IFRS 17 Actuarial Disclosure Items: What the Standard Actually Asks For

The single most useful thing a finance or actuarial team can do at the start of an IFRS 17 build is to work backwards from the disclosure note. Every architectural choice — projection granularity, discount-curve persistence, grouping logic, data lineage — either makes the disclosure note producible or makes it an ongoing reconciliation exercise. Teams that start from the measurement engine tend to rediscover this the hard way, usually in the second parallel run, when the disclosure note cannot be generated without a day of manual allocation.

What follows is the practitioner-level list of actuarial disclosure items the standard asks for, and what each one implies for the projection and measurement engines upstream of it. The aim is to give the build team a concrete view of what the finance team will need to produce every reporting period, so the architecture can be specified to produce it natively rather than reconstruct it after the fact.

1. Reconciliation of Contractual Service Margin

The CSM reconciliation — opening balance, CSM released, new business CSM, changes in estimates that adjust the CSM, and the effect of discount-rate unwinding — is the disclosure item that sets the structure of the measurement engine. Every line has to be traceable to a specific cash flow or assumption movement.

What this implies: the projection engine must output cash flows that separate experience from assumption change, and the measurement engine must persist the locked-in curve for every group so the discount-rate unwinding line can be calculated. If either of these is missing, the reconciliation becomes a set of balancing figures, and that is not an audit-defensible disclosure.

2. Reconciliation of Risk Adjustment

A similar reconciliation is required for the risk adjustment — opening balance, release, new business risk adjustment, and changes in the non-financial risks it is measuring. For many insurers, the risk adjustment is a newer component than the CSM and the reconciliation is the first place its calibration gets audited externally.

What this implies: the risk adjustment methodology has to produce period-by-period movements, not just point-estimates. A confidence-level approach that regenerates the adjustment end-to-end each period cannot explain the movement categorisation the disclosure requires. A roll-forward methodology — even a simple one — makes the reconciliation tractable.

3. Reconciliation of Liability for Remaining Coverage and Liability for Incurred Claims

The separation of liabilities between coverage not yet provided (LRC) and claims already incurred (LIC) is a structural disclosure item. The LRC reconciliation shows premiums, insurance revenue, amortisation, and experience. The LIC reconciliation shows claims incurred, claims paid, and changes in estimates.

What this implies: the actuarial model must tag cash flows by whether they relate to coverage already provided or coverage still to be provided. For contracts measured under the Premium Allocation Approach the split is structural; for GMM it is a tagging exercise through projection. Either way, it has to be consistent through the life of the contract.

4. Expected future cash flows in time buckets

The standard requires a time-bucketed disclosure of expected future cash flows — typically one-year, two-to-five-year, over-five-year buckets, though jurisdictions vary. The disclosure is for remaining coverage only.

What this implies: the projection engine has to produce cash flows at a granularity that can be aggregated into the disclosure buckets without interpolation. Annual projection buckets are insufficient if the first disclosure bucket is sub-annual. Monthly is the safe minimum.

5. Loss component on onerous groups

Where a group of contracts is onerous at initial recognition (or becomes onerous subsequently), a loss component is recognised and disclosed. The loss component has its own reconciliation — initial recognition losses, losses arising in the period, reversals.

What this implies: profitability testing at group level has to be automated, and onerous classification has to be persisted. A group that is borderline onerous, flipping between remaining-profitable and onerous across periods, triggers disclosure every time it flips. The engine has to produce a clean audit trail of when the classification changed and why.

6. Insurance revenue decomposition

Insurance revenue under IFRS 17 is the sum of expected claims, risk adjustment release, CSM release, and other amortised items. The disclosure requires those components to be separately identifiable.

What this implies: the measurement engine cannot treat insurance revenue as a single emitted number. Each component has to be produced separately and traced to the driver — expected claims to the claims projection, risk adjustment release to the risk adjustment engine, CSM release to the coverage unit schedule. If any of those drivers is a single aggregated output, the revenue decomposition collapses.

7. Significant judgements and inputs

The qualitative disclosure on significant judgements — grouping decisions, onerous classification methodology, risk adjustment calibration, discount-rate derivation — is often the most negotiated part of the first-year note. Every architectural choice made upstream shows up here.

What this implies: the methodology paper written at the start of the build becomes the audit-defence document for this disclosure. The discipline is to write it as if it were the disclosure, and then configure the engine to implement it literally. The version of the methodology paper that exists at go-live should be the version the external auditor reads for this note.

Closing thought

The disclosure note is not the last thing the IFRS 17 architecture produces; it is the specification the architecture is built against. Every reconciliation line, every time bucket, every decomposition drives back to something the projection engine, measurement engine, or methodology paper has to produce. Teams that write the first-year disclosure note in draft before the build is complete tend to find the gaps in the architecture early. Teams that wait for the measurement engine to produce its first set of numbers and then try to assemble the note typically rediscover the same gaps, less cheaply, two parallel runs later.