ISA 540 (Revised) Accounting Estimates: Auditing Assumptions, Data, and Methods

Most firms roll the ISA 540 procedures forward from the prior year. The risk assessment gets copied, the three response approaches get notionally ticked, and the file emerges with what looks like ISA 540 (Revised) coverage but is really SALY with a methodology shield. It is the finding the FRC has raised in four consecutive Audit Quality Review reports and the AFM flagged in its 2024 thematic on expected credit loss provisions.

The standard gives us three response approaches for accounting estimates (the fair value of investment property, the warranty provision, the useful life estimate embedded in every fixed asset, and the ECL on any loan portfolio). Picking the wrong one costs us either time or quality. Most of what the revised standard is really trying to stop is the instinct to just roll it forward.

ISA 540 (Revised) requires us to evaluate the reasonableness of accounting estimates by understanding the entity's estimation process ( ISA 540.13 -15), assessing risks of material misstatement ( ISA 540.17 -19), and responding through one or a combination of three approaches. Developing an independent estimate, testing management's process, or testing the data and assumptions used ( ISA 540.22 -28).

Understanding the entity's estimation process

ISA 540.13 requires us to obtain an understanding of the entity's process for making accounting estimates. This includes understanding the method, data, and assumptions the entity uses. The understanding is not optional and cannot be skipped in favour of jumping straight to testing.

What does "understanding the method" mean in practice? ISA 540.13 (a) asks us to understand how management identifies transactions, events, and conditions that give rise to the need for accounting estimates. A company with an expected credit loss provision should have a process for determining which receivables are credit-impaired and which forward-looking information to incorporate. We need to know what that process is before evaluating whether it produces a reasonable result.

ISA 540.13 (b) covers the data used. This includes internal data (historical loss rates, sales return percentages, warranty claim frequencies) and external data (macroeconomic forecasts, industry benchmarks, discount rates). We identify the data sources and consider their relevance and reliability.

ISA 540.14 requires us to understand the relevant controls over the estimation process. For a provision calculated annually by the CFO in a spreadsheet with no review, the control environment is weak. For the same provision calculated by a dedicated actuarial team with independent review and board-level approval, the control environment is different. This understanding feeds directly into the risk assessment.

ISA 540.15 requires us to understand how management assesses estimation uncertainty. Some entities quantify it (for example, by running sensitivity analyses showing the provision under different assumptions). Others do not. The level of management's own assessment of estimation uncertainty affects what we need to do independently.

Risk assessment for accounting estimates

ISA 540.17 requires us to identify and assess the risks of material misstatement for accounting estimates at the assertion level. This assessment uses the same inherent risk factors from ISA 315.12 (f) (complexity, subjectivity, change, uncertainty, and susceptibility to management bias), but accounting estimates tend to score higher on several of them.

Complexity is inherent in many estimates. An IFRS 9 expected credit loss model with multiple forward-looking scenarios is more complex than an inventory count. A deferred tax calculation under IAS 12 with temporary differences across multiple jurisdictions adds layers of complexity we must understand before testing.

Subjectivity is the defining characteristic of accounting estimates. Every estimate requires judgement. ISA 540.18 requires us to consider the degree to which the estimate depends on management's judgement and the extent to which the inputs are observable or unobservable. An estimate based entirely on observable market data (the fair value of a listed equity investment) has low subjectivity. An estimate based on management's projection of future cash flows (the recoverable amount of a cash-generating unit under IAS 36 ) has high subjectivity.

Estimation uncertainty is ISA 540 's equivalent of the uncertainty factor in ISA 315 . ISA 540.19 asks us to consider the degree of estimation uncertainty. The range of possible outcomes for the estimate. A warranty provision for a manufacturer with 15 years of claims data and stable products has low estimation uncertainty. A litigation provision where the outcome depends on a pending court decision has high estimation uncertainty.

When estimation uncertainty is high, inherent risk is positioned higher on the ISA 315 spectrum. This may result in the estimate being classified as a significant risk under ISA 315.17 (b), which triggers the additional requirements under ISA 330.15 and ISA 330.21 .

The three response approaches

ISA 540.22 -28 sets out three approaches for responding to assessed risks of material misstatement related to accounting estimates. We may use one approach or a combination. The choice depends on the nature of the estimate, the assessed risk, the availability of data, and the strength of management's own controls.

The first approach ( ISA 540.22 -23) is developing an auditor's independent estimate. We use our own method, data, or assumptions (or a combination) to develop an estimate and compare it to management's. This approach works best when we have access to independent data and can build an estimate from different inputs. For a fair value of investment property, we might obtain an independent valuation. For an expected credit loss provision, we might recalculate using our own forward-looking assumptions.

The advantage of an independent estimate is that it provides direct evidence about the reasonableness of management's number. The disadvantage is cost. It requires us to have the expertise, data, and tools to produce a credible alternative estimate. For small and mid-size firms, this approach can be disproportionately expensive for routine estimates.

The second approach ( ISA 540.24 -25) is testing management's process. We evaluate the method, test the data, and challenge the assumptions, but we do not produce an independent number. ISA 540.24 requires us to evaluate whether the method is appropriate given the applicable financial reporting framework and the nature of the estimate. ISA 540.25 requires us to test the data and to evaluate the reasonableness of the assumptions.

This approach is the most common in practice. It works when the entity's estimation process is well-structured and we can test the inputs and logic without needing to rebuild the model. The risk is superficial testing. Accepting management's model, checking a few inputs, and concluding without actually challenging the key assumptions does not satisfy ISA 540.25 . The point estimate almost always sits suspiciously close to management's number, and on review the working papers rarely explain why.

The third approach ( ISA 540.26 -28) is testing the data and significant assumptions used in management's estimate without evaluating the full method. This is a more targeted version of the second approach, focused on the inputs rather than the process. It works for estimates where the method is standard (for example, a straight-line depreciation calculation) and the real question is whether the inputs (useful life, residual value) are reasonable.

ISA 540.28 permits us to test data and assumptions by evaluating whether the assumptions are consistent with each other, with observable market conditions, with historical data, and with the entity's planned course of action. This is where most of the audit work actually happens on accounting estimates.

Choosing the right approach depends on the assessed risk. For a significant risk estimate (positioned at the upper end of the ISA 315 spectrum), ISA 330.21 requires specific substantive procedures. Testing a few data points under approach three may be insufficient. We may need approach one (independent estimate) or a rigorous version of approach two (testing management's full process with independent recalculation of key assumptions).

Estimation uncertainty and the point estimate

ISA 540.31 requires us to evaluate whether the accounting estimate and related disclosures are reasonable in the context of the applicable financial reporting framework, or are misstated. This evaluation considers whether the point estimate selected by management falls within our range of reasonable outcomes.

Estimation uncertainty creates a range. Our job is not to determine one correct number. It is to determine whether management's number falls within a range that the evidence supports. ISA 540 .A125 explains that when estimation uncertainty is high, the range of reasonable outcomes may be wide. When it is narrow, the point estimate is more constrained.

If management's estimate falls outside our range, the difference is a misstatement. ISA 540.32 requires us to determine whether a difference between our range and management's point estimate is reasonable. If it is not reasonable, we treat the difference as a misstatement and evaluate it under ISA 450 .

Management may also select a point estimate that is within our range but consistently at the aggressive end. ISA 540.32 -33 requires us to evaluate whether management's selections of point estimates and related disclosures indicate possible management bias. If each estimate is individually reasonable but all estimates are biased in the same direction (for example, all provisions are at the low end of the range), ISA 540.33 requires us to consider the effect on the financial statements as a whole. This connects to the ISA 240 fraud evaluation. Directional bias across estimates is a fraud indicator.

Retrospective review of prior-year estimates

ISA 540.14 (b) requires us to review the outcome of accounting estimates included in the prior-year FS (or their subsequent re-estimation). This retrospective review serves two purposes.

First, it provides evidence about the reliability of the entity's estimation process. If management's warranty provision has been materially understated for four consecutive years, that tells you the process has a systematic bias. The comparison of estimate to actual outcome (or current re-estimate) is direct evidence.

Second, it informs the current-year risk assessment. A history of reasonable estimates reduces inherent risk for the current-year estimate (though it does not eliminate it). A history of significant differences between estimates and outcomes increases inherent risk and may push the estimate toward significant risk territory.

The review is not mechanical. ISA 540 .A64 notes that the retrospective review considers the nature of the estimate and whether the conditions that gave rise to it are comparable. An expected credit loss provision set in 2019 (pre-pandemic) and compared to actual losses in 2020 does not tell you the model was wrong. It tells you the conditions changed. We exercise judgement about which PY comparisons are informative.

Disclosure requirements and what auditors miss

ISA 540.34 requires us to evaluate whether the disclosures related to accounting estimates are adequate in the context of the applicable financial reporting framework. This is not limited to checking that the note exists. We evaluate whether the disclosure communicates the nature of the estimate, the method used, the key assumptions, the sensitivity to those assumptions, and the degree of estimation uncertainty.

IFRS 7 requires specific disclosures for credit risk and expected credit losses. IAS 36 requires specific disclosures for impairment testing, including the key assumptions used in value-in-use calculations and their sensitivity. IAS 37 requires disclosures about the nature of provisions, including the uncertainties surrounding the amount and timing. For each estimate, we cross-reference the applicable IFRS disclosure requirement to the actual disclosure and evaluate whether the information is sufficient.

The common gap is disclosures about estimation uncertainty. IFRS practice statement 2 and IAS 1.125 -133 require entities to disclose the assumptions about the future and other sources of estimation uncertainty that have a significant risk of resulting in a material adjustment within the next financial year. Auditors frequently test the estimate itself but do not evaluate whether the disclosure of estimation uncertainty is adequate. Nobody enjoys rewriting a client's note 18, and that is exactly how the disclosure issue gets waved through.

Worked example: auditing warranty provisions at a Belgian manufacturer

Scenario: Janssen Machinebouw N.V. is a Belgian manufacturer of industrial packaging equipment with €38M revenue. The company provides a two-year warranty on all machines sold. The warranty provision at year-end is €1.2M, calculated by management using historical claims data from the past five years, adjusted for a new product line launched 14 months ago.

1. The engagement team obtains an understanding of Janssen's estimation process ( ISA 540.13 ). Management calculates the provision by multiplying the number of machines under warranty by the average claim cost per machine over the past five years. For the new product line (the X-400 series, representing 22% of current-year sales), management uses the average claim rate from the existing product range because the X-400 has only 14 months of warranty history.

Documentation note: "Understanding obtained per ISA 540.13 . Management uses a historical claim rate method. Data source: internal warranty claims database, five-year history. New product line (X-400, 22% of revenue) has only 14 months of claims data. Management applies the existing product average claim rate as a proxy for the X-400. No external data used."

2. The team assesses inherent risk. Complexity is low (simple multiplication model). Subjectivity is moderate (the assumption that the X-400 claim rate equals the existing range is a judgement). Change is elevated because the new product line represents a change in the product mix, and new products often have higher warranty claim rates than mature products. Uncertainty is elevated for the X-400 component because limited historical data means a wider range of possible outcomes. Susceptibility to management bias is moderate because the warranty provision reduces reported profit, and Janssen is approaching a bank covenant on EBITDA. The team positions the estimate at the mid-upper range of the ISA 315 spectrum for the X-400 component but lower for the mature product range.

Documentation note: "Risk assessment per ISA 540.17 -19. X-400 warranty component: mid-upper spectrum. Primary drivers: limited claims history (14 months, uncertainty factor), new product with untested reliability profile (change factor), proximity to EBITDA covenant (susceptibility to bias). Mature products: lower-mid spectrum. 5 years of stable claims data. Not classified as significant risk."

3. The team selects the second response approach (testing management's process, ISA 540.24 -25) for the mature product provision. They test the claims data by agreeing a sample of recorded claims to supporting documentation, recalculate the average claim rate, and verify the count of machines under warranty against the sales ledger. For the X-400 component, they use a combination of approach two and approach one. Testing management's data but also developing an independent estimate using first-year claim data from comparable product launches at peer companies (obtained from Janssen's industry association data).

Documentation note: "Response approach per ISA 540.22 -28. Mature products: approach two (test management's process). X-400: combination of approach two (test data) and approach one (independent estimate using industry peer data for new product launch claim rates). Peer data obtained from EUROPEM industry association 2024 survey of first-year warranty claim rates for industrial packaging equipment."

4. The retrospective review ( ISA 540.14 (b)) compares the PY warranty provision to actual claims settled during the current year. The PY provision was €980K. Actual claims settled against it totalled €910K, a 7% overstatement. The team notes this is within a reasonable tolerance but documents the directional pattern. The provision has been overstated in four of the past five years, with overstatements ranging from 3% to 12%.

Documentation note: "Retrospective review per ISA 540.14 (b). Prior-year provision: €980K. Actual claims: €910K. Overstatement: €70K (7%). Pattern observed: overstatement in 4 of 5 prior years (range: 3%-12%). Direction: consistently overstated, reducing reported profit. This pattern is opposite to the expected bias direction (management incentive to understate provisions to improve EBITDA). No management bias indicator identified. Pattern suggests conservative estimation methodology."

5. The independent estimate for the X-400 component produces a range of €310K to €420K. Management's estimate for the X-400 component is €340K. The team concludes that management's estimate falls within our range. The team evaluates the disclosure of estimation uncertainty per ISA 540.34 and confirms that the FS disclose the sensitivity of the warranty provision to the claim rate assumption for the new product line.

Documentation note: "Auditor's independent estimate range for X-400 component: €310K-€420K. Management's estimate: €340K. Falls within auditor's range. Disclosure per ISA 540.34 evaluated: Note 18 discloses the sensitivity of the warranty provision to claim rate assumptions, including specific disclosure for the X-400 product line. Disclosure assessed as adequate under IAS 37.85 ."

The file demonstrates a differentiated approach. A simple test-the-process response for the mature product range and a dual-approach response for the higher-risk new product component.

Practical checklist for ISA 540 engagements

Common mistakes in accounting estimate audits

• Testing management's model inputs without challenging the key assumptions. The FRC's Audit Quality Review 2023 found that auditors frequently verified the data accuracy (the numbers going into the model) but did not evaluate whether the underlying assumptions (growth rates, discount rates, claim frequencies) were reasonable. ISA 540.25 requires both.

• Failing to perform the retrospective review for estimates where the PY FS are not being re-audited. ISA 540.14 (b) applies regardless. The review of PY estimate versus actual outcome is required even on recurring engagements where the prior year was audited by the same firm.

• Accepting management's assessment of estimation uncertainty without independent evaluation. ISA 540.15 requires us to understand how management assesses estimation uncertainty, but ISA 540.19 requires us to make our own assessment. Using management's sensitivity analysis as the sole basis for the auditor's conclusion conflates the two.

• Writing the response approach memo as a formality. We've seen files that cite ISA 540.22 -28, tick "testing management's process," and then test two inputs. That is SALY with a methodology shield, and it does not survive an FRC or AFM review.

Frequently asked questions

Related content

• IFRS 9 ECL calculator: Supports the independent estimate approach ( ISA 540.22 -23) for expected credit loss provisions by providing a calculation framework separate from management's model.
• IAS 36 impairment calculator: Supports the testing of value-in-use calculations, including discount rate and cash flow assumption evaluation, directly relevant to ISA 540 procedures on impairment estimates.
• ISA 315 (Revised 2019) risk assessment: the spectrum of inherent risk: Explains the spectrum concept that drives the ISA 540 risk assessment for accounting estimates.

Related tools