hierarchy table read as a finding rather than a gap all follow from those three.

What is the fair value hierarchy actually for?

The street is a better starting point than the standard. Two people state what their home is worth. The first lives in a block of forty identical flats, three of which changed hands last month within a few thousand rupees of each other, and she gives her number in four seconds. The second lives in a converted mill, one of a kind, and nothing remotely like it has been sold in eleven years. He gives his number in four seconds too. Neither is lying. Both numbers are the same length and both are said with the same confidence. Only one of them can be checked by anybody other than the person saying it.

The hierarchy exists because a reported amount carries none of the difference between those two answers, and the level is the label that carries it instead. The arrangement buys a great deal. Nobody has to second-guess either measurement, argue that one business is more honest than the other, or refuse to accept a figure. The two businesses each report what they believe, and then each states, in a table in the notes, the class of evidence its belief rests on. A reader who wants to test a balance sheet then knows exactly which parts can be tested from outside and which parts cannot, and can spend attention accordingly. The hierarchy is a remarkably cheap piece of machinery for the amount of information it produces.

So hold on to the sentence that governs everything below it. The level is a statement about the evidence behind a measurement, and it is never a statement about the merit, the safety or the worth of the thing measured. A Level 1 holding can halve next week. A Level 3 holding can be the steadiest thing a business has bought. The hierarchy is not looking at the holding at all, so it has nothing to say about either possibility. The hierarchy looks instead at what a reader would have to accept on trust in order to believe the number.

What puts a measurement in Level 1?

Level 1 is the narrowest of the three and the easiest to state. A measurement is Level 1 when it uses a quoted price, in an active marketA market in which transactions in the asset happen often enough and in enough volume that prices are available on a continuing basis. Occasional trades in a thin market do not make one., for the identical asset, taken exactly as the market gives it. Four conditions, and every word does work. A price, so a valuation somebody produced does not count. Quoted in an active market, so a price from somewhere that trades once a quarter does not count. For the identical asset, so a price for something similar does not count. And used as it stands, the condition people forget.

The everyday version is the packet of biscuits with the price printed on it. Nobody estimates the packet's worth. The price is read off, and the next person in the queue can read the same one, and so can anybody who walks into any other shop selling the same packet. Level 1 is exactly that: the number belongs to nobody in particular, the holder did not make it, and anybody can go and get the same one.

Adjusting a quoted price generally takes the measurement out of Level 1, however small the adjustment and however good the reason for it. The unforgiving rule surprises people, and the reason for it is worth understanding. The moment a business changes a market price, the reported amount is no longer something a reader can reproduce. The reported amount is the market price plus a decision, and the decision belongs to the business. The decision may be perfectly sensible. A decision it remains, and a reader who wants to check the figure now has to check that decision too. Checking a decision instead of a price is precisely the difference the hierarchy was built to record. So a measurement that starts from an unadjusted quoted price and ends there stays in Level 1; one that starts there and then moves generally drops to Level 2, and drops further if the adjustment itself rests on something nobody can observe.

What counts as Level 2, and why is it the widest of the three?

Level 2 is everything built from observable inputsInputs that come from outside the business and can be seen by anybody looking: market prices, published rates, and data about transactions that actually happened. other than a quoted price for the identical asset in an active market. Three shapes cover most of it. A quoted price for a similar asset, where nothing identical trades but something close enough does. A quoted price for the identical asset in a market that is not active, where the thing does trade but rarely. And observable rates, yields or transaction data fed into a standard technique that turns them into an amount.

Here is the street version. A five year old scooter, and the question of its resale price. The scooter has a particular mileage and a dented panel, so nothing identical to it is for sale today. But four of the same model and year are advertised, and the dealer down the road will say what a dent takes off. Every ingredient in the answer came from outside the owner's head. The model price was not invented, and neither was the deduction for the dent. A Level 2 measurement is exactly that, in one sentence: the answer was assembled, but no part of it was authored.

Level 2 is the widest of the three because most things a business holds do not have a live screen price and yet are not mysteries either, so the honest description of them is assembled from public parts. That width is also why Level 2 covers measurements of very different character. A quoted price for a similar asset is barely a step from Level 1. A long chain of observable inputs run through a technique with several stages is a long way from it, even though nothing in the chain was invented. The hierarchy does not grade within a level, and that is one of its real limitations. A reader who wants to know how far a Level 2 measurement travelled has to read the note describing the technique rather than the level itself.

What does a Level 3 measurement actually involve?

Level 3 is where the business runs out of things to look up. The level applies when the inputs that matter are unobservable inputsInputs for which no market data exists, so the business uses the best information available to it, which in practice means its own assumptions about the holding., meaning there is no market data for them and the business has to use the best information available. In practice the business uses its own belief about the holding. The belief is not idle. The belief is built from what the business knows about the holding, and it goes into the accounts as a number like any other.

A Level 3 measurement, described without teaching any technique, usually involves three things. An expected stream of amounts the holding should produce. A rate at which those amounts are brought back to todayReduced to what a future amount is worth now, because money arriving in five years is worth less than the same money arriving this afternoon. The size of the reduction depends on the rate chosen.. And a set of assumptions that shape both, such as how long the arrangement runs and how fast the underlying business grows. None of those three is available on a screen. All three come from the business. The standard therefore requires them to be disclosed and requires a sensitivity disclosureA statement of how much the reported amount would change if a key assumption were altered, given alongside a Level 3 measurement so a reader can see how much rests on that assumption. alongside them, showing how the amount moves if the assumptions move.

Here is an example closer to home. A cousin runs a small tailoring unit near a bus depot, and money went into it four years ago for a quarter share. Somebody asks what that quarter share is worth today. There is no market in quarter shares of tailoring units near bus depots. So the answer comes from what the unit brings in, what it might bring in next year, and what somebody would want as a return for taking it on. The answer is a real number and the best one available. The answer is also unverifiable by the person asking, and that unverifiability is exactly and only what Level 3 records.

Level 3 is not a failure state, a warning, or a sign that anything has gone wrong; it is what the accounts must say about a holding for which the world supplies no price. Shares in a private supplier can never be anything else. Nor can a holding in an unlisted arrangement, an interest in a small local business, or anything else that simply does not trade. If a business holds such a thing and measures it at fair value, Level 3 is the truthful answer and any other answer would be false. The disclosure is doing its job when it says so.

Why does one unobservable input decide the level of the whole measurement?

Most measurements are not built from a single input. Most are built from several inputs, and the several rarely share a level. So a placement rule is needed, and it is the rule most readers of accounts have never been told. A measurement takes the level of the lowest input that is significant to it. One significant unobservable input therefore pulls the entire measurement to Level 3, however observable everything else in it is.

A restaurant bill makes the same point. Four of the five lines are printed on the menu and every one can be checked against the board on the wall. The fifth line says the fish was charged at whatever the kitchen judged it was worth this morning, and the fish is most of the bill. How reliable is the total? The four checkable lines cannot rescue the one that is not, so the total is exactly as reliable as the fish line. The total is a single number and it inherits the weakest thing inside it that actually matters.

The word significant carries the entire rule, and it is not a computed threshold. An input is significant when it matters to the answer, and how much it has to matter is a judgement made by the business preparing the accounts and reviewed by whoever audits them. No percentage at which an input becomes significant is stated in the standard. An unobservable input that genuinely makes almost no difference to the amount does not drag the measurement down. One that decides most of the answer does. And a wide band lies in between. Two careful people would disagree inside that band, and saying so is the honest description of the rule.

The lowest-significant-input rule is why a table in a set of accounts can show a large Level 3 figure for a business whose measurements look, from the outside, mostly market based. Such a table is not a contradiction and it is not evasion. The rule is working. A measurement is described by its weakest significant link, deliberately, and that link is the one a reader most needs to know about.

Four settings on the panel are worth reading in prose. At the default, a single unadjusted quoted price for the identical holding gives Level 1 on Rs 4,00,000, or 30.8 per cent of the Rs 13,00,000 of illustrative holdings measured at fair value. Add an observable adjustment and mark it insignificant, and the level falls to Level 2 anyway: a price that has been adjusted is no longer used as it stands. Now set the starting price to a quoted price for the identical holding in an active market, keep the adjustment observable and significant, and make only the rate one the business builds itself and marks significant: the panel returns Level 3, driven by that one input. The lowest-significant-input rule is doing exactly what the prose describes. Then mark that same rate arguable instead of significant. The panel refuses to return a level and says so. Whether an input is significant enough to pull a measurement down is a judgement about the amount at stake and the technique used, and no switch on the panel holds that fact.

Why do businesses disclose movements between levels?

A holding does not have to change for its level to change. The evidence available about the holding changes instead. A business therefore discloses transfers between levelsMovements of a holding from one level of the hierarchy to another between one reporting date and the next, disclosed along with the business's stated policy on when it treats such a movement as having happened., and states when it treats a transfer as having occurred, so a reader can see the movement rather than infer it from two tables that no longer agree.

The usual cause of a move from Level 1 to Level 2 is that the market for the holding stopped being active. The shares are the same shares. The business did nothing. Trading in them thinned out until a quoted price was no longer a price in an active market, so the measurement had to be built from other observable things instead. A transfer of that kind is information about the market the holding sits in, not about its holder, and reading it the other way round is the most common misreading of the whole disclosure.

Which leads to the reading that is worth more than any single business's table. When many businesses transfer holdings into Level 3 in the same period, that is a statement about liquidity conditions across a market, not a statement about those businesses simultaneously deciding to be less transparent. Prices went away. The same screen went blank for all of them, so everybody who held those instruments had to start estimating at the same time. A reader who notices the pattern across several sets of accounts has learned something about the market that no individual filing would have told them.

Ind AS 113: Fair Value Measurement, what does the standard settle?

The hierarchy is a principle about evidence rather than a local rule, so everything above holds wherever it is read. The named document that carries that principle in India, and the exact conditions inside it, are a separate matter and belong in one clearly marked place.

Where do four holdings fall, and what does Anjani Stationers' own table show?

The whole structure runs on a portfolio small enough to hold in mind. Four holdings, every one of them hypothetical: a fixed deposit of Rs 10,00,000 maturing in fourteen months, an investment of Rs 6,00,000 in a debt mutual fund redeemable on demand, Rs 4,00,000 of quoted equity shares, and Rs 3,00,000 of shares in an unquoted private supplier. Rs 23,00,000 in total. Anjani Stationers Private Limited holds none of it.

The Rs 4,00,000 of quoted shares is the easy one. A quoted price for the identical shares exists in an active market and is used as it stands, so the measurement is Level 1. The Rs 3,00,000 of unquoted supplier shares is the other easy one and lands at the far end: no market at all, so an expected stream, a rate and a set of assumptions, all supplied by the business, and the measurement is Level 3.

The Rs 6,00,000 debt fund is the interesting one, and it is interesting because the answer is a judgement rather than a fact. Whether the fund's own published value counts as a quoted price in an active market for the identical holding is exactly the question the placement turns on, and two careful preparers can land in different places on it. If that published value is a price at which the holding can actually be transacted, in a market that is active, the measurement is Level 1. If the holding is redeemable against the fund rather than traded, or the published value is derived rather than quoted, the measurement is assembled from observable inputs and is Level 2. Naming that question matters more than the answer. No single answer holds for every debt fund everywhere.

The Rs 10,00,000 fixed deposit is the one people place wrongly. Their mistake is trying to place it at all. A deposit of that kind is carried at amortised cost, not at fair value, so there is no fair value measurement to rank and it does not appear in the measurement hierarchy table at all. The absence is not a gap; it is a different question. There is one refinement worth carrying: where a business discloses the fair value of something it carries at amortised costA way of carrying a financial asset at what was paid for it, adjusted over time for interest earned and amounts received, rather than remeasuring it to a current value at each reporting date., that disclosed figure does carry a level of its own. The distinction to hold is between the table of things measured at fair value and any separate disclosure of fair values for things measured otherwise.

Now the proportions, the figures a reader actually computes. Of the Rs 13,00,000 measured at fair value, the Rs 3,00,000 of unquoted supplier shares is 23.1 per cent, and that figure is the same whichever way the debt fund is placed. If the fund is Level 1, then Rs 10,00,000 or 76.9 per cent is Level 1 and 23.1 per cent is Level 3. If the fund is Level 2, then Level 1 is Rs 4,00,000 or 30.8 per cent, Level 2 is Rs 6,00,000 or 46.2 per cent and Level 3 is 23.1 per cent, and the three come to 100 per cent before rounding. Against the whole Rs 23,00,000 portfolio, including the deposit that is not ranked, the Level 3 share is 13.0 per cent. Three different denominators, three different answers, all of them correct about different questions. The note therefore has to say which set it is describing.

Anjani Stationers Private Limited itself supplies the finding. Anjani Stationers reports total assets of Rs 1,80,00,000 in its standalone accounts. Its financial assets are cash and cash equivalents of Rs 5,00,000, trade receivables net of Rs 86,00,000, and the Rs 21,00,000 holding in Chitra Binding Works. Not one of those is measured at fair value. Cash is cash, with nothing to measure. Receivables are carried at amortised cost. And the Chitra Binding holding is a subsidiary carried at cost in the standalone accounts rather than a fair-valued financial asset of the kind this guide is about. So Anjani Stationers' fair value hierarchy table has no rows in it at all.

An empty hierarchy table is a finding, not an omission, and it says something precise: not one rupee of this balance sheet rests on a fair value judgement. That is a fact worth knowing about a business, and it is unusual enough to be worth stating. The empty table is also the honest position for Anjani Stationers, whose 143.1 day working capital cycle absorbed the surplus that would otherwise have been available to hold anything, and whose Rs 5,00,000 of year-end cash sat alongside a seasonal facility drawn at an average of about Rs 26,40,000 through the year. A business in that shape does not have a portfolio to measure. Read the empty table alongside the cycle and the two facts explain each other.

What should a reader actually do with the level information?

Four moves, in order, and none of them takes long. The first is the hierarchy table, and what proportion of the holdings measured at fair value sits in Level 3. The second is the assumptions disclosed for those Level 3 measurements. The third is the sensitivity beside them, showing how much the reported amount moves if those assumptions move. The fourth is the transfers, showing whether anything shifted level during the period and why. Four moves are the whole procedure, and at the end of them a reader knows how much of a balance sheet rests on judgement and how sensitive that part is.

Watch three different people do it in the same week. A lender's question is what it can rely on if things go wrong, so it treats the Level 3 portion as the part of the asset base whose amount it would want to test independently before lending against it, and asks for the assumptions rather than accepting the total. An analyst comparing two businesses wants to know whether a difference in reported assets is a difference in holdings or a difference in measurement, and the hierarchy tables answer that in about ninety seconds. And a finance controller in Vaidehi Rao's position at Anjani Stationers reads it from the other side: she has to be able to defend every input she cannot point at. The assumptions and the sensitivity are therefore prepared long before anybody asks for them.

A Level 3 label describes what evidence exists and says nothing whatever about whether a holding is sound, so it is a reason to read more carefully and never in itself a reason to distrust. Some entirely ordinary holdings can never be anything else. No market for such shares will ever exist, so shares in a private supplier a business has held for years, taken as part of a long trading relationship, are Level 3 on the day they are bought and Level 3 on the day they are sold. Marking that business down for carrying doubtful assets is not analysis. The screen is mistaking a description of evidence for a verdict on substance, and it would penalise exactly the businesses whose holdings are least liquid rather than the ones whose measurements are most fragile.

Anjani Stationers Private Limited, Chitra Binding Works and Vaidehi Rao are invented.
Educational material. Not advice on any investment, tax, budget or market position.