IFRS 17 Risk Adjustment Methods

By Jonas Osman Abdelghafour.

The IFRS 17 risk adjustment (RA) is the compensation an insurer requires for bearing the uncertainty in the amount and timing of cash flows arising from non-financial risk. It is one of the three building blocks of the fulfilment cash flows, and it feeds directly into the contractual service margin (CSM) at initial recognition and the pattern of profit emergence over the coverage period. The standard is deliberately silent on methodology, which puts real weight on the entity's method choice and calibration discipline.

What IFRS 17 actually requires

Paragraph 37 requires the RA to reflect the compensation the entity requires; paragraph 119 requires disclosure of the confidence level to which the chosen technique corresponds. The technique need not itself be a confidence-level (VaR) approach — but the result must be translatable into one for disclosure. In practice, three families are used.

Confidence-level (VaR) approach

The RA is set as the difference between a chosen percentile of the fulfilment cash flow distribution and the mean. Common calibrations range from the 65th to the 85th percentile for non-life short-tail, with lower percentiles for longer-tailed business where the underlying distribution is heavier.

The approach is intuitive and directly disclosure-compatible. Its weaknesses are that it depends on the shape of a distribution that is often estimated with substantial uncertainty, and that a fixed percentile can produce counter-intuitive results when the tail is heavy or the sample is small.

Cost-of-capital approach

The RA is calculated as the present value of the cost of holding capital against the non-financial risk over the run-off of the liabilities. A capital measure (typically an internal or standard-formula analogue restricted to non-financial risk) is projected each year, multiplied by a cost-of-capital rate (Solvency II uses 6%; IFRS 17 leaves the rate to the entity), and discounted.

The approach is coherent with capital-based pricing and is familiar to Solvency II reporters. Its weaknesses are the projection assumptions for capital run-off, sensitivity to the cost-of-capital rate, and the need to strip out financial risk from the underlying capital measure.

Conditional tail expectation (CTE) / TVaR

CTE at level α is the expected value of the loss conditional on being beyond the α percentile. It is coherent (in the Artzner sense), sensitive to tail shape, and generally preferred where distributions are heavy-tailed. Its weaknesses are heavier data requirements, greater sensitivity to model choice for the tail, and the need to translate the CTE calibration into an equivalent VaR percentile for disclosure.

Choosing a method

Method choice should reflect the nature of the business. Short-tail non-life with rich data supports a VaR approach with defensible percentile calibration. Long-tail casualty and life protection with material tail sensitivity often argue for CTE or cost-of-capital. Groups with a Solvency II framework already in place may find cost-of-capital administratively efficient because much of the capital projection infrastructure already exists.

Whichever method is chosen, the disclosure percentile must be genuinely representative of the RA, not reverse-engineered to a preferred number.

Calibration and diversification

The RA is set at the level at which the entity manages risk — usually the group or entity level — and the benefit of diversification across portfolios can be reflected. Diversification assumptions are among the most contested elements of RA calibration and warrant explicit documentation, sensitivity testing, and challenge from the independent validation function.

Implementation traps

Three recurring issues appear in reviews. Financial risk creeping into the RA base measure — the standard restricts the RA to non-financial risk only. Static percentile or cost-of-capital rate assumptions that never change even when the underlying risk environment does. And insufficient linkage between the RA release pattern and the actual emergence of uncertainty over the coverage period.

Limitations

Every RA method depends on a distributional or capital-projection assumption that is only weakly identified by data, particularly for long-tail lines. The disclosure percentile provides comparability across entities in form but not always in substance. Users of financial statements should read the RA alongside the methodology and calibration disclosures, not in isolation.

Conclusion

The IFRS 17 risk adjustment is a genuine actuarial judgement dressed as a disclosure. The choice among VaR, cost-of-capital, and CTE approaches should be driven by portfolio characteristics and governance capacity, not by whichever method produces the most convenient CSM. Related notes: actuarial model governance and the three lines of defence and loss reserving explained.

Written by Jonas Osman Abdelghafour, actuary and financial risk manager. Background and contact details are on the about page.