An illustrative diagnostic scenario for South African insurers where fragmented contract boundary data affects IFRS 17 grouping, CSM, risk adjustment and subledger confidence.
The issue is not always the actuarial model. In many South African insurers, the IFRS 17 calculation can be technically sound while the underlying contract data is split across policy administration, underwriting, claims, broker, reinsurance, finance and manual files.
That matters because the contract boundary is not a minor classification field. It determines which future cash flows belong inside the IFRS 17 measurement. If the boundary is wrong or inconsistently applied, the insurer may misstate fulfilment cash flows, grouping, contractual service margin (CSM), loss components, risk adjustment and the finance journals posted through the subledger.
This article sets out an illustrative diagnostic scenario. It is not a client case. It reflects a pattern that can arise in South African insurers with legacy platforms, product variation, broker-led distribution, manual actuarial workarounds and reporting pressure. For broader context, see Zorinthia’s insurance data strategy page and related insurance examples.
Consider a South African insurer with a mix of individual risk products, group schemes and selected short-term cover. Some products sit on an older policy administration system. Newer products are administered on a different platform. Group scheme amendments arrive through employer files and broker channels. Claims data is held separately. Reinsurance information is maintained by a specialist team. Finance receives summarised outputs into the IFRS 17 subledger.
At year-end, the IFRS 17 results are directionally plausible. The actuarial team can explain the movement in CSM for major portfolios. Finance can reconcile the final journals to the general ledger. The audit file contains model documentation, assumption papers and committee minutes.
But when the CFO asks why one group of contracts has produced an unexpected loss component, the answer becomes less firm.
The finance actuary traces the issue back to contract boundary assumptions. Some policies are treated as having cash flows beyond the next repricing date. Others, apparently similar, are cut off earlier. A scheme amendment loaded late in the year changes expected premiums but is not reflected consistently in the actuarial extract. Certain commission patterns are captured in broker files but not in the policy system. The reinsurance data uses its own treaty effective dates, which do not align neatly to the underlying policy records.
No single issue proves the IFRS 17 number is wrong. The concern is more practical: the insurer cannot show, quickly and cleanly, that the same contract boundary logic has been applied across all relevant data sources.
That is the heart of the IFRS 17 insurance data quality South Africa problem. The board does not only need a compliant result. It needs confidence that the result can be defended under audit scrutiny, regulatory questioning and internal challenge.
Contract boundary issues often arise because the information needed for IFRS 17 was never designed as one controlled data set.
A policy administration system may hold inception date, renewal date, premium frequency and product code. Underwriting may hold repricing rights and review conditions. Broker files may contain commission terms or scheme-level changes. Claims systems may hold benefit utilisation and incurred claims patterns. Reinsurance may hold treaty start dates, termination clauses and quota share percentages. Finance may hold product mappings used for ledger posting.
Each area may be accurate for its own purpose. The problem is that IFRS 17 needs these elements to work together at a more granular level.
For example, a group funeral scheme may have annual repricing rights, monthly premium collection, broker commission adjustments and member-level changes during the year. The policy system may show the scheme as active for a full year. The actuarial model may need to know whether future premiums beyond the repricing point are inside or outside the contract boundary. If scheme amendments are captured in spreadsheets before they are loaded into the administration system, the IFRS 17 extract may reflect only part of the commercial position.
This is not unusual in South Africa. Many insurers have grown through product extensions, distribution partnerships, acquisitions and regulatory change. Replacing all legacy systems is rarely a realistic near-term answer. The more immediate requirement is to know which data elements are critical, where they originate, how they are transformed, and who is accountable for their business meaning.
Contract boundary weakness can affect several IFRS 17 outputs at once.
Grouping is one example. If contracts are allocated to groups using incomplete product, issue date, profitability or portfolio data, the insurer may combine contracts that should be assessed separately. A small mapping error can affect whether onerous contracts are identified early or whether profitable and unprofitable experience is blurred.
CSM is another pressure point. The CSM depends on future service and expected future cash flows within the contract boundary. If premium projections include cash flows that the insurer has no substantive obligation to provide, or exclude cash flows that should be included, the pattern of profit recognition may be distorted. The finance actuary may still be able to produce a movement analysis, but the driver of the movement becomes harder to defend.
Risk adjustment can also be affected. If claims uncertainty, lapse behaviour or expense allocations are measured on populations that do not align with the contract boundary, the risk adjustment may reflect the wrong exposure base. In a South African environment where consumer affordability, load-shedding-related business interruptions and claims inflation can alter experience, weak boundary data makes it harder to separate genuine actuarial judgement from poor input quality.
The subledger then becomes the point where these weaknesses become visible to finance. Journals may post correctly according to configured rules, but the subledger cannot repair inconsistent source data. A clean posting process is not the same as a controlled measurement process.
A useful diagnostic does not begin by rebuilding the IFRS 17 model. It begins by isolating the data elements that determine the contract boundary and tracing them from source to subledger.
For this scenario, the CFO and finance actuary would focus on a manageable set of high-impact elements, such as:
The test is not whether these fields exist somewhere. The test is whether the insurer can show which version is authoritative for IFRS 17, how changes are approved, how late amendments are handled, and how the final value reaches the actuarial model and subledger.
This exercise should be done on actual reporting populations, not theoretical system diagrams. Select one material product line, one group business segment and one area with known manual intervention. Trace a sample of contracts from administration source through actuarial extract, model input, grouping logic, CSM calculation and subledger posting.
The outcome is usually revealing. Executives often discover that the most important data risks do not sit in the model itself. They sit in unclear ownership, undocumented transformations, spreadsheet adjustments and reconciliations that depend on a small number of experienced staff.
In South African insurers, the control design must reflect operating conditions.
Load-shedding remains a practical risk where batch processes, file transfers and overnight data loads are part of the reporting cycle. If a policy extract is incomplete because a process failed and was restarted manually, the insurer needs evidence of the failure, rerun approval and completeness check. A verbal assurance that “the team fixed it” is not sufficient for a controlled IFRS 17 environment.
POPIA also matters. IFRS 17 work often requires policyholder, member, claims and beneficiary information. Diagnostic work should use only the personal information needed for the purpose, with access restricted to the right people. Where samples are used for testing, the insurer should consider masking or minimising fields that are not required for the analysis.
Data immaturity is another reality. Some insurers still rely on spreadsheets for product mappings, actuarial adjustments or reinsurance overlays. Spreadsheets are not automatically unacceptable. They become a problem when there is no owner, no version control, no review evidence and no plan to reduce dependency where the risk is material.
The executive constraint is time. A CFO cannot wait for a multi-year data programme before improving IFRS 17 confidence. The practical route is to prioritise the contract boundary data elements that most affect reported results and audit effort.
A contract boundary diagnostic should produce more than a list of data defects. It should give finance leadership a decision-ready view of risk.
The CFO and finance actuary should expect to see:
This is where independent advisory work is useful. The point is not to blame actuarial, finance, IT or operations. The point is to create a shared factual picture that executives can act on.
For wider guidance on building that capability beyond a single reporting issue, see Zorinthia’s insurance data strategy work.
For a South African insurer, IFRS 17 contract boundary data is not a narrow actuarial concern. It is part of the financial reporting control environment.
The next executive question is simple:
Can we take one material IFRS 17 group of contracts and prove, from source system to subledger journal, that the contract boundary data is complete, consistent, approved and explainable?
If the answer is uncertain, the insurer should not start with a broad transformation slogan. It should run a focused diagnostic on the highest-risk product area, identify the weak data hand-offs, and agree who will own the corrections before the next reporting cycle.