As companies move beyond IFRS 17 implementation and shift focus to streamlining their actuarial processes, one area that often requires attention is the Analysis of Change (“AOC”) calculations. These calculations are critical in IFRS 17 reporting, as they provide insights into the drivers of movements in key balances like Best Estimate Liabilities (“BEL”), Risk Adjustment (“RA”), and Contractual Service Margin (“CSM”). However, if not set up efficiently, AOC can easily become one of the most resource-intensive aspects of the entire roll-forward process.
II. Limit Prophet Runs into 2 Runs, i.e., Opening Run & Closing Run
AOC typically involves breaking down movements into various categories—such as the impact of new business, model changes, interest unwinding, experience variance, and assumption changes. The level of detail required can vary depending on internal policies or external audit requirements, but even a modest AOC breakdown requires multiple layers of calculations.
For many Prophet users, the common approach is to run individual Prophet runs for each AOC item. While straightforward in concept, this method comes with significant downsides:
- Extensive setup time to define, test, and maintain each run
- Large run result files, leading to storage and performance issues
- Prolonged run time, especially for high-volume portfolios
- Increased risk of human error from managing multiple inputs and outputs
A More Efficient Approach: Two Prophet Runs with Calculation Looping
To reduce the operational burden while maintaining accuracy and auditability, we recommend consolidating AOC calculations into just two Prophet runs:
- Opening run: Uses model point files combining previous month’s in-force policies and current month’s new business
- Closing run: Uses model point files for current month’s new business only
All AOC breakdowns are then handled within a single run using calculation looping technique, a built-in feature in Prophet that allows multiple calculations to be performed in a controlled sequence.
Here’s how it works in practice:
- Each loop corresponds to a different AOC component (e.g., Loop 1 = change in model, Loop 2 = effects of new business, etc.)
- Model logic and control tables are adjusted dynamically for each loop to reflect the specific change being measured
- Result storage is limited to what’s needed, avoiding the need to retain all projection details for every variable
- NO_CALC function is used to prevent certain variables from being overwritten during subsequent loops
By using this setup, Prophet performs AOC breakdowns efficiently within a single run structure, greatly reducing run time, file size, and the complexity of managing separate model versions.
Why This Matters
Streamlining AOC calculations in this way:
- Improves maintainability of Prophet models
- Reduces operational effort during monthly reporting cycles
- Increases transparency and control over roll-forward logic
- Supports better alignment with audit and governance requirements
As IFRS 17 becomes a steady-state reporting process, adopting efficient modeling techniques like this helps actuarial teams stay agile and focused on analysis, not administration.
