Key Takeaways
- Unobservable inputs drive Level 3 fair value measurements, the tier with the widest estimation uncertainty and the heaviest disclosure requirements.
- IFRS 13.89 requires the entity to develop unobservable inputs using the best information reasonably available, including the entity's own data.
- Over 80% of AFM and FRC fair value inspection findings relate to Level 2 and Level 3 measurements, not Level 1.
- The auditor must evaluate each significant unobservable input individually, not accept the output figure alone.
What are Unobservable Inputs?
IFRS 13.87 defines unobservable inputs as inputs for which market data is not available and that are developed using the best information available about the assumptions market participants would use. That phrasing is deliberate: the entity does not project what it would pay or accept. It estimates what a hypothetical buyer or seller with market knowledge would assume.
Common examples include long-term growth rates in a discounted cash flow model, customer attrition rates applied to acquired intangible assets, marketability discounts for unlisted equity stakes, and credit spreads derived from internal default experience. IFRS 13.89 permits the entity to start with its own data but requires adjustment if reasonably available information indicates that other market participants would use different assumptions. ISA 540.18 directs the auditor to evaluate whether the data supporting the estimate (including each unobservable input) is relevant, reliable, and sufficient. That evaluation is where most audit hours concentrate on Level 3 measurements, because the auditor cannot simply corroborate the input against an external price feed.
The classification consequence matters. IFRS 13.73 requires the entire fair value measurement to be classified based on the lowest-level input that is significant to the measurement as a whole. A single significant unobservable input pushes the entire measurement into Level 3, triggering the expanded disclosure requirements in IFRS 13.93(d) and (h).
Worked example: Dupont Ingenierie S.A.S.
Client: French engineering services firm, FY2025, revenue EUR 92M, IFRS reporter. Dupont acquired a 100% stake in a specialised geotechnical consultancy in March 2023 for EUR 14,200,000 and recognised a customer relationship intangible asset of EUR 3,800,000 at acquisition under IFRS 3. The asset is amortised over 12 years. At 31 December 2025, impairment indicators exist because two anchor clients representing 18% of the consultancy's revenue did not renew their framework agreements.
Step 1 — Identify the unobservable inputs required
Dupont measures the recoverable amount of the customer relationship intangible using a multi-period excess earnings method (income approach per IFRS 13.B10). The model requires four inputs without observable market equivalents: projected revenue from the remaining customer base, the customer attrition rate, a contributory asset charge for the assembled workforce, and a risk-adjusted discount rate reflecting the uncertainty of the cash flows.
Documentation note: record each input, classify it as observable or unobservable per IFRS 13.86–89, and state why no observable equivalent exists. The customer attrition rate and contributory asset charge are entity-specific; no active market provides this data.
Step 2 — Develop the unobservable inputs
Management estimates annual customer attrition at 8%, based on the consultancy's five-year historical retention data adjusted upward from 6% to reflect the loss of the two anchor clients. The contributory asset charge for the assembled workforce is set at 4% of projected revenue, derived from replacement-cost analysis. The risk-adjusted discount rate is 13.5%, built from the entity's WACC of 9.1% plus a 4.4% premium for customer-concentration risk and intangible-asset illiquidity.
Documentation note: record the historical attrition data (source: CRM system export, 2020–2025), the adjustment rationale for the two lost clients, the replacement-cost calculation for the workforce charge, and the discount rate build-up per IFRS 13.B14. Reference ISA 540.18 for the auditor's evaluation of each assumption.
Step 3 — Calculate fair value
Applying the multi-period excess earnings method with 8% attrition, 4% contributory charge, and 13.5% discount rate produces a present value of EUR 2,640,000 for the remaining customer relationship. The carrying amount at 31 December 2025 (after 33 months of amortisation) is EUR 2,950,000.
Documentation note: record the full model output, attach the cash flow schedule showing each year's projected excess earnings, and document the present value calculation. Cross-reference to IFRS 13.93(d) for the required quantitative disclosure of each unobservable input.
Step 4 — Perform sensitivity analysis
Varying the attrition rate between 6% and 10% produces a fair value range of EUR 2,420,000 to EUR 2,870,000. Varying the discount rate between 12.0% and 15.0% produces a range of EUR 2,480,000 to EUR 2,810,000. Under every tested combination, fair value remains below carrying amount.
Documentation note: record the sensitivity table, confirm that the impairment conclusion holds across all reasonable scenarios per ISA 540.18, and document the resulting impairment loss of EUR 310,000 recognised in profit or loss per IAS 36.59.
Conclusion: the fair value of EUR 2,640,000 is defensible because each unobservable input traces to entity-specific data adjusted for market-participant perspective, the sensitivity analysis confirms the impairment conclusion across a reasonable range, and the disclosures satisfy IFRS 13.93(d) and (h).
Why it matters in practice
Teams accept the entity's unobservable inputs at face value without testing whether the assumptions reflect a market-participant perspective. IFRS 13.89 permits entity-specific data as a starting point but requires adjustment if available information suggests market participants would assume different values. ISA 540.18 expects the auditor to evaluate the reasonableness of each significant assumption. Recording the entity's number without that evaluation is a documentation gap that regulators flag consistently.
Entities classify a measurement as Level 2 when one or more significant inputs are unobservable. IFRS 13.73 requires classification at the lowest level of input significant to the entire measurement. Misclassifying Level 3 as Level 2 avoids the expanded sensitivity disclosures in IFRS 13.93(d) and (h), which understates estimation uncertainty for users of the financial statements.
Unobservable inputs vs. observable inputs
| Dimension | Unobservable inputs | Observable inputs |
|---|---|---|
| Data source | Entity's own data, internal models, assumptions about market-participant behaviour | Market prices, yield curves, broker quotes, publicly available indices |
| Fair value hierarchy | Level 3 (when significant to the measurement) | Level 1 (quoted prices) or Level 2 (adjusted observable data) |
| Disclosure burden | IFRS 13.93(d): quantitative table of inputs; IFRS 13.93(h): sensitivity narrative | IFRS 13.93(b): level classification and valuation technique description |
| Audit effort | ISA 540.18 requires individual evaluation of each significant assumption, source data, and adjustment | Verification of price source, measurement date, and active-market status |
| Typical instruments | Unlisted equity stakes, customer relationship intangibles, contingent consideration, long-dated options without liquid markets | Listed equities, government bonds, exchange-traded derivatives, interest rate swaps valued from yield curves |
The practical consequence: when an engagement involves Level 3 measurements, the auditor builds a separate testing programme for each significant unobservable input. That programme does not exist for Level 1 or most Level 2 measurements, which is why Level 3 work often accounts for a disproportionate share of the ISA 540 audit effort on fair value engagements.
Related terms
Frequently asked questions
How do I document an unobservable input in the audit file?
Record the input value, the data source (entity records, external studies, comparable transactions), the adjustments applied to move from entity-specific data to a market-participant assumption, and the range of reasonable alternatives tested. ISA 540.20 requires the auditor to evaluate whether the data and significant assumptions are relevant to the measurement date and internally consistent.
Can an entity use its own data as an unobservable input?
Yes. IFRS 13.89 explicitly permits entity-specific data when no market-participant data is reasonably available without undue cost and effort. The entity must not ignore reasonably available information that suggests market participants would use different assumptions. The auditor's role under ISA 540.18 is to test whether the entity made that assessment and whether the conclusion is supportable.
When does an unobservable input push a measurement into Level 3?
An unobservable input triggers Level 3 classification only when it is significant to the fair value measurement as a whole (IFRS 13.73). If the unobservable input affects less than 5% of the total fair value and all other significant inputs are observable, the measurement may remain at Level 2. Significance is a matter of judgment, and the auditor documents the rationale for the classification under ISA 540.13(a).