Show investors the combined ratio before discounting

I think European property and casualty insurers should lead their investor reporting with the combined ratio before discounting, and publish a consistent bridge to the discounted figure every time they report. This is a minority view among my peers, and I hold it knowing that it makes our own results look weaker in a period of higher interest rates. The reason is simple. The combined ratio is the one number most investors use to judge underwriting, and since IFRS 17 it can improve materially because of the interest rate environment rather than the quality of the underwriting.

Some insurers lead with a discounted ratio, some with an undiscounted one, and several disclose both under definitions that differ in detail. An investor comparing two insurers on their headline ratios may be comparing two different things. That is not a failure of the standard, which was designed for many purposes. It is a failure of presentation, and presentation is a choice each management team makes.

What the discount is really measuring

Under IFRS 17, most property and casualty business is measured using the premium allocation approach, but claims that have been incurred and not yet paid are generally discounted to reflect the time value of money. When a claim is recorded, the expected payments are discounted at current rates, which reduces the insurance service expense. Over the following years, that discount unwinds through insurance finance expense, outside the insurance service result. The combined ratio, built from the insurance service result, therefore benefits from the discount in the year the claim occurs, while the cost of unwinding it appears elsewhere.

The size of that benefit depends on two things management does not control in the short term: the level of interest rates and the length of time before claims are paid. A book dominated by liability claims that take years to settle gains far more from discounting than a book of property claims settled within months. When rates were near zero, the effect was negligible. When they rose, it became worth several points of combined ratio for many insurers. At Talvermark it was worth 3.5 points in 2025: our undiscounted combined ratio was 95.8% and our discounted ratio 92.3%.

The discount is not wrong. It reflects real economics: an insurer holding money for several years before paying a claim earns a return on it. But it is an investment and interest-rate effect, and a combined ratio that includes it no longer measures underwriting alone. If rates fall, the benefit shrinks and the reported ratio deteriorates, even if every risk was priced and handled exactly as before. Investors who have learnt to read the combined ratio as a measure of underwriting skill will then draw the wrong conclusion, and so may the insurer's own underwriters.

Why we lead with the harder number

When we prepared for IFRS 17 in 2022, our chief financial officer and head of investor relations argued for leading with the discounted figure. They had good reasons. It is the figure the standard produces, it was what several peers intended to report, and leading with a less flattering number risks being penalised by investors who compare headlines. I agreed with their analysis of the risk and disagreed with their conclusion.

My first concern was durability. A headline that interest rates have improved will, sooner or later, be a headline that interest rates worsen. I would rather explain a stable underwriting figure in a falling rate environment than explain why our combined ratio deteriorated without any change in our business.

My second concern was internal. The figures we report externally shape what we measure internally. If the headline combined ratio includes a discount benefit, it becomes tempting for underwriting teams, and for the incentive schemes that pay them, to treat that benefit as part of their performance. Long-tail liability business would look more attractive simply because its claims take longer to pay. That is exactly the sort of mispricing an insurer should avoid, and it begins with an apparently harmless reporting choice.

My third concern was comparability across time. Investors who have followed an insurer for twenty years have a sense of what its combined ratio means. The undiscounted figure is closer to what they knew before IFRS 17. It allows them to judge whether underwriting has improved since the old standard without having to reconstruct the effect of discounting themselves.

The cost was real. In our first year-end results under IFRS 17, published in March 2024, our headline ratio was higher than most peers'. Some commentators described our performance as weaker when the underlying underwriting was comparable or better. We spent time with analysts explaining the difference, and I accept that some investors simply looked at the headline and moved on.

One bridge, published every time

Leading with the undiscounted ratio is only half of the argument. The other half is a bridge, published in the same format every reporting period, showing how the undiscounted figure becomes the discounted one. Ours separates three elements: the discount effect itself, changes in the risk adjustment for non-financial risk, and other items. We also disclose the confidence level at which our risk adjustment is set, as the standard requires, with a short explanation of what that level means in practice.

The bridge did more for our credibility than the choice of headline. Before we published it, analysts' estimates of our discount effect ranged from 1.9 to 5.2 points. Afterwards, the range narrowed to within half a point. Investors who prefer the discounted figure can use it, and those who prefer the undiscounted figure can use that. What matters is that both can see how one becomes the other, and that we do not change the definitions when one of them becomes less convenient.

I would make the same case for the relationship between the IFRS 17 result and Solvency II. The two frameworks measure liabilities differently: Solvency II technical provisions use the risk-free curves EIOPA publishes and include a risk margin, while IFRS 17 uses discount rates and a risk adjustment that each insurer determines under the standard's principles. Investors do not need every difference explained in each release, but a short reconciliation once a year would help them understand why capital and earnings can move in different directions.

None of this requires new regulation. It requires management teams to decide that the purpose of a headline figure is to inform rather than to compare favourably. I would ask any insurance board to put one question to its finance team: if interest rates fell by two percentage points tomorrow, how much would our reported combined ratio deteriorate with no change in our underwriting? If the answer is more than a point or two, the board should consider whether investors are being shown the figure that best describes the business it oversees.