Practical Guidance: Streamlining Prophet Models for IFRS 17 Reporting (4)

Guide 4: Watch Out for Result File Size Growth

One of the major challenges in IFRS 17 actuarial reporting is the need to perform calculations at the Insurance Cohort Group (“ICG”) level, which require high level of granularity in the calculations. While this requirement is essential for traceability and transparency, it also introduces a significant burden on model performance and result file management — especially when working with tools like Prophet.

In most cases, new ICGs are introduced every financial year (e.g., from new policies issued), while existing ICGs slowly run off over time. This means that the number of ICGs accumulates as years pass, and your result file grows steadily larger with each financial year cycle. What many actuarial teams fail to realize is how quickly this buildup can affect not just file size, but also run time and overall system performance. Waiting until your Prophet model hits a performance bottleneck is risky and can result in delays, rework, or even missed reporting deadlines.

If you are currently using SPCODE or Flexible Results to group results by ICG, now is the time to step back and assess:

1. How fast is your result file growing each financial year cycle?
2. How much longer does your Prophet run take compared to earlier periods?
3. Is your current hardware (memory, storage, processing speed) still able to handle the model runs efficiently?

These are not just technical questions — they have real operational implications. Larger result files can strain your Prophet environment, increase the risk of run failures, and make post-processing more time-consuming.

Don’t Rely Only on SPCODE

To manage this growth more effectively, one practical strategy is to combine SPCODE with Accumulation for ICG grouping. This technique allows you to multiply your grouping capacity without needing to expand the number of SPCODEs in your model point file. For instance, using 5 SPCODEs with 10 Accumulations gives you 50 unique ICG combinations. This helps reduce the need for a long list of SPCODEs while still achieving the granularity required for IFRS 17.

That said, SPCODE is often used for other purposes too, such as product segmentation, distribution channel analysis, or model testing. Therefore, it’s even more effective to build dedicated ICG grouping logic into your Prophet model — one that decouples operational grouping from business analysis grouping. With the right setup, you can retain flexibility for ad-hoc analysis while keeping your model efficient and scalable for future reporting cycles.

In short, proactive ICG management is critical. By planning early and embedding the right grouping mechanism, you can avoid performance bottlenecks, reduce file size inflation, and ensure your Prophet model continues to serve both regulatory and internal needs reliably.

---

Previous Guide (Guide 3) | Resources | Next Guide (Guide 5)