IFRS 17 Corner
By Bruce Rosner
The Financial Reporter, September 2023
At the time this column is being written, we already have a substantial number of companies that have issued results under IFRS 17, often including the full 2022 comparison to IFRS 4, as well as Q1 financials under IFRS 17 alone. I’ve been leafing through these looking for the common denominators, as well as the occasional unexpected tidbit. My sense right now is that results are largely playing out as we have expected over the last couple of years.
Many companies produced nice information packages that were simple enough for me to understand (a good thing). I studied seven of those, and here are some of my takeaways:[1]
- Equity has been successfully stabilized through better matching of Other Comprehensive Income (OCI) in assets and liabilities. There were actually instances where more of the assets moved into fair value through profit and loss accounting (FVPL) as a result of IFRS 9 adoption, and that will be one of the many ongoing challenges. I may have also inadvertently selected from regions that made heavier use of the election to disaggregate (e.g., Germany and Canada)—I have heard anecdotally that the UK and some other countries are seeing more P&L elections for the IFRS 17 liability.
- Some companies noted that their 2022 earnings appear less than ideal under IFRS 17 because their hedging programs were tailored to IFRS 4. That was obviously a difficult decision, being impossible in many cases to minimize volatility under both IFRS 4 and 17. And US subsidiaries have the additional task of managing their statutory reserves/capital, so may or may not attempt to minimize IFRS volatility.
- In general, we see lower equity, as the margins tend to be larger than the various types of provisions for adverse deviation inherent in prior accounting bases. However, as noted by Allianz, companies could consider taking credit for some portion of the contractual service margin (CSM[2]) when calculating the leverage ratio.
- IFRS 17 really shines when it comes to showing the sources of income. In the comparatives between IFRS 4 and 17, it’s generally easier to understand where the IFRS 17 income is coming from (e.g., amortization of profit tied up in the CSM).
- For years now, I have been trying to get a sense for how large the risk adjustment (RA) will be relative to the CSM or the best estimate liability (BEL). It’s very difficult to boil down into a metric that’s comparable across companies, particularly as the size of the RA will likely vary substantially between different types of products. With the latest batch of results, I can still see that there’s a wide range—comparing the RA release to the CSM amortization, I see ratios of 10%, 50%, 90%. I’ll keep an eye on this one.
- I see instances where companies focus on new business CSM as the metric of value added for long-duration business (e.g., Zurich), and other instances where companies show a reconciliation back to their internal definition (e.g., Generali).
The other big story right now are the “day two” challenges that companies are facing. Among those—reporting timelines and model run times are too long, staff still have trouble interpreting results, and companies are left with laundry lists of approximations put in place to get over the finish line.
It does seem like model run time sits in the middle of a lot of that, and companies should be thinking of ways to improve run time, and use it to improve their reporting timeline. IFRS 17 requires a series of runs to perform the necessary attribution steps and split results between P&L, OCI and CSM, and also requires some form of sensitivities for disclosures. Beyond these, companies often want to add runs to split the attribution further for management’s understanding, or add additional sensitivities for ALM purposes beyond the minimum requirements.
There are some clever techniques that have a lot of potential in this context. For example, is it really necessary to have the same level of precision in the attribution steps or sensitivities relative to the final quarter-end result? Here are two variations on a theme to exploit this:[3]
Method 1:
- Run a base run with full precision (as defined internally).
- Select a subset of scenarios that achieve materially correct results.
- Using the same random seed, run all subsequent scenarios only on the subset of scenarios.
Method 2:
- When running sensitivities, instead of running n scenarios at 100bps (for example), run n/10 scenarios at each increment of 20bps.
- Draw a regression line through those results, and report based on the regression line.
- This is considered to have a similar precision level to the full n scenarios (but the days when I knew how to prove that mathematically are gone).
Beyond that, there’s a long list of techniques that companies commonly employ both under IFRS 17 and elsewhere.
Good luck to all of you working through the reporting now for the first time.
Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the newsletter editors, or of Ernst & Young LLP or other members of the global EY organization.
Bruce Rosner, FSA, MAAA, is a managing director at Ernst & Young LLP. He can be reached at bruce.rosner@ey.com.