Fair Value Under IFRS 13: The Hierarchy That Governs Your Inputs

Fair value sounds like a single idea, but IFRS 13 turns it into a classification problem. Two funds can hold the same company, use the same multiple and still report under different levels of scrutiny, because what counts is where the inputs came from.

For private-market holdings the answer is nearly always Level 3. That is not a failing, it is the nature of the asset. What matters is knowing it, and carrying the documentation that goes with it.

Fair value is an exit price, not a cost

The definition is precise: the price that would be received to sell an asset in an orderly transaction between market participants at the measurement date. Three words in that sentence do the work.

Received makes it an exit price. Not what you paid, not what it would cost to build. Market participants makes it their view rather than yours, which rules out valuing on a strategic premium only you would pay. Measurement date fixes it in time, so events after the reporting date do not belong in the number.

Cost is not fair value. A recent transaction price is evidence of fair value at the transaction date, and its weight decays as time passes and facts change.

The three levels, in plain terms

Level 1 is a quoted price in an active market for the identical asset. A listed share you hold. Nothing to estimate.

Level 2 is other observable inputs: quoted prices for similar assets, or inputs derived from observable market data. A multiple taken from a documented set of listed comparables sits here as an input, even though the asset itself does not trade.

Level 3 is unobservable inputs, meaning inputs reflecting your own assumptions about what market participants would use. A discount for lack of marketability, a company-specific adjustment, a forecast nobody else can see.

The levels describe the inputs, not the quality of the work. A carefully built Level 3 valuation is worth more than a careless Level 2 one.

The lowest significant input decides the level

This is the rule most often misapplied. Your valuation may draw on an observable peer multiple, which feels like Level 2. But if you then apply a marketability discount you estimated yourself, and that discount is significant to the result, the whole measurement falls to Level 3.

So a typical private holding is Level 3 even when most of its inputs are observable, because the adjustments that turn a public multiple into a private valuation are yours.

The practical consequence: do not fight to classify a private holding as Level 2. Accept Level 3 and carry the documentation it requires.

What Level 3 actually requires of you

Beyond the valuation itself, Level 3 brings disclosure obligations that shape how the file must be built.

You need a description of the valuation technique and the inputs used. You need quantitative information about significant unobservable inputs, which in practice means stating the multiple, the discount and the range you considered. You need a reconciliation of opening to closing balances, separating what moved because of performance from what moved because of the market. And you need a sensitivity analysis showing what a reasonable change in the key input does to the value.

That last one is where most reporting gaps appear. It is also the easiest to produce if the valuation was built from documented, dated inputs in the first place.

Building a file that matches the level

Record the technique and why it suits the asset. Record the peer set and how it was constructed. Record every input with its source and the date it was observed, and keep the vintage you used each period.

Then record the adjustments separately from the base figure, each with its rationale and magnitude. This is what makes the sensitivity analysis a lookup rather than a project, and it is what lets you show that the level was assessed rather than assumed.

A valuation that can be reconstructed from its inputs satisfies the hierarchy almost as a side effect. One that cannot will fail the disclosure test even if the number itself is right.