IAS 37 provisions: expected value vs most likely outcome, with worked examples
IAS 37 gives two different ways to put a number on a provision: expected value (a probability-weighted average) and most likely outcome (the single most probable figure). Picking the wrong one is one of the most common provisioning mistakes, because the trigger isn't "how many outcomes were given" — it's whether you're measuring one obligation or a large population of similar ones.
The rule in one table
| What you're measuring | Method | IAS 37 reference |
|---|---|---|
| A large population of similar, largely independent items (warranties, returns, a class of claims) | Expected value — probability-weight every outcome and sum | para 39 |
| A single obligation (one lawsuit, one contract dispute, one onerous arrangement) | Most likely outcome — but adjusted if other outcomes are skewed to one side | para 40 |
Large population: expected value, worked example
A manufacturer sells 50,000 units of an appliance with a one-year warranty. Historical claims data shows: 92% of units need no repair, 6% need a minor repair averaging $80, and 2% need a major repair or replacement averaging $600. Because this is a large population of similar, independent items, IAS 37.39 requires expected value — you weight every outcome by its probability, not just the most common one.
- Expected cost per unit = (6% × $80) + (2% × $600) = $4.80 + $12.00 = $16.80
- Total provision = 50,000 units × $16.80 = $840,000
Journal entry at recognition: debit warranty expense $840,000, credit warranty provision $840,000. Note that 92% of units contribute nothing to the figure individually, yet they're still part of the weighting — the provision isn't the cost of the "worst case" units, it's the average cost spread across the whole population.
Single obligation: most likely outcome, worked example
A company is defending a breach-of-contract claim. Legal counsel assesses a 75% chance of losing and paying $3.2m, a 15% chance of losing and paying a reduced $1.8m, and a 10% chance of winning outright. This is one obligation, not a population, so IAS 37.40 calls for the single most likely figure: $3.2m, the 75% outcome.
The mistake to avoid: probability-weighting this the way you would a warranty population gives (75% × $3.2m) + (15% × $1.8m) + (10% × $0) = $2.67m — a figure nobody involved in the case actually expects to pay. IAS 37.40 exists precisely to stop a single dispute's provision from being dragged toward a blended average that doesn't correspond to any real outcome.
The nuance: when "most likely" isn't the final answer
Paragraph 40 adds a qualification: where the other possible outcomes are mostly higher or mostly lower than the most-likely figure, the best estimate must be adjusted up or down from that figure. Example: a single obligation has a 45% chance of $4.0m (the largest single probability), a 40% chance of $4.6m, and a 15% chance of $3.0m. The raw "most likely" figure is $4.0m, but the other outcomes are skewed upward — 40% sits above $4.0m against only 15% below it — so the best estimate should be adjusted upward from $4.0m, not left at the unadjusted modal figure. This is the detail most summaries of the "most likely outcome" rule leave out.
When to discount: paragraph 45 and 47
Where the time value of money effect is material, IAS 37.45 requires the provision to be measured at the present value of the expenditure expected to settle it. Paragraph 47 specifies the rate: a pre-tax rate reflecting current market assessments of the time value of money and the risks specific to the liability — but the rate must not also reflect risks you've already built into the cash flow estimate itself. Double-counting risk in both the cash flows and the rate is a common error.
Worked example: a company recognises an environmental remediation provision of $2,000,000, expected to be paid in five years. The effect of discounting is material at this horizon, so it's measured at present value using a 4% pre-tax rate:
- Present value = $2,000,000 ÷ 1.04 compounded over 5 years ≈ $1,643,854
Initial journal entry: debit remediation expense $1,643,854, credit provision $1,643,854. In each following year, the discount unwinds through finance cost, not the original expense line:
| Year | Opening balance | Unwind of discount (4%) | Closing balance |
|---|---|---|---|
| 1 | $1,643,854 | $65,754 | $1,709,608 |
| 2 | $1,709,608 | $68,384 | $1,777,992 |
| 3 | $1,777,992 | $71,120 | $1,849,112 |
| 4 | $1,849,112 | $73,964 | $1,923,076 |
| 5 | $1,923,076 | $76,924 | $2,000,000 |
Each year's unwind entry: debit finance cost, credit provision. By year 5, the provision has accreted back up to the full $2,000,000 settlement amount.
By contrast, a product recall provision of $500,000 expected to be paid out within two months would not be discounted — paragraph 45's test is materiality of the time value of money effect, not simply "does this extend past the current period." Short-horizon provisions are usually recognised at their full undiscounted amount.
Common mistakes
- Treating "the question gives several outcomes and probabilities" as the signal to use expected value. The real test is population vs. single obligation (paras 39 vs 40), not how many outcomes are listed.
- Applying unweighted "most likely outcome" without checking whether the other outcomes are skewed to one side, which para 40 says should shift the estimate.
- Using a post-tax discount rate, or a rate that reflects a risk already priced into the cash flows — paragraph 47 requires a pre-tax rate and prohibits reflecting the same risk twice.
- Discounting every provision that settles more than 12 months out. Paragraph 45 only requires discounting where the time value of money effect is material, with no fixed time threshold.
These measurement rules sit alongside the rest of the Conceptual Framework and IAS 1/8/10/24 material tested in the IFRS Concepts & Framework quiz, which also covers provision recognition, contingent assets, and onerous contracts under IAS 37.
This is educational material, not accounting or legal advice — apply the standard's full text and current interpretive guidance to your own facts.
Source: IAS 37 Provisions, Contingent Liabilities and Contingent Assets, paragraphs 39–41, 45 and 47 (IFRS Foundation), cross-verified against the Australian Accounting Standards Board's verbatim adoption of the same paragraphs (AASB 137).